1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
3 * new feature: MAKE-HASH-TABLE now experimentally accepts a
4 :SYNCHRONIZED argument, which makes the hash-table safe for
5 concurrent accesses (but not iteration.) See also:
6 SB-EXT:WITH-LOCKED-HASH-TABLE, and
7 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
8 * bug fix: number of thread safety issues relating to SBCL's internal
9 hash-table usage have been fixed.
10 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
11 be returned from its body when the values were being returned
12 using unknown-values return convection and the W-P-O was wrapped
13 inside an UNWIND-PROTECT.
14 * bug fix: sb-posix should now compile again under Windows, enabling
17 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
18 * incompatible change: hash-table accessor functions are no longer
19 automatically protected by locks. Concurrent accesses on the same hash-table
20 from multiple threads can give inconsistent results or even corrupt the
21 hash-table completely. Multi-threaded applications should do their own
22 locking at the correct granularity. In the current implementation it is
23 still safe to have multiple readers access the same table, but it's not
24 guaranteed that this property will be maintained in future releases.
25 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
26 and will signal an error at runtime.
27 * enhancement: SB-THREAD package now exports a semaphore interface.
28 * enhancement: CONS can now stack-allocate on x86 and
29 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
31 * enhancement: nested lists can now be stack allocated on
32 platforms providing stack allocation support.
33 * enhancement: dynamic-extent support has been extended to support
34 cases where there are multiple possible sources for the stack
36 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
37 if the mutex is uncontested on Linux.
38 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
39 as specified by the CLHS. (thanks to Tobias Rittweiler)
40 * bug fix: NaN comparison now works on x86-64.
41 * bug fix: CLOSE :ABORT T on a stream with pending output now
43 * bug fix: instances of non-standard metaclasses using standard
44 instance structure protocol sometimes missed the slot type checks
46 * bug fix: known functions can be inlined in high-debug code.
48 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
49 * minor incompatible change: the MSI installer on Windows no longer
50 associates .lisp and .fasl files with the installed SBCL.
51 * minor incompatible change: :UNIX is no longer present in *FEATURES*
52 on Windows. (thanks to Luis Oliviera)
53 * new feature: SB-CLTL2 contrib module now implements
54 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
55 * optimization: scavenging weak pointers is now more efficient,
56 requiring O(1) instead of O(N) per weak pointer to identify
57 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
58 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
59 method bodies using constant slot-names when the first argument is
60 a specializer parameter for the method.
61 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
62 names, when the class of the instance is a direct instance
63 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
64 as slow as the constant slot-name case.)
65 * optimization: member type construction is now O(N) instead
67 * optimization: UNION and NUNION are now O(N+M) for large
68 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
70 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
71 now more readable in environments like Slime which display it.
72 (thanks to Tobias C. Rittweiler)
73 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
74 was able to infer the correct argument type for the object on which
75 the CAS operation was being performed.
76 * bug fix: copy propagation interfered with parallel assignment
77 semantics in local calls. (reported by Paul Khuong)
78 * bug fix: the signed modular fixnum shift compiled to wrong code on
79 x86 and x86-64. (spotted by a slight modification to some of
82 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
83 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
84 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
85 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
87 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
88 and generic function now print the object identity for anonymous
89 instances. (thanks to Attila Lendvai)
90 * enhancement: as an extension to MOP, SBCL now supports SETF of
91 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
92 (thanks to Attila Lendvai)
93 * enhancement: sb-cover has an alternate annotation mode which puts
94 the form annotations at the car of the form, instead of annotating
95 it completely. The benefit of this mode is that it shows better
96 which forms were instrumented by the compiler.
97 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
98 by printer control variables. (reported by Dan Corkill)
99 * bug fix: system leaked memory when delayed output was performed by
100 the OS in smaller chunks then expected. (thanks to David Smith)
101 * bug fix: system leaked memory when file streams were not closed
103 * bug fix: large objects written to slow streams that were modified
104 after the write could end up with the modified state written to
105 the underlying file descriptor.
106 * bug fix: multiple threads operating in parallel on the same stream
107 could cause buffer-overflows.
108 * bug fix: source location information is stored correctly
109 (broken since 1.0.6). This bug would generally show up as the
110 Slime debugger higlighting the wrong form when the sldb-show-source
112 * bug fix: Branch forms are again annotated as branches in the sb-cover
114 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
115 on x86-64. (reported by Gregory Vanuxem)
116 * bug fix: EQUALP could return wrong results for structures with raw
117 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
118 word). (reported by Vjacheslav Fyodorov)
119 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
122 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
123 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
124 atomic compare-and-swap operations on threaded platforms.
125 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
126 allows assining a global minimum value to optimization qualities
127 (overriding proclamations and declarations).
128 * enhancement: closed over variables can be stack-allocated on x86
130 * performance bug fix: GETHASH and (SETF GETHASH) are once again
132 * optimization: slot definition lookup is now O(1). This speeds up
133 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
134 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
135 * optimization: ASSOC and MEMBER can now be open-coded for all
136 combinations of keyword arguments when second argument is constant
137 and SPEED >= SPACE. In other cases a specialized version is
139 * bug fix: using obsoleted structure instances with TYPEP and
140 generic functions now signals a sensible error.
141 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
142 (reported by Kristoffer Kvello)
143 * bug fix: backtrace construction is now more careful when making
144 lisp-objects from pointers on the stack, to avoid creating bogus
145 objects that can be seen by the GC.
146 * bug fix: defaulting of values in contexts expecting more than 7
147 variables now works on x86-64. (reported by Christopher Laux)
148 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
150 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
151 as the property-list of a symbol.
152 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
153 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
154 situations requested, are once again file-compileable. (reported
157 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
158 * MOP improvement: support for user-defined subclasses of
159 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
160 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
161 the expansion of DEFMETHOD, is responsible for generating a form
162 which creates a list of specializers when evaluated. Additional
163 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
164 debugging and introspective support.
165 * minor incompatible change: the (unsupported) spinlock interface
166 has changed: free spinlock now has the value NIL, and a held spinlock
167 has the owning thread as its value.
168 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
169 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
171 * enhancement: name of a socket-stream is now "a socket" instead of
173 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
174 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
175 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
176 for generic functions. (Thanks to Tobias C. Rittweiler)
177 * optimization: bignum printing speed has been improved by 20-40%
178 (depending on the bignum size.)
179 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
181 * bug fix: the cache used by the CLOS to store precomputed effective
182 methods, slot offsets, and constant return values is now thread and
184 * bug fix: generic function dispatch function updating is now thread
185 and interrupt safe (in the sense that the known issues have been
187 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
188 * bug fix: interning EQL-specializers is now thread and interrupt safe.
189 * bug fix: asdf systems with dependencies to the SB-POSIX or
190 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
191 * bug fix: interrupt safety of applicable method computation has been
194 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
195 * new contrib: sb-cover, an experimental code coverage tool, is included
197 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
198 significantly faster.
199 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
200 produce more efficient code on x86-64 if the compiler can determine
201 that (+ INDEX <constant>) does not require a bounds check and FOO
202 has an element type at least 8 bits wide.
203 * enhancement: a new, experimental synchronous timeout facility is
204 provided. Refer to SB-SYS:WITH-DEADLINE for details.
205 * enhancement: when a symbol name conflict error arises, the
206 conflicting symbols are always printed with a package prefix.
207 (thanks to Kevin Reid)
208 * enhancement: stepping is now once again supported on the SPARC and
209 MIPS platforms. (It is also now more likely to work on CheneyGC
211 * enhancement: sb-sprof can now also track and report accurate call
213 * bug fixes: the treatment of non-standard subclasses of
214 SB-MOP:SPECIALIZER is more correct.
215 * incompatible change: PURIFY no longer copies the data from the
216 dynamic space into the static and read-only spaces on platforms
217 that use the generational garbage collector
218 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
220 * bug fix: GC race condition occasionally resulting in crashes with
221 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
223 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
225 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
226 a STYLE-WARNING to be signalled (regression from 1.0.4.)
227 * bug fix: an asynchronous interrupt could previously leave the
228 system running with GC inhibited.
229 * bug fix: a DECLARE form evaluated at top-level now causes an error
230 rather than silently (or verbosely) returning NIL.
231 * bug fix: trying to dissassemble functions compiled at high DEBUG could
232 result in a error being signalled due to source form lookup errors.
233 (reported by Peter Graves)
235 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
236 * incompatible change: removed writer methods for host-ent-name,
237 host-ent-addresses -- changing the values did not update the DNS
239 * minor incompatible change: changed experimental JOIN-THREAD interface
240 * documentation: the manual now lists reader and writer methods
241 in class slot documentation sections. (thanks to Richard M Kreuter)
242 * documentation: unwinding from asyncronous events has been
243 documented as unsafe.
244 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
245 in multithreaded application code.
246 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
247 platforms. (thanks to James Anderson for the optimization hint)
248 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
250 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
251 variants no longer cons.
252 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
253 their NOT- variants no longer cons.
254 * optimization: EQUAL hash tables no longer use SXHASH for objects
255 of all data types, but instead use an EQL hash for types for which
256 EQUAL is the same as EQL.
257 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
258 are significantly faster.
259 * optimization: new STRING-OUTPUT-STREAM implementation is much
260 faster for large outputs and conses 30% less on average.
261 * enhancement: XREF information is now collected for references made
262 to global variables using SYMBOL-VALUE and a constant argument.
263 * enhancement: SIGINT now causes a specific condition
264 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
265 * bug fix: FILE-POSITION used to signal an error in some cases where
266 ANSI requires it to return NIL.
267 * bug fix: ADJUST-ARRAY is now interrupt-safe.
268 * bug fix: adding and removing fd-handlers is now interrupt-safe.
269 * bug fix: inlined calls to C now ensure 16byte stack alignment on
271 * bug fix: bad type declaration in the CLOS implementation has
272 been fixed. (reported by James Anderson)
273 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
274 (thanks to Lutz Euler)
275 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
276 required by the CLHS. (reported by Samium Gromoff)
277 * bug fix: dead unbound variable references now signal an error.
278 (reported by Marco Monteiro)
279 * bug fix: / with an unused value was being deleted in safe code.
280 (thanks to Marco Monteiro and Kevin Reid)
281 * bug fix: number of characters that can be written onto a single
282 line in a file is unlimited.
283 * bug fix: some GC deadlocks caused by asynchronous interrupts have
284 been fixed by inhibiting interrupts for when GC is disbled.
285 * bug fix: some interrupt safety issues with GETHASH, (SETF
286 GETHASH), CLRHASH and REMHASH have been fixed.
287 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
288 specifier no longer causes infinite recursion.
289 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
290 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
291 * bug fix: an &environment argument with the correct variable information
292 is passed to macros that are expanded during byte compilation.
293 (reported by Samium Gromoff)
294 * bug fix: a logic error was causing the hash function for bignums to
295 have excessive amounts of collisions. (reported by Faré Rideau)
296 * bug fix: modifying the contents of an array could change the return
297 value of SXHASH on that array, which is only allowed for strings
298 and bit vectors. (bug introduced in 0.9.16)
299 * improvement: the x86-64/darwin port now passes all tests (except
300 for the debugger tests) but should still be considered
301 experimental until this is fixed.
302 * improvement: a style-warning is signaled for CASE (etc) clauses with
303 duplicate keys. (thanks to Kevin Reid)
304 * improvement: macos/x86 and macos/x86-64 now use mach exception
305 handlers for dealing with illegal instructions (for trapping and
306 error handling) and memory protection violations (for GC).
308 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
309 * new platform: experimental support for x86-64/darwin (MacOS).
310 * incompatible change: the thread-safe (on most platforms) getaddrinfo
311 and getnameinfo sockets functions are used instead of gethostbyaddr
312 and gethostbyname, on platforms where the newer functions are available.
313 As a result, the ALIASES field of HOST-ENT will always be NIL on these
315 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
316 don't print the startup banner, but behave as if --noinform was passed
317 as a command line argument. (thanks to Kevin Reid)
318 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
319 * optimization: code using alien values with undeclared types is much faster.
320 * optimization: the compiler is now able to open code SEARCH in more cases.
321 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
322 * bug fix: using standardized COMMON-LISP special variables as loop
323 variables no longer signals bogus package lock violations. (reported
325 * bug fix: declaring local loop variables to be of a range-limited type
326 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
327 (reported by Andras Simon)
328 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
329 bugs remain on x86-64.)
330 * bug fix: the #= and ## reader macros now interact reasonably with
331 funcallable instances.
332 * bug fix: type-checks for function arguments were compiled using the
333 compiler policy settings of the wrong lexical environment (in 1.0.2
335 * bug fix: SHADOW accepts characters as string designators, as required
336 by the spec (thanks to Eric Marsden)
337 * bug fix: fixed GC safety issues when foreign functions are called with
338 non-base strings as arguments
339 * bug fix: more consistent error output for fatal-compiler-warnings, like
341 * bug fix: functions with proper names were showing up as NIL in
342 backtraces (reported by Edi Weitz, regression in 1.0.2)
344 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
345 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
346 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
347 produce more efficient code on the x86 if the compiler can determine
348 that (+ INDEX <constant>) does not require a bounds check and FOO
349 has an element type at least 8 bits wide.
350 * bug fix: references to undefined variables are handled the same way
351 in toplevel forms as in the normal compiler
352 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
354 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
355 (thanks to Jon Buller)
356 * improvement: faster compilation times for complex functions
357 * improvement: added readlink support to SB-POSIX (thanks to Richard
360 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
361 * improvement: experimental support for mach exception handling on
362 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
364 * improvement: support for GBK external format.
365 (thanks to Chun Tian (binghe))
366 * improvement: the debugger now displays variables that have been closed
367 over, in code compiled with (DEBUG 2) or higher
368 * improvement: support for executable cores on NetBSD (thanks to
370 * new feature: added a RESTART-FRAME debugger command
371 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
372 be used to provide an implementation for FILE-POSITION on Gray streams
373 (thanks to Eric Marsden)
374 * new feature: add syslog, openlog and closelog support to SB-POSIX
375 (thanks to Richard Kreuter)
376 * optimization: the function call overhead in code compiled with
377 a high DEBUG optimization setting is significantly
378 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
379 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
380 (thanks to Tony Martinez)
381 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
382 evaluated once (reported by Kevin Reid)
383 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
384 works again (thanks to Kilian Sprotte)
385 * bug fix: an error is signaled for tagbodies with duplicate tags
386 (thanks to Stephen Wilson)
387 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
388 * bug fix: Win32 port can now handle foreign code unwinding Lisp
389 stack frames from alien callbacks.
390 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
391 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
392 65k pages (thanks to David Woodhouse)
393 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
395 changes in sbcl-1.0.1 relative to sbcl-1.0:
396 * new platform: FreeBSD/x86-64, including support for threading.
397 * new feature: the compiler stores cross-referencing information
398 abount function calls (who-calls), macroexpansion (who-macroexpands)
399 and special variables (who-binds, who-sets, who-references) for code
400 compiled with (< SPACE 3). This information is available through the
401 sb-introspect contrib.
402 * new feature: users may subclass SEQUENCE, and have instances of
403 these classes interoperate with standard Common Lisp functions if
404 a number of methods are defined. (This feature is experimental
405 and the interface subject to change based on feedback from SBCL
406 users and the general community)
407 * improvement: sb-sprof traces call stacks to an arbitrary depth on
408 x86/x86-64, rather than the previous fixed depth of 8
409 * improvement: another pthread back-end of mutex "pthread-futex".
410 * bug fix: non-ascii command-line arguments are processed correctly
411 (thanks to Yaroslav Kavenchuk)
412 * bug fix: non-required arguments were not passed correctly when a method
413 defined using DEFMETHOD was called from a mop-generated method using
414 CALL-NEXT-METHOD (reported by Pascal Costanza)
415 * bug fix: recursion is now permitted in accessors through
416 SLOT-UNBOUND. (reported by Pascal Costanza)
417 * bug fix: an error was signaled at startup if the HOME environment
418 variable was defined, but had an empty value (reported by Peter Van Eynde)
419 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
420 signal a PROGRAM-ERROR, not a BUG.
421 * bug fix: SB-POSIX:READDIR works when built with large file support.
422 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
424 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
425 for better automatic indentation support. (thanks to Matt Pillsbury)
426 * optimization: loading generic functions no longer takes O(n^2) time,
427 proportional to the amount of methods in the generic function
428 (reported by Todd Sabin and Jeremy Brown)
429 * optimization: the FIND and POSITION family of sequence functions
430 are significantly faster on arrays whose element types have been
432 * improvements to the Windows port:
433 ** Intermittent heap corruption problems have been fixed. (thanks
434 to Alastair Bridgewater)
435 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
437 ** Lisp is able to unwind foreign exception frames from alien
438 callbacks. (thanks to Alastair Bridgewater)
440 changes in sbcl-1.0 relative to sbcl-0.9.18:
441 * improvement: experimental support for threading on FreeBSD/x86.
442 (thanks to NIIMI Satoshi)
443 * improvement: runtime option --dynamic-space-size can be used
444 to set the size of the dynamic space reserved on startup.
445 * improvement: floating point modes in effect are now saved in
446 core, and restored on startup.
447 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
448 startup, not time since first call to GET-INTERNAL-REAL-TIME.
449 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
450 threads are not running after *SAVE-HOOKS* have run.
451 * improvement: writes to CLOS instance slots are type-checked in code
452 compiled with (SAFETY 3)
453 * improvement: floating-point exception handling on FreeBSD (thanks to
455 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
456 (thanks to Zach Beane)
457 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
459 * improvement: added support for the Shift-JIS external format.
460 (contributed by NIIMI Satoshi)
461 * improvement: callbacks are supported on Linux/PPC. (thanks to
463 * bug fix: compiler bug triggered by a (non-standard) VALUES
464 declaration in a LET* was fixed. (reported by Kaersten Poeck)
465 * bug fix: file compiler no longer confuses validated and already
466 dumped structurres. (reported by Kaersten Poeck)
467 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
468 fill-pointer signals a type-error as required. (thanks to
470 * bug fix: disassemly of funcallable instances works.
471 * bug fix: single stepping on PPC.
472 * bug fix: fix thread-safety problems in the type system (generally
473 manifesting as nonsensical errors like "STRING is a bad type specifier
474 for sequences" or "The value 1 is not of type FIXNUM")
475 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
476 (reported by Josip Gracin)
477 * bug fix: an error is signaled for attempts to displace arrays with
478 incompatible element types (thanks to Mario Mommer)
479 * bug fix: more correct handling of wide characters in debug info
480 (bug reported by Attila Lendvai and fixed by Juho Snellman)
481 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
482 and don't cause extra consing
483 * optimization: MAP and MAP-INTO are significantly faster on vectors
484 whose elements types have been declared.
485 * Improvements to SB-SPROF:
486 ** Support for allocation profiling
487 ** Reduced profiling overhead, especially for long profiling runs
488 * Improvements to the Windows port:
489 ** floating point exceptions are now reported correctly.
490 ** stack exhaustion detection works partially.
491 ** more accurate GET-INTERNAL-REAL-TIME.
492 ** better RUN-PROGRAM behaviour in case of errors in the child
494 ** PROBE-FILE now simplifies pathnames correctly.
495 ** DIRECTORY now works correctly with :WILD-INFERIORS.
496 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
497 1st 1970 (although time timezone data will be unreliable).
498 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
500 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
501 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
502 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
504 * bug fix: two potential GC deadlocks affecting threaded builds.
505 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
506 returns T (reported by Anton Kazennikov)
507 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
508 with non-variable places
509 * bug fix: remove a race condition in the setting of
510 funcallable-instance functions, this should make threaded CLOS
511 code more stable against memory faults.
512 * bug fix: corruption of specials when unbinding is interrupted by an
513 asynchronous unwind (reported by Hannu Koivisto)
514 * improvement: the debugger will now also display local variables that
515 are only used once, for code compiled with a DEBUG optimization quality
518 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
519 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
520 * incompatible change: External-format support for FFI calls. The
521 SB-ALIEN:C-STRING no longer implies an ASCII
522 external-format. Instead the string is subject to external-format
523 conversion. Additionally return values of type SB-ALIEN:C-STRING
524 are by default of type (SIMPLE-ARRAY CHARACTER), not
525 SIMPLE-BASE-STRING. For an alien type that behaves like the old
526 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
527 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
528 * incompatible change: SB-EXT package no longer contains the
529 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
530 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
532 * incompatible change: the single-stepper is no longer available
533 on Alpha, Mips and Sparc platforms.
534 * minor incompatible change: the direct superclasses of
535 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
536 not (STANDARD-OBJECT FUNCTION). This makes the
537 class-precedence-lists of GENERIC-FUNCTION and
538 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
540 * minor incompatible change: the default stream external format on
541 non-unicode SBCL is no longer detected from the locale, but is always
543 * new feature: Add a version of evaluator that uses an interpreter instead
544 of the compiler. EVAL still uses the compiler by default, to switch it
545 to use the interpreter, set the value of the variable
546 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
547 * minor incompatible change: the single-stepper REPL has been merged
548 with the normal debugger (see the "Stepping" heading of the debugger help
549 for more details). The debugger command STEP will no longer switch
550 to the single-stepper REPL.
551 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
552 for a type now works.
553 * bug fix: loading of default sysinit file works. (thanks to Leonid
555 * bug fix: better detection of circularities in the file-compiler.
556 (reported by Marco Monteiro)
557 * bug fix: the CL pathname functions now work with files that have
558 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
559 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
560 systems with Unix98 pty semantics.
561 * bug fix: ASDF-INSTALL will now work with bsd tar.
562 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
564 * bug fix: timers expiring in dead threads no longer cause a
565 type-error (reported by Paul "Nonny Mouse").
566 * bug fix: thanks to more lightweight single-stepper instrumentation,
567 code compiled with (DEBUG 3) will compile and execute significantly faster,
568 and will have more accurate type-inferencing than before
569 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
570 whose bindings are modified
571 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
572 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
573 ** replace dummy implementations of CL:MACHINE-INSTANCE and
574 CL:SOFTWARE-VERSION with proper version
576 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
577 * feature: implemented the READER-METHOD-CLASS and
578 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
579 as specified by AMOP.
580 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
582 * optimization: faster LOGCOUNT implementation on x86 and x86-64
583 (thanks to Lutz Euler)
584 * optimization: hashing of general arrays and vectors has been
585 improved. (reported by Any Fingerhut)
586 * enhancement: SB-INTROSPECT is now able to find definitions of
587 profiled functions. (thanks to Troels Henriksen)
588 * enhancement: compiler-macro expansion applies now to FUNCALL forms
589 as well, allowing compiler-macros for SETF-functions to expand.
590 * enhancement: step-instrumentation no longer wraps known
591 single-value functions with multiple-value context, allowing
592 better type inference.
593 * fixed bug #337: use of MAKE-METHOD in method combination now works
594 even in the presence of user-defined method classes. (reported by
595 Bruno Haible and Pascal Costanza)
596 * fixed bug #339(c): if there are applicable methods not part of any
597 long-form method-combination group, call INVALID-METHOD-ERROR.
598 (reported by Bruno Haible)
599 * fixed bug #361: the :FUNCTION initarg in the protocol for
600 initialization of methods can now be used to override
601 internally-produced optimized functions. (reported by Bruno
603 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
604 system-provided lambda expression no longer cause warnings about
605 unbound #:|pv-table| symbols.
606 * bug fix: improved the handling of type declarations and the
607 detection of violations for keyword arguments with non-constant
609 * bug fix: potentially erronous calls to PATHNAME and
610 MERGE-PATHNAMES were being flushed in some cases.
611 (reported by Richard Kreuter)
612 * bug fix: compiled calls to TYPEP were mishandling obsolete
613 instances. (reported by James Bielman and Attila Lendvai)
614 * bug fix: format strings with ~> without matching ~< no longer
615 trigger an AVER, but signal an understandable error instead.
616 (reported by Antonio Martinez)
617 * bug fix: specifying an output-file without a pathname-type for
618 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
619 of input-file instead of "fasl". (reported by Robert Dodier)
620 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
621 (reported by James Y Knight).
622 * bug fix: compiler-macros-function did not consider the environment
623 argument for shadowing by local functions.
624 * bug fix: compiler-macros expansion was inhibited by local INLINE
626 * bug fix: inline expansions of known functions were subject to
627 step-instrumentation in high debug policies, leading to problems
629 * bug fix: compiler failed to differentiate between different CONS
631 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
632 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
633 * bug fix: #368: incorrect use of expressed vs. upgraded array
635 * bug fix: #306a: more precise unions of array types.
636 * thread-safety improvements:
637 ** CONDITION-WAIT could return early on Linux, if the thread was
638 interrupted and subsequently continued with SIGCONT.
639 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
641 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
642 * added support for the ucs-2 external format. (contributed by Ivan
644 * minor incompatible change: pretty printing of objects of type
645 (cons symbol) is, in the default pprint-dispatch-table, now
646 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
648 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
649 called later in a class's lifetime, possibly as late as when the
650 first instance of the class is created. Previously,
651 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
652 class became finalizeable.
653 * fixed bug: FILE-POSITION sometimes returned inconsistent results
654 for multibyte external-format streams. (thanks to "vbzoli")
655 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
656 with :ALLOCATION :CLASS inherited from superclasses of the
658 * fixed bug: anonymous classes can now be created using the :NAME
659 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
660 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
661 * fixed bug: core-files saved with :EXECUTABLE T can again be
662 executed when SBCL_HOME isn't set. (reported by James Knight)
663 * fixed bug: toplevel LOCALLY forms with declarations could
664 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
665 * fixed bug: printing from several different threads using different
666 values of *print-case* could cause invalid output, due to
667 some internal special variables of the printer not being bound
668 thread-locally (reported by Max Mikhanosha)
669 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
670 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
671 (reported by Pascal Costanza)
672 * fixed bug: FINALIZE-INHERITANCE is called from
673 REINITIALIZE-INSTANCE on classes when the class has previously
674 been finalized, as required by AMOP.
675 * minor code generation optimizations:
676 ** better register allocation in CLOS dispatching functions
677 ** overflow detection when coercing signed bytes to fixnums on x86-64
678 ** is now implemented with one IMUL instruction instead of three shifts
679 ** more efficient bit-vector access on x86 and x86-64
680 ** more efficient access to raw structure slots on x86 and x86-64
681 * fixed some bugs revealed by Paul Dietz' test suite:
682 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
685 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
686 * feature: thread support on Solaris/x86, and experimental thread support
688 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
690 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
691 default method for SLOT-UNBOUND.
692 * minor incompatible change: prevent the user from specializing the
693 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
694 somewhat counter to the intent of the protocol, I (CSR) think, and
695 additionally it just doesn't work in SBCL as currently
696 implemented, thanks to optimizations (that are always valid for
697 the other three SLOT-VALUEish functions, but not for the setter).
698 * minor incompatible change: the :SB-LDB feature is now enabled by
699 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
700 the low-level debugger.
701 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
702 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
703 down to an SB-SYS:FD-STREAM, allowing direct communication in
704 more cases. Temporary files and pipes are still used as a fallback
706 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
707 on the non-reentrant C functions localtime(3) and gmtime(3).
708 * bug fix: global optimization policy was not visible in LOCALLY and
710 * bug fix: class objects can be used as specializers in methods.
711 (reported by Pascal Costanza)
712 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
713 with a type error. (reported by blitz_ on #lisp)
714 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
715 failed with a type error. (reported by Pascal Bourguignon)
716 * bug fix: merging pathnames against defaults with :DIRECTORY
717 starting with '(:RELATIVE :BACK) should preserve the :BACK.
718 (reported by James Y Knight)
719 * bug fix: saving large (>2GB) cores on x86-64 now works
720 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
721 constant index and a (simple-array (signed-byte 32)) array
722 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
723 error (patch by Robert J. Macomber)
724 * feature: TIME now displays the amount of run-time spent in GC
725 * bug fix: The debugger now does a better job of respecting
726 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
727 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
728 workaround for bug 403.)
729 * bug fix: floating point exception handling now works on FreeBSD
730 (thanks to NIIMI Satoshi)
731 * fixed some bugs revealed by Paul Dietz' test suite:
732 ** MISC.641: LET-conversion were not supposed to work in late
734 * improvements to the Win32/x86 port:
735 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
736 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
738 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
739 * new feature: source path information is generated for macro-expansion
740 errors for use in IDE's like Slime (thanks to Helmut Eller)
741 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
742 * bug fix: compiler does not loop forever on an invalid type in
744 * improvement: compilation of most CLOS applications is significantly
746 * optimization: added a limited bytecode compiler for simple toplevel
747 forms, speeding up compilation and FASL loading
748 * bug fix: the statistical profiler now properly distinguishes anonymous
751 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
752 * minor incompatible change: in sbcl-0.9.11 (but not earlier
753 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
754 into subclasses of the finalized class. Now user calls to
755 FINALIZE-INHERITANCE finalize just the one class, and calls by the
756 system return before any subclasses are finalized.
757 * minor incompatible change: The reader no longer ignores errors
758 regarding non-existent packages in #+ and #- feature tests.
759 * new feature: command line options --no-sysinit, --no-userinit to
760 inhibit loading the corresponding init files
761 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
762 for sending data through UDP sockets (thanks to François-René Rideau)
763 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
764 error is signalled instead (thanks to François-René Rideau)
765 * minor incompatible change: Error signalling behaviour of lexical
766 operations violating package locks has changed slightly. Refer to
767 documentation on package locks for details.
768 * enhancement: EVAL can process IF-expressions without resorting to the
770 * bug fix: Some operations involving SIN, COS, or TAN weren't being
771 constant-folded properly due to missing out-of-line functions.
772 (reported by Mika Pihlajamäki)
773 * bug fix: LISTEN sometimes returned T even in cases where no data was
774 immediately available from the stream
775 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
776 were reversed. (reported by Levente Mészáros)
777 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
778 when given a vector displaced to an adjustable array. (reported
780 * bug fix: derivation of float boundaries from numbers outside the
781 appropriate float range (reported by John Wiseman)
782 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
784 * bug fix: precomputing cache entries for generic functions with
785 some subclasses of specializers as yet invalid does not attempt to
786 fill a cache line with a negative offset. (reported by Levente
788 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
789 allow condition class objects as arguments to DOCUMENTATION and
790 (SETF DOCUMENTATION); only find and set documentation for
791 structure names for the STRUCTURE doc-type. (suggested by Gary
793 * improvements to the Win32/x86 port:
794 ** (user-homedir-pathname) and default initialization file
795 locations now know about the user's "Documents and Settings"
796 directory (thanks to Yaroslav Kavenchuk)
797 ** run-program is implemented (thanks to Mike Thomas)
798 ** sockets support (thanks to Timothy Ritchey)
799 ** better backtrace support (thanks to Alastair Bridgewater)
800 ** sb-grovel supported
801 ** asdf-install and sb-posix work somewhat
802 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
803 * improvements to the Solaris/x86 port:
804 ** works on Solaris 11/Solaris Express
805 ** floating-point exception handling support
806 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
807 * fixed some bugs revealed by Paul Dietz' test suite:
808 ** REMOVE-METHOD returns its generic function argument even when
809 the method is not one of the generic functions' methods.
810 ** objects declared as MEMBER types can be admissible arguments to
812 ** printing characters should simply be printed by the FORMAT ~:C
814 ** compiler failure when compiling functions with hairy constant
815 defaults for optional parameters.
816 ** compiler produces wrong code when MAYBE-INLINE-expanding a
817 function, which is already optimized.
819 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
820 * new platform: experimental support for SBCL x86/Darwin, including
821 MacOS X 10.4.5 on Intel.
822 * new feature: Unicode character names are now known to the system
823 (through CHAR-NAME and NAME-CHAR).
824 * new feature: the filesystem location of SBCL's core file is
825 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
826 * minor incompatible change: the contrib modules SB-POSIX and
827 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
828 this change is to make it easier to distribute
829 location-independent binaries.
830 * bug fix: as implied by AMOP, standardized classes no longer have
831 slots named by external symbols of public packages. (reported by
833 * optimization: calling functions via constant symbols -- as in
834 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
835 via the function object as in (FUNCALL #'FOO).
836 * optimization: on x86 and x86-64, the calling convention for the
837 case of dealing with an unknown number of values has been altered
838 to be friendlier to the prediction heuristics implemented,
839 particularly with reference to CALL and RET pairing. (thanks to
840 Alastair Bridgewater)
841 * optimization: on x86, the code for access to thread-local symbol
842 values and binding/unbinding of thread-local symbols is smaller.
843 (thanks to Alastair Bridgewater)
844 * enhancement: CONSTANTP is now able to determine constantness of
845 more complex forms, including calls to constant-foldable standardized
846 functions and some special forms beyond QUOTE.
847 * fixed bug: occasional GC crashes on Solaris/x86
848 * optimization: x86-64 supports stack allocation of results of simple
849 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
850 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
851 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
852 (thanks to James Knight)
853 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
854 by the standard (thanks to Stephen Compall)
856 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
857 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
858 be used for bundling the runtime and the core file into one
859 executable binary. This feature is not currently supported on all SBCL
860 platforms. (thanks to James Bielman and NIIMI Satoshi)
861 * new feature: a generational or ephemeral garbage collector is now
862 the default on the PowerPC platform (both Linux and Darwin). The
863 old Cheney (stop and copy) collector is a build-time option.
864 (thanks to Cyrus Harmon, after Raymond Toy)
865 * minor incompatible change: the method by which SBCL finds its
866 contributed modules has changed; it no longer relies on symbolic
867 links from an $SBCL_HOME/systems directory, but searches directly
868 in the subdirectories of $SBCL_HOME.
869 * enhancement: the dynamic heap size on the Linux/PPC platform is
870 markedly larger, even using the older Cheney garbage collector.
871 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
872 certain complicated string types. (reported by Gary King)
873 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
874 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
875 * fixed bug: DOCUMENTATION on structures no longer signals an
876 error if no structure type of the right name exists. (reported by
878 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
879 applicable non-standard primary or :AROUND method on
880 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
881 no longer cause unbound variable errors. (reported by Kevin Reid)
882 * fixed bug: metacircle resolution in cases where methods have slots
883 added before the slots from STANDARD-METHOD. (reported by Jean
885 * fixed bug: the Power PC assembly code for calling into lisp
886 sometimes computed its return address wrongly (depending on the
887 15th bit of the address, fixed at link time). This used to
888 manifest itself in a segmentation violation while building PCL.
889 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
890 many others over the years)
891 * fixed bug: ROOM no longer reports silly numbers for stack usage if
892 the stack pointer (treated as a signed integer) is negative.
893 (thanks to Peter van Eynde)
895 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
896 * new platform: experimental support for the Windows operating
897 system has been added. (thanks to Alastair Bridgewater)
898 * fixed several bugs in and robustified the PPC FFI (including
899 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
900 * bug fix: allow non-simple string symbol names (reported by Paul
902 * bug fix: interrupt handling on NetBSD (thanks to Richard M
904 * bug fix: saving a core corrupted callbacks on x86/x86-64
905 * bug fix: closed a loophole in metacircularity detection and
906 grounding in the PCL implementation of CLOS.
907 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
908 (reported by tomppa on #lisp)
909 * bug fix: clear the direction flag on Lisp -> C transitions, as
910 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
911 (reported by Andrej Grozin and Hendrik Maryns)
912 * optimization: major improvements to GC efficiency on GENCGC platforms
913 * optimization: faster implementation of EQUAL
914 * optimization: emit more efficient opcodes for some common
915 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
916 * optimization: several other minor code-generation improvements on x86-64
917 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
919 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
920 * minor incompatible change: (SETF CLASS-NAME) and (SETF
921 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
922 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
923 by ANSI as a generic function, and for consistency (SETF
924 GENERIC-FUNCTION-NAME) is treated likewise.
925 * fixed bug #233.b: make constraint propagation notice when a variable
926 value is changed after it is referenced but before it is used
927 * fixed bug #296: no more arbitrary behaviour from filenames with
928 odd characters as --load arguments. (reported by Adam Warner)
929 * fixed bug #390: :CHARACTER-SET pathname components now work as
930 expected. (reported by Tim Daly Jr)
931 * fixed bug #391: complicated :TYPE intersections in slot
932 definitions no longer cause an error in PCL internals.
933 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
934 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
935 returning the number of octets which would be written to the
936 file-stream. (thanks to Robert J. Macomber)
937 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
938 arguments to the internal stream functions. (thanks to David
940 * bug fix: stack exhaustion is now handled on the x86/SunOS
941 platform. (thanks to Daisuke Homma)
942 * bug fix: more accurate ROOM results on GENCGC platforms
943 * optimization: improved type inference for arithmetic-for
944 index variables in LOOP
945 * optimization: faster floating-point SQRT on x86-64
946 * fixed some bugs revealed by Paul Dietz' test suite:
947 ** DOCUMENTATION returns NIL instead of "" for method combinations
948 that don't have a docstring
950 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
951 * minor incompatible change: (SETF CLASS-NAME) and (SETF
952 GENERIC-FUNCTION-NAME) are no longer generic functions, and
953 therefore are not customizeable by user code (as seems to be at
954 least permitted and maybe required by AMOP). As a consolation,
955 however, the SBCL implementation of these functions now calls
956 REINITIALIZE-INSTANCE as specified by AMOP.
957 * bug fix: slot-definition documentation is propagated as per ANSI
958 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
959 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
960 argument, as do eslotd initialization methods. (from Pascal
961 Costanza's "Closer" project)
962 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
963 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
965 * bug fix: it is now possible to have more than one subclass of
966 STANDARD-GENERIC-FUNCTION without causing stack overflow.
967 (reported by Bruno Haible, Pascal Costanza and others)
968 * bug fix: the dependent update protocol now works for generic
969 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
971 * bug fix: condition-class instances corresponding to
972 DEFINE-CONDITION forms are now created eagerly. (reported by
973 Kalle Olavi Niemitalo on comp.lang.lisp)
974 * bug fix: floating point printing is more accurate in some
975 circumstances. (thanks to Simon Alexander)
976 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
977 merged with *DEFAULT-PATHNAME-DEFAULTS*.
978 * bug fix: callbacks on OS X now preserve stack-alignment.
979 (thanks to Cyrus Harmon)
980 * enhancement: the x86-64 disassembler is much better at
981 disassembling SSE instructions. (thanks to Lutz Euler)
982 * enhancement: CHECK-TYPE on a variable now allows type-inference
983 (thanks to Vincent Arkesteijn)
984 * optimization: improved performance of EUC-JP external format.
985 (thanks to NIIMI Satoshi)
986 * optimization: performance improvements to IO on file streams of
987 :ELEMENT-TYPE CHARACTER
988 * optimization: much faster memory allocation on x86-64
989 * optimization: faster garbage collections (latency and throughput)
991 * optimization: faster bignum subtraction and division on x86 and x86-64
992 (thanks to Lutz Euler)
993 * optimization: more accurate type-inference for loops that use a
994 floating point index variable or a negative step.
996 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
997 * bug fix: add a workaround to SBCL looping infinitely at startup on
998 Linux kernels with apparently buggy implementations of personality().
999 (thanks to Svein Ove Aas)
1000 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1001 (thanks to David Lichteblau)
1002 * bug fix: local bindings shadow global symbol macros for
1003 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1004 * optimization: non-open coded uses of numeric comparison operators
1005 (e.g. >) no longer cons when called with more than one parameter
1006 on platforms supporting dynamic-extent allocation.
1007 * enhancement: saving cores with foreign code loaded is now supported
1008 on MIPS/Linux in addition to the previously supported platforms.
1009 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1010 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1011 * bug fix: division by zero in sb-sprof when no samples were collected
1012 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1013 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1014 (thanks to Alexander Kjeldaas)
1015 * new feature: ignoring the return values of destructive functions
1016 like NREVERSE signals a compile-time style-warning.
1017 (thanks to Kevin Reid)
1018 * new feature: the alignment of alien structure fields can be
1019 explicitly specified. (thanks to Cyrus Harmon)
1020 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1021 and dump core on SIGQUIT
1023 ** incompatible change: threads do not inherit values of specials
1024 from their parents (see manual)
1025 ** bug fix: threads stacks belonging to dead threads are freed by the
1026 next exiting thread, no need to gc to collect thread stacks anymore
1027 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1028 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1029 ** bug fix: dynamic variable and thread start related gc lossage
1030 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1032 * fixed some bugs revealed by Paul Dietz' test suite:
1033 ** SUBTYPEP is slightly more accurate on heinously complicated
1034 CONS types where some of the members have uncertain (in the
1035 NIL, NIL sense) type relationships to each other.
1037 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1038 no longer purified unless :PURIFY T is explicitly specified.
1039 ** Non-purified cores are significantly smaller than before
1041 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1042 * new feature: timers based on Zach Beane's excellent timer package
1043 * added support for the following external formats: koi8-u,
1044 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1045 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1046 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1047 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1048 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1049 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1050 * incompatible change: a threaded SBCL will no longer revert to
1051 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1052 * bug fix: interrupts are disabled until startup is complete; no
1053 more sigsegvs when receiving a signal to soon
1054 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1056 * bug fix: add a workaround for the memory randomization features in
1057 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1058 workaround will only be in effect on systems with the proc filesystem
1060 * bug fix: printing objects of type HASH-TABLE signals a
1061 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1063 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1064 have been running for over 50 days. (reported by Gilbert Baumann)
1065 * bug fix: the logic for getting names of functions gets less
1066 confused when confronded with alternate-metaclass
1067 funcallable-instances. (reported by Cyrus Harmon)
1068 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1069 consistent, even on internal alternate-metaclass objects.
1070 * bug fix: flush closure information collected by physical
1071 environment analysis prepass before the main pass. (bug reported
1073 * bug fix: compiler pack phase does not modify a hashtable, which is
1074 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1075 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1076 may be expanded. (reported by Alexander <malishev> on c.l.l)
1077 * bug fix: no more style-warnings when compiling code using LAST
1078 under high-SPEED low-SPACE settings. (reported by David Wallin
1080 * fixed some incompatibilities between SBCL's MOP and the MOP
1082 ** the METAOBJECT class is now implemented;
1083 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1084 STANDARD-OBJECT, as required;
1085 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1086 now compatible (as required by VALIDATE-SUPERCLASS); there
1087 remains a consistency requirement over the presence or absence
1088 of the FUNCTION class in the superclass list of finalized
1089 classes; see the manual for more details;
1090 ** the :AROUND method for COMPUTE-SLOTS on
1091 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1092 requested slot ordering.
1094 ** bug fix: parent thread now can be gc'ed even with a live
1096 ** bug fix: binding a special with PROGV to no value is not
1097 the same as that symbol not having been bound (thanks to
1099 ** bug fix: binding specials is thread safe (thanks to
1101 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1102 and SB-SPROF do not die with 'no handler for signal XX in
1103 interrupt_handle_now(..)' anymore
1104 ** bug fix: WITH-TIMEOUT works with multiple running threads
1105 * fixed some bugs revealed by Paul Dietz' test suite:
1106 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1107 the :method-class keyword argument.
1109 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1110 * new port: the Solaris operating system on x86 processors is now
1111 mostly supported, though some rough edges in the environment
1112 remain. (thanks to Daisuke Homma)
1113 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1114 space for its heap. (thanks to Thiemo Seufer)
1115 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1116 closures. (thanks to Thiemo Seufer)
1117 * minor incompatible change: eof selects abort in the debugger.
1118 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1119 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1121 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1122 deprecated, and will go away in a future revision of SBCL.
1123 * minor incompatible change: GC-ON and GC-OFF are no longer
1124 implemented with a counter, it does not matter how many times gc
1125 is switched on or off
1126 * bug fix: discriminating functions for generic function classes
1127 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1128 make invalid assumptions about method precedence order. (reported
1130 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1131 platforms. (thanks to Thiemo Seufer)
1132 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1133 * bug fix: correct pathname printing with printer escaping is on.
1134 (thanks to Kevin Reid)
1135 * bug fix: complex VOP definitions in "user-space" no longer trigger
1136 package locks. (reported by Zach Beane)
1137 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1138 for funcallable-instance functions, and is no different from
1140 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1141 characters in the current readtable. (reported by Nicholas Neuss)
1142 * bug fix: SUBTYPEP on various CONS types returns more a more
1143 accurate acknowledgment of its certainty.
1144 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1145 lists in some cases. This partially fixes bug 384.
1146 * flush all standard streams before prompting in the REPL and the
1148 * bug fix: signal handling and triggering gc do not conflict
1149 directly anymore, in particular a high frequency sb-sprof does
1150 not prevent gc from running
1151 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1152 approximation for timezone and DST information between the
1153 universal time epoch and the smallest negative 32-bit time_t.
1154 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1155 year 1899 so long as the encoded time is non-negative.
1156 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1157 the correct answer for non-constant rotate amounts when performing
1158 an inline 32-bit rotation.
1160 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1161 there is only one thread in the session
1162 ** bug fix: memory leak for streams created in one thread and
1163 written to in another
1164 ** bug fix: lockup when compiled with gcc4
1165 ** bug fix: race that allows the gc to be triggered when gc is
1167 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1168 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1170 ** bug fix: finished threads are gc'ed properly
1171 * fixed some bugs revealed by Paul Dietz' test suite:
1172 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1173 has arguments with a different set of applicable methods from
1174 the orignal arguments.
1175 ** The type error thrown by MAP now has a correct expected-type
1177 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1178 name a compiled function.
1179 ** Three MISC tests where a large bignum was improperly coerced to
1180 a float that couldn't represent that bignum during type
1181 derivation were fixed.
1182 ** SUBTYPEP can now handle types involving the negation of a
1183 list-form FUNCTION type.
1184 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1185 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1186 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1188 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1189 * New feature: Experimental support for bivalent streams: streams
1190 opened with :element-type :default now allow character and binary
1191 (unsigned-byte 8) I/O
1192 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1193 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1195 * bug fix: no more highly sporadic "couldn't check whether ~S is
1196 readable" when reading a stream and an interrupt hits in the middle
1197 of a select system call
1198 * compiler better recognizes complex arrays (reported by Eduardo
1200 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1201 for variables on SBCL built with linkage-tables. (reported by Luis
1203 * various error reporting improvements.
1204 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1205 (Thanks to Thiemo Seufer)
1206 * bug fix: nested reader invokations work correctly
1207 * bug fix: it is possible to have simultaneous references to foreign
1208 code and foreign data with the same name.
1210 ** added x86-64 support
1211 ** incompatible change: the threading api now works with thread
1212 objects instead of thread ids
1213 ** bug fix: threads are protected from signals and interruption when
1214 starting up or going down
1215 ** bug fix: a race where an exiting thread could lose its stack to gc
1216 ** bug fix: don't halt on infinite error in threads if possible
1217 ** fixed numerous gc deadlocks introduced in the pthread merge
1218 ** bug fix: fixed thread safety issues in read and print
1219 ** bug fix: debugger doesn't hang on session lock if interrupted at
1220 an inappropriate moment
1221 ** bug fix: run-program is now thread safe(r)
1222 ** bug fix: inner with-recursive-lock no longer releases the mutex
1223 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1224 * fixed some bugs revealed by Paul Dietz' test suite:
1225 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1226 EXPECTED-TYPE slots filled.
1227 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1229 ** MAKE-CONDITION accepts classes as type-designators.
1230 ** COMPILE may never return NIL.
1231 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1232 range before calling Unix time functions
1234 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1235 * numerous signal handling fixes to increase stability
1236 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1237 * minor incompatible change: we now correctly canonize default
1238 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1239 as per the MOP, rather than the historical (INITARG INITFUNCTION
1240 INITFORM). (reported by Bruno Haible)
1241 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1242 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1243 for more information.
1244 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1245 pathname is a directory pathname.
1246 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1247 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1249 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1250 :METACLASS argument in addition to a class name. (reported by
1251 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1252 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1253 (reported by Bruno Haible; patch from Gerd Moellmann)
1254 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1256 * bug fix: more cleanups to the floating point exception handling on
1257 x86-64 (thanks to James Knight)
1258 * bug fix: the compiler does not try to constant fold calls of
1259 COERCE and BIT-* functions when they return freshly constructed
1260 objects. (reported by Nikodemus Siivola and Paul Dietz)
1261 * optimization: improved the MIPS versions of generic (in the
1262 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1263 * optimization: direct conversion of (unsigned-byte 32) to floats on
1264 the PowerPC platform.
1265 * optimization: structure instances with raw slots now use less
1266 memory, and probably show better memory locality. (thanks to
1268 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1269 are treated as requests for stack allocation on the x86-64,
1270 Alpha-32, PPC and SPARC platforms.
1271 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1272 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1273 (thanks to Tony Martinez)
1275 ** gcing a dead thread can no longer lead to lockups
1276 ** threads block signals until they are set up properly
1277 ** errno is no longer shared by threads
1278 ** interrupt-thread restores the eflags register on x86
1279 ** fixed some lockups due to gc/thread interaction
1280 * fixed some bugs revealed by Paul Dietz' test suite:
1281 ** invalid dotted lists no longer raise a read error when
1282 *READ-SUPPRESS* is T
1283 ** signal an error if a symbol that names a declaration is used
1284 as the name of a type, or vice versa
1285 ** allow using the (declare (typespec var*)) abbreviation for
1286 (declare (type typespec var*)) with all type specifiers
1287 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1288 ** always return NIL from PPRINT-POP when OBJECT is NIL
1289 ** don't signal errors when pretty-printing malformed LABELS,
1290 FLET or MACROLET forms
1291 ** declarations in a DOLIST body are also in scope for the
1293 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1295 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1298 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1299 * fixed cross-compiler leakages that prevented building a 32-bit
1300 target with a 64-bit host compiler.
1301 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1302 opened with :IF-EXISTS :OVERWRITE.
1303 * fixed bug 281: error for an invalid qualifier in a short-form method
1304 combination method is not signalled until the faulty method is called.
1305 * bug fix: iteration variable type inferrer failed to deal with open
1306 intervals. (reported by Alan Shields)
1307 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1308 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1309 or more runtime options were provided to the sbcl binary.
1310 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1311 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1312 * optimization: there's now a fast-path for fixnum arguments in the
1313 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1314 * optimization: the code generated on x86-64 is more compact thanks
1315 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1316 * bug fix: floating-point exception handling now partly works on
1317 x86-64. (Thanks to James Knight)
1318 * improvement to the MIPS backend from Thiemo Seufer: C-style
1319 64-bit long long arguments and return values to alien functions
1321 * fixed some bugs revealed by Paul Dietz' test suite:
1322 ** the type-error signalled from WARN has a filled-in DATUM slot.
1323 ** the type-error required when a stream is not associated with
1324 a file has the stream as its datum.
1325 ** type-errors on single-floats on x86-64 no longer have
1326 :INVALID-OBJECT as the datum
1327 ** the type-errors signalled for invalid function names now have
1328 a correct expected type
1329 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1330 for structures defined with a :TYPE. Documentation strings for
1331 typed structures are no longer immediately discarded
1332 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1333 broadcast streams and synonym streams. FILE-LENGTH now also works
1334 on broadcast streams.
1336 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1337 * incompatible change: the --noprogrammer option, deprecated since
1338 version 0.7.5, has been removed. Please use the equivalent
1339 --disable-debugger option instead.
1340 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1342 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1343 inoperational for a while now) has been completely removed.
1344 * null lexical environments are now printed as #<NULL-LEXENV>,
1345 significantly reducing the amount of clutter in typical backtraces.
1346 * documentation on weak pointers, finalization, and after GC hooks
1347 has been added to the manual.
1348 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1349 as other specialized array types, is much faster. SUBSEQ and
1350 COPY-SEQ on such arrays have also been sped up.
1351 * optimization: EQL is now more efficient when at least other argument
1352 is known to be of type (OR FIXNUM (NOT NUMBER)).
1353 * fixed bug: the runtime is now less vulnerable to changes in the
1354 size of the SBCL object on OS X, and virtual memory is reserved for
1355 all spaces that need to be at a fixed address.
1356 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1357 * fixed bug: finalizers and after GC hooks that cause consing are now
1359 * fixed bug: compiler error messages and summaries are now printed to
1360 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1361 * fixed inference of the upper bound of an iteration variable.
1362 (reported by Rajat Datta).
1363 * fixed bug 211e: calling local functions with duplicated constant
1364 keyword argument no longer causes a bogus style warning about an
1366 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1367 ftype declaration to be disregarded. (reported by Dave Roberts)
1368 * fixed bug 373: caused by erronous compilation of references to alien
1369 variables in the runtime on ppc/darwin.
1370 * fixed bug 376: CONJUGATE type deriver.
1371 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1372 (reported by Baughn on #lisp)
1373 * a cross-compiler bug on non-x86 platforms has been identified and
1374 fixed. (thanks to Bruno Haible)
1375 * improvements to the MIPS runtime code for increased stability.
1376 (thanks to Thiemo Seufer)
1377 * increased the maximimum compact environment size to allow
1378 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1379 * partial workaround for bug 135: don't name the function we're
1380 calling for hairy cases of EVAL, so as not to accumulate one environment
1381 entry per call to EVAL. (reported by Kevin Reid)
1382 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1383 * optimization: type testing for non-vector arrays should be faster.
1384 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1386 * bug fix: setting 31st element of a bit vector to zero did not work
1388 * bug fix: redefining a class definition which failed due to a
1389 previous accessor / function clash now works (but see BUGS entry
1390 #380 for more problems in this area). (thanks to Zach Beane)
1391 * the long form of DEFINE-METHOD-COMBINATION disables method group
1392 checking when given a single method group with pattern *. (thanks
1394 * on x86 compiler supports stack allocation of results of simple
1395 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1396 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1397 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1398 * fixed some bugs related to Unicode integration:
1399 ** the restarts for recovering from input and output encoding
1400 errors only appear when there is in fact such an error to
1402 * fixed some bugs revealed by Paul Dietz' test suite:
1403 ** MISC.549 and similar: late transformation of unsafe type
1404 assertions into derived types caused unexpected code
1406 ** SCALE-FLOAT type deriver is less wrong.
1407 ** type derivers for EXP, LOG and similar functions compute result
1408 types for complex arguments better.
1409 ** (MISC.563) CONJUGATE type deriver works for very restricted
1411 ** out-of-line type testers for character strings are available.
1412 ** EQUAL compiler transform understands specialness of objects
1414 ** accessing double-floats stored on the stack now works on x86-64.
1415 ** debugger internals could sometimes create invalid lispobjs,
1416 resulting in GC crashes.
1417 ** MISC.548: type check weakening can convert required type into
1419 ** initialization forms for bindings are not in scope of free special
1422 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1423 * incompatible change: thread support for non-NPTL systems has
1424 been removed - locking is buggy and unreliable. A threaded
1425 SBCL build will now warn at startup and refuse to create
1426 new threads, unless futex support is detected in the kernel
1427 * incompatible change: the top level REPL now has only an ABORT
1428 restart associated with it, not TOPLEVEL and ABORT as it used to.
1429 TOP and TOPLEVEL are now available as debugger commands for
1430 returning to the top level.
1431 * incompatible change: forms evaluated in the REPL now use the
1432 global optimization policy.
1433 * incompatible change: user- and system-initialization files are
1434 no longer processed with LOAD, but by READ and EVAL; hence the
1435 global optimization policy, startup package, readtable, etc,
1437 * The .fasl file format number has been incremented because of
1438 various incompatible changes.
1439 * internal entry point details and argument counts no longer appear
1440 in backtraces unless explicitly requested by setting
1441 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1442 * built-in and standard functions no longer have names like "top
1443 level local call to FOO".
1444 * fixed bug 32: functions defined in non-null lexical environments
1445 now have more legible printed representation
1446 * fixed bug 33: functions defined in non-null lexical environemnts
1447 are now more amenable to inspection by INSPECT.
1448 * workaround for bug 354: XEPs no longer appear in backtraces unless
1449 explicitly requested.
1450 * fixed bug: receiving the signal which results from attempting to
1451 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1452 SIGBUS on other *BSDs) on architectures where the C stack is also
1453 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1454 to debugger confusion or wild instability. (reported by Cheuksan
1456 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1457 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1458 (reported by Lutz Euler)
1459 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1460 compiled code. (reported by Paul Dietz)
1461 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1462 specialized parameter to a method to something that is not TYPEP
1463 the specializer is now possible.
1464 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1465 face of package deletion.
1466 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1467 pathname. (reported by David Tolpin on c.l.l)
1468 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1469 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1470 STRING-STREAM and FILE-STREAM, respectively.
1471 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1472 than previously; in addition, its test suite is now run on build.
1473 (thanks to Luke Gorrie)
1474 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1475 * optimization: added a immediate representation for single-floats
1477 * fixed some bugs revealed by Paul Dietz' test suite:
1478 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1479 ** attempting to create a package with a colliding nickname causes
1480 correctable errors to be signalled.
1481 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1482 ** several x86-64 backend bugs related to sign-extension of immediate
1485 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1486 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1487 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1489 * fixed loading of multiply forward-referenced layouts.
1490 (thanks to Cheuksan Wang)
1491 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1492 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1493 (see COMPILE-FILE documentation for details of :PRINT option)
1494 * fixed bugs 19 and 317: fixed-format floating point printing is
1495 more accurate. This also fixes a bug reported by Adam Warner
1496 related to the ~@F format directive.
1497 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1499 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1500 dispatch table if the from-char is a dispatch macro character.
1501 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1502 by the wordsize no longer ignore the last word. (reported by Lutz
1504 * fixed bug in type checking of dynamic-extent variables. (reported
1506 * optimization: sequence traversal functions use their freedom to
1507 coerce function designators to functions.
1508 * optimization: code with many calls to CLOS methods specialized on
1509 CLOS classes has had redundant type checks removed.
1510 * optimization: added declarations to speed up operations that access
1511 the internal character database (for example STRING-UPCASE)
1512 * optimization: comparison operations between floats and sufficiently small
1513 fixnums no longer create extra rationals
1514 * fixed some bugs related to Unicode integration:
1515 ** portions of multibyte characters at the end of buffers for
1516 character-based file input are correctly transferred to the
1517 start of the buffer at the next read.
1518 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1519 passing it through to OPEN.
1520 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1521 argument given to internal calls to OPEN.
1522 ** the built-in comment readers (introduced by character sequences
1523 ";" and "#|") are more forgiving to encoding errors; they will
1524 STYLE-WARN and then attempt to resync the stream at a character
1525 boundary. (thanks to Teemu Kalvas)
1526 * fixed some bugs in the x86-64 port:
1527 ** Negative short int return values from c-calls are sign-extended
1529 ** The stack is aligned to 16-bytes for c-calls, as required by
1530 the ABI. (thanks to Cheuksan Wang)
1531 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1532 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1533 for enums. (thanks to Vincent Arkesteijn)
1534 ** Multiple small optimizations and bugfixes for floating point
1536 * fixed some bugs revealed by Paul Dietz' test suite:
1537 ** Space, Tab, Linefeed, Return and Page have the invalid
1538 secondary constituent character trait.
1539 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1541 ** WITH-INPUT-FROM-STRING should only update the index place on
1543 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1544 works more reliably.
1545 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1546 with one-dimensional arrays with fill pointers.
1547 ** TYPE-OF failed on a complex with an integer realpart and a
1549 ** compiler failure during type inference for the code of form
1550 (IF (EQL X (THE ...)) ...) (MISC.535).
1552 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1553 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1554 platform. The port supports 61-bit fixnums, large memory spaces
1555 and reloading shared object files.
1556 * enhancement: saving cores with foreign code loaded is now
1557 supported on ppc/Darwin in addition to the previously supported
1559 * enhancement: the statistical profiler now walks deeper into the
1560 call stack for more meaningful call-graphs and accrued time
1561 reports (x86/x86-64 only). It also now reports time spent in
1563 * enhancement: it is now possible to trace most individual methods
1564 of a generic function in addition to tracing the generic function
1566 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1568 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1569 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1570 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1571 produces an error. (thanks to Vincent Arkesteijn)
1572 * bug fix: NAMESTRING on pathnames with :WILD components in their
1573 directories works correctly. (thanks to Artem V. Andreev)
1574 * fixed bug 125: compiler preserves identity of closures. (reported
1576 * bug fixed: functions with &REST arguments sometimes failed with
1577 "Undefined function" when compiled with (DEBUG 3). (reported by
1579 * bug fix: overflow during compiling of setting element of a bit
1580 vector with constant index and value. (reported by Timmy Douglas)
1581 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1582 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1583 * on x86 compiler partially supports stack allocation of dynamic-extent
1585 * GO and RETURN-FROM do not check the extent of their exit points
1586 when compiled with SAFETY 0.
1587 * fixed some bugs related to Unicode integration:
1588 ** encoding and decoding errors are now much more robustly
1589 handled; it should now be possible to recover even from invalid
1590 input or output to the terminal. (thanks to Teemu Kalvas)
1591 ** provided a first cut at implementing STRING-TO-OCTETS and
1592 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1593 ** altered the SB-MD5 contributed module slightly, changing the
1594 interface just enough for it to be supportable for builds where
1595 lisp characters are not eight bits.
1596 * fixed some bugs revealed by Paul Dietz' test suite:
1597 ** the FORMATTER-generated functions for ~V[ conditionals require
1598 the correct number of arguments.
1599 ** READ-FROM-STRING returns the mandated second value when applied
1600 to displaced strings.
1601 ** the #\Rubout and #\Backspace characters are treated as invalid
1602 constituent characters by the tokenizer.
1604 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1605 * new feature: reloading changed shared object files with
1606 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1607 * new feature: references to foreign variables and functions
1608 can now be compiled and loaded before the corresponding shared
1609 object file is loaded, as long as the foreign definitions are
1610 available at runtime.
1611 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1612 supported. (thanks to Dan Debertin)
1613 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1614 just keywords. (thanks to Vincent Arkesteijn)
1615 * fixed bug #331: structure-class instances corresponding to
1616 DEFSTRUCT forms are now created eagerly.
1617 * fixed bug #345: backtraces from calls to undefined functions work
1618 on x86 as well. Related bug #61 is now also partially fixed on x86
1619 (backtraces from throws to unknown catch tags.)
1620 * bug fix: lambda-list parsing is now stricter vrt. order and number
1621 of lambda-list keywords.
1622 * bug fix: as specified by AMOP, an error is signalled if a
1623 class-option appears multiple times in a DEFCLASS form. (reported
1625 * bug fix: RANDOM can be compiled when the compiler derives the type
1626 of its numeric argument as a disjoint set of small integers.
1627 (reported by Paul Dietz)
1628 * bug fix: starting a core saved with shared objects loaded when
1629 those objects are not available no longer causes threaded SBCL to
1630 hang. (reported by Sean Ross)
1631 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1633 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1634 argument, which must always be NIL. (reported by Kalle Niemitalo)
1635 * bug fix: printing 1.0d+23 no longer results in an error.
1636 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1637 * bug fix: structure slot setters preserve evaluation order. (thanks
1639 * fixed some bugs related to Unicode integration:
1640 ** RUN-PROGRAM can allow its child to take input from a Lisp
1641 stream. (reported by Stefan Scholl)
1642 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1643 locales. (reported by Ken Causey)
1644 * fixed some bugs revealed by Paul Dietz' test suite:
1645 ** INCF, DECF and REMF evaluate their place form as specified in
1647 ** FORMATTER expands ~{ ~} iteration directives with V or #
1648 parameters correctly.
1649 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1650 consequent uses no arguments correctly.
1651 ** the system has a partial understanding of the (COMPLEX RATIO)
1653 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1656 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1657 * new feature: a build-time option (controlled by the :SB-UNICODE
1658 keyword feature, enabled by default) for building the system with
1659 support for the entire 21-bit character space defined by the
1661 * new feature: the system now has rudimentary external-format
1662 support; the primary user-visible change at this time is that
1663 characters with the high bit set (such as lower-case-e-acute) will
1664 print correctly to a terminal in a UTF-8 environment.
1665 * minor incompatible change: BASE-CHAR no longer names a class;
1666 however, CHARACTER continues to do so, as required by ANSI.
1667 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1668 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1669 has been moved to the SB-EXT package (temporarily re-exported from
1671 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1672 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1674 * fixed bug #351: better error-handlind and reporting for malformed
1676 * fixed bug #350: bignum-printing is now more memory-efficient,
1677 allowing printing of very large bignums, eg. (expt 2 10000000).
1678 (reported by Bruno Haible)
1679 * fixed bug #302: better primitive-type selection for intersection
1681 * fixed bug #308: non-graphic characters now all have names, as
1682 required. (reported by Bruno Haible)
1683 * bug fix: redefining a class with different superclasses now correctly
1684 removes it from the direct-subclasses of its previous superclasses.
1685 (reported by David Morse)
1686 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1687 new value now works. (reported by Bruno Haible)
1688 * bug fix: correct canonicalization of multiple non-standard slot
1689 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1690 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1691 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1692 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1693 now exists, an signals an error.
1694 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1695 during macroexpansion for non-iteration variables as well. (reported
1696 by Bruno Haible for CMUCL)
1697 * bug fix: Cyclic structures and unprintable objects in compiler
1698 messages no longer cause errors. (reported by Bruno Haible)
1699 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1700 and READ-FROM-STRING all now return a primary value of NIL if
1701 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1702 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1703 (reported by Bruno Haible for CMUCL)
1704 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1705 do not designate a condition. (reported by Bruno Haible for
1707 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1708 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1709 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1710 permitting resolution in favour of any of the conflicting symbols.
1711 (reported by Bruno Haible for CMUCL)
1712 * bug fix: EQUAL compiler optimizations is less aggressive on
1713 strings which can potentially compare true despite having distinct
1714 specialized array element types.
1715 * bug fix: unit enumerations can be defined without dividing by
1716 zero. (thanks to Vincent Arkesteijn)
1717 * FORMAT compile-time argument count checking has been enhanced.
1718 (report from Bruno Haible for CMUCL)
1719 * a partial workaround for the bug 262: the compiler does not try to
1720 inline-expand a local function doing RETURN-FROM from a deleted
1721 BLOCK. (thanks to Peter Denno for the bug report and to David
1722 Wragg for the simple test case)
1723 * fixed some bugs revealed by Paul Dietz' test suite:
1724 ** RENAME-PACKAGE allows all package designators as new package
1726 ** constraint propagation and lambda variable substitution are
1727 more cautious in dealing with partially deleted code.
1728 ** compiler sometimes lost reoptimization passes.
1729 ** CERROR, given a condition as condition designator, passes its
1730 remaining arguments to the continue format control without
1732 ** Case-altering FORMAT directives work correctly on non-ASCII
1734 ** The REAL type specifier handles bounds outside the floating
1735 point ranges without signalling FLOATING-POINT-OVERFLOW.
1736 ** Functions with IR1-transformations can create intercomponent
1737 references to global functions.
1738 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1740 ** FORMAT ~R treats a nil value for its first parameter correctly.
1742 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1743 * enhancement: saving cores with foreign code loaded is now
1744 supported on x86/NetBSD and sparc/Linux in addition to the previously
1745 supported platforms.
1746 * bug fix: on some platforms repeated installations caused multiple
1747 copies of HTML documentation to be installed -- should not happen
1748 any more. (reported by Stefan Scholl)
1749 * bug fix: parsing self-recursive alien record types multiple times
1750 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1751 original patch by Helmut Eller for CMUCL)
1752 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1753 (thanks to Richard Kreuter)
1754 * bug fix: defining classes whose accessors are methods on existing
1755 generic functions in other (locked) packages no longer signals
1756 bogus package lock violations. (reported by François-René Rideau)
1757 * bug fix: special variables as DEFMETHOD parameters no longer have
1758 associated bogus type declarations. (reported by David Wragg and
1760 * bug fix: read-write consistency on streams of element-type
1761 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1762 * bug fix: redefiniton of the only method of a generic function with
1763 no DEFGENERIC no longer emits a full WARNING. In addition,
1764 redefinition of generic functions with no DEFGENERIC to an
1765 incompatible lambda list now signals an error. (thanks to Zach
1767 * bug fix: DEFGENERIC now works even when there's a function of the
1768 same name in an enclosing lexical environment. (thanks to Zach
1770 * fixed compiler failure, caused by instrumenting code during
1771 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1772 * optimization: added loop analysis and improved register allocation
1773 * fixed some bugs revealed by Paul Dietz' test suite:
1774 ** POSITION on displaced vectors with non-zero displacement
1775 returns the right answer.
1776 ** (SIMPLE-STRING) is a valid type specifier for sequence
1778 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1780 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1781 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1783 ** The justification version of the FORMAT ~< directive treats
1784 non-zero minpad parameter correctly.
1786 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1787 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1788 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1789 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1790 the supported interface.
1791 * new feature: Single-stepping of code compiled with DEBUG 2 or
1792 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1793 * new feature: saving cores with foreign code loaded is now
1794 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1795 Timothy Moore's work for CMUCL)
1796 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1797 parameters to * instead of NIL if no initform is supplied.
1798 (reported by Johan Bockgaard)
1799 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1800 correctly when FUNCALL appears as the car of the form. Note:
1801 despite this FUNCALL forms are not currently subject to
1802 compiler-macro expansion. (port of Raymond Toy's fix for the
1803 same from CMUCL, reported by Johan Bockgaard)
1804 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1805 (thanks for Teemu Kalvas)
1806 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1807 case instead of terminating the iteration (thanks for Julian
1808 Squires, Sean Champ and Raymond Toy)
1809 * bug fix: incorrect expansion of defgeneric that caused a style
1810 warning. (thanks for Zach Beane)
1811 * bug fix: slot accessor effective method computation works properly
1812 for classes with multiple non-standard applicable methods on
1813 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1814 * on x86 compiler supports stack allocation of results of LIST and
1815 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1816 CMUCL implementation by Gerd Moellmann)
1817 * fixed some bugs revealed by Paul Dietz' test suite:
1818 ** FORMAT strings with both the ~<~:;~> form of the justification
1819 directive and pretty-printing directives cause an error.
1821 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1822 * incompatible change: the internal functions
1823 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1824 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1825 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1826 instead of the old functions.
1827 * new feature: on platforms where "dladdr" is available foreign
1828 function names now appear in backtraces. (based on Helmut Eller's
1830 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1831 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1833 * documentation: ASDF manual is now distributed with SBCL.
1834 * bug fix: SBCL can now load its contributed modules with REQUIRE
1835 even if the system-provided entries have been removed from
1836 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1838 * bug fix: backtraces involving undefined functions or assembly
1839 routines are more informative. (thanks to Brian Downing)
1840 * bug fix: mutually referent alien structures now work correctly.
1841 (reported by Rick Taube)
1842 * bug fix: structures defined by WITH-ALIEN can be referred to
1843 within other definitions in the same WITH-ALIEN.
1844 * bug fix: division operators (MOD, TRUNCATE and the like) with
1845 constant zero divisors and integer dividends no longer generate
1847 * bug fix: provide default methods for INPUT-STREAM-P and
1848 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1849 * bug fix: improve the detection and resolution of MOP metacycles
1850 (where effective method computation depends on the generic function
1851 having its effective method computed). (reported by Bruno Haible)
1852 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1853 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1854 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1855 * bug fix: pathnames with relative directory components can be
1856 represented relative to default pathnames.
1857 * optimization: in taking the GCD of bignums, reduce the two bignums
1858 to approximately the same size (using Euclid's algorithm) before
1859 applying the more sophisticated binary GCD. (thanks to Juho
1861 * optimization: COUNT on bitvectors now operates word-at-a-time.
1862 * optimization: ASH with a positive, but not necessarily constant,
1863 (leftwards) shift, when in a modular context, is compiled to a
1865 * fixed some bugs revealed by Paul Dietz' test suite:
1866 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1867 if the corresponding argument is NIL.
1868 ** FORMAT directives accepting a minpad parameter treat negative
1870 ** literal commas as character parameters to FORMAT directives are
1872 ** literal spaces directly after ~<Newline> directives within a
1873 format-logical-block (~:< ~@:>) do not induce :FILL-style
1874 conditional newlines.
1875 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1876 as specified: it no longer includes conditional newlines).
1877 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1879 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1880 :PER-LINE-PREFIX argument does not evaluate to a string.
1882 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1883 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1884 the manual for details; for now, package locks can be disabled by
1885 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1886 if no major problems are found then it is likely that they will be
1887 compiled in unconditionally.
1888 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1889 unsupported operators on all platforms. To load a shared library
1890 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1891 object file, link it into a shared library outside of SBCL and
1892 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1894 * bug workaround: ROOM T and the GC were not working together
1895 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1896 reporting facility aren't preserved. That reporting has been
1897 disabled, so now until and unless someone figures out how to make it
1898 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1899 * minor incompatible change: as threatened around sbcl-0.8.0, the
1900 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1901 The symbols are also exported from SB-PCL for backwards
1902 compatibility, but more so than before SB-PCL should be treated as
1903 an implementation-internal package.
1904 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1906 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1907 It is known as of this release not to work on the Alpha, however.
1908 * fixed bug #167: errors signalled due to illegal syntax in method
1909 bodies are now more legible.
1910 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1911 designators and can hence be used with TYPEP.
1912 * fixed bug #333: CHECK-TYPE now ensures that the type error
1913 signalled, if any, has the right object to be accessed by
1914 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1915 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1916 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1918 * fixed bug #334: programmatic addition of slots using specialized
1919 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1920 and :ALLOCATION :CLASS slots.
1921 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1922 float exponents. (rereported by Peter Seibel)
1923 * fixed a bug: #\Space (and other whitespace characters) are no
1924 longer considered to be macro characters in standard syntax by
1925 GET-MACRO-CHARACTER.
1926 * fixed bug: initialization of condition class metaobjects no longer
1927 causes an instance of the condition to be created. (reported by
1929 * fixed bug: it is now possible to have slots such that
1930 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1931 description is neither :INSTANCE nor :CLASS.
1932 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1933 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1934 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1935 non-standard SLOT-DEFINITION-ALLOCATION values.
1936 * fixed bugs: various race conditions handling exiting threads.
1937 CL-PPCRE's thread torture test now passes.
1938 * fixed arguably-a-bug: GC time proportional to number of threads in
1939 system even when most of them are idle
1940 * optimization: improved performance of BIT and SBIT on bit-vectors.
1941 * .fasl file incompatibility: The fasl file version number has
1942 been incremented because of changes associated with package locks.
1944 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1945 * minor incompatible change: the system no longer provides
1946 optimization hints (conditions which are TYPEP
1947 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1948 mode; these hints are still emitted when higher SPEED optimization
1950 * new contrib module: a sampling profiler (profiling by statistical
1951 sampling, rather than by instrumenting functions) is available as
1952 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1953 * the behaviour of the standard function ED is now customizeable by
1954 third parties through a hook variable: see ED's documentation
1955 string for information on the protocol.
1956 * the compiler no longer emits efficiency notes for (FUNCALL X)
1957 when the type of X is uncertain under default optimization
1959 * fixed bug 276: mutating a binding of a specialized parameter to a
1960 method to something that is not TYPEP the specializer is now
1962 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1963 exhibit the required behaviour on the x86 platform. (thanks to
1964 Peter van Eynde, Eric Marsden and Bruno Haible)
1965 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1966 even for difficult arguments. (reported by Peter Graves)
1967 * fixed bug 141a: the backquote printer now descends quoted
1969 * fixed another bug in backquote printing: no more destructive
1970 modification of the form's list structure. (reported by Brian
1972 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1973 move between its address being taken and the call to
1974 interrupt_thread, fixing a crashing race condition.
1975 * the SB-POSIX contrib implementation has been adjusted so that it
1976 no longer exhibits ridiculously poor performance when constructing
1977 instances corresponding to C structs.
1979 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1980 * minor incompatible change: the sb-grovel contrib now treats C
1981 structures as alien (in the sense of SB-ALIEN) objects rather than
1982 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1983 has implications for memory management of client code
1984 (sb-grovel-returned objects must now be manually managed) and for
1985 type safety (alien objects now have full types).
1986 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1987 used to control emission of compiler diagnostics, rather than the
1988 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1989 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1990 quality should be considered deprecated.
1991 * install.sh now installs the user manual as well
1992 * (not quite a new documentable feature, but worth considering in
1993 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1994 beginnings of a semantically meaningful condition hierarchy is
1995 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1997 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1998 designator as the defaults argument.
1999 * fixed bug: Displaced arrays whose displaced-to array has become
2000 too small now cause ARRAY-DIMENSION to signal an error, providing
2001 for safer bounds-checking. (reported by Bruno Haible)
2002 * fixed bug: DEFCLASS slot definitions with identical :READER and
2003 :WRITER names now signal a reasonable error. (reported by Thomas
2005 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2007 * fixed bug: Invalid slot specification errors now print correctly.
2008 (thanks to Zach Beane)
2009 * fixed bug 320: Shared to local slot value transfers in class
2010 redefinitions now happen corrently from superclasses as
2011 well. (reported by Bruno Haible)
2012 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2014 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2015 improper lists correctly. (reported by Bruno Haible)
2016 * fixed bug 313: source-transform for <fun-name> was erroneously
2017 applied to a call of a value of a variable with name <fun-name>.
2018 (reported by Antonio Menezes Leitao)
2019 * fixed bug 307: The obsolete instance protocol ensures that
2020 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2021 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2022 remove dead unknown-values globs from the middle of the stack.
2023 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2024 easier to resume long computations after using *BREAK-ON-SIGNALS*
2025 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2026 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2027 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2028 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2029 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2030 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2032 * fixed a (fairly theoretical) bug in string printing: if
2033 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2034 does not have array-element-type equal to the most general string
2036 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2037 function name. (thanks to Nikodemus Siivola)
2038 * fixed bug: compile-time format string checker failed on
2039 non-closed ~{. (reported by Thomas F Burdick)
2040 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2041 don't warn on legal loop code involving more than one aggregate
2042 boolean. (thanks to Nikodemus Siivola)
2043 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2044 in keyword position to MAKE-PACKAGE was not being reported as
2046 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2047 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2049 * fixed bugs 280 and 312: the checking for multiple definitions in a
2050 file is less likely to become confused by uses of inline
2052 * fixed bug: the #S reader macro performs the keyword coercion
2053 specified for slot names. (reported by Kalle Niemitalo)
2054 * fixed bug: lambda lists may contain symbols whose names start with
2055 & but are not lambda-list-keywords; their occurrence triggers a
2057 * fixed bug 321: define-method-combination argument lambda lists do
2058 not cause invalid code to be generated when &optional or &aux
2059 variables have default values. (reported by Bruno Haible)
2060 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2061 have CLOS classes; this appears also to have fixed reported
2062 instability in stack exhaustion detection.
2063 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2064 MUFFLE-WARNING when no associated restart is present is now
2066 * optimization: rearranged the expansion of various defining macros
2067 so that each expands into only one top-level form in a
2068 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2070 * optimization: used a previously little-used slot in symbols to
2071 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2073 * fixed some bugs revealed by Paul Dietz' test suite:
2074 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2076 ** Fixed an optimization bug related to inheritance of initforms
2077 from local to shared slots.
2078 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2079 ** CAST optimizer forgot to flush argument derived type.
2080 ** print/read consistency on floats is now orders of magnitude
2081 more likely. (thanks also to Bruno Haible for a similar report
2083 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2084 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2085 for objects of type REAL. Make it so.
2086 ** SXHASH is specified (infelicitously) to respect similarity,
2087 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2088 it so. (thanks to Markus Ziegler)
2089 ** on the Alpha, the compiler succeeds in compiling functions
2090 returning a known number of arguments greater than 63.
2091 ** fixed handling of invalid NIL arguments in keyword position
2093 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2095 ** provided a readably-printable representation for RANDOM-STATE
2097 ** ensured that pathnames, where they have a namestring, always
2098 print using #P"..." syntax.
2100 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2101 * Support for the forthcoming 2.0 version of the NetBSD kernel
2102 running on x86 hardware has been added. (thanks to Perry
2103 E. Metzger most immediately, and others for their past work)
2104 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2105 current release version; SBCL's OpenBSD support had been broken
2106 since about the time of OpenBSD's switch to ELF binary format.)
2107 * [placeholder for DX summary]
2108 ** user code with &REST lists declared dynamic-extent, under high
2109 speed or space and low safety and debug optimization policy.
2110 * The manual has been converted to Texinfo format and the debugger
2111 chapter from the cmucl manual has been added.
2112 * A facility has been added to extract documentation strings from
2113 sbcl and store them as Texinfo-formatted snippets for inclusion in
2114 the manual (via Texinfo's @include directive)
2115 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2116 the test case to Dave Roberts)
2117 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2118 pointers. (reported by Sean Ross)
2119 * bug fix: PROFILE output is printed nicely even for large numerical
2120 values. (thanks to Zach Beane)
2121 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2122 greater than 32 handle EOF correctly.
2123 * bug fix: on X86 an immediate argument of the IMUL instruction is
2124 correctly printed in disassembly. (thanks to Lutz Euler)
2125 * bug fix: class slots in redefined classes preserve their old
2126 values. (thanks to Bruno Haible and Nikodemus Siivola)
2127 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2128 succeeds. (reported by Marco Baringer)
2129 * fixed some bugs revealed by Paul Dietz' test suite:
2130 ** READ-SEQUENCE now works on ECHO-STREAMs.
2131 ** RATIONALIZE works more according to its specification. (thanks
2134 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2135 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2136 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2137 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2138 should matter to you only if you rebind the printer control
2139 variables and then find you want different bindings in the
2140 debugger than in the ordinary execution of your program.)
2141 * The runtime build system has been tweaked to support building
2142 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2143 assembler and linker. (thanks to Nikodemus Siivola)
2144 * Unbound, undefined, undeclared variables now trigger full
2145 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2146 more often programmer error than deliberate exploitation of undefined
2148 * optimization: the hash algorithm for strings has changed to one
2149 that is less vulnerable to spurious collisions. (thanks to Juho
2151 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2152 needless bounds checking. (thanks to Juho Snellman)
2153 * optimization: implemented multiplication as a modular
2154 (UNSIGNED-BYTE 32) operation on the PPC backend.
2155 * fixed some bugs revealed by Paul Dietz' test suite:
2156 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2158 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2159 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2160 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2162 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2164 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2166 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2167 their output stream on EOF from read.
2168 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2169 have been read to end-of-file.
2170 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2172 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2173 description of determination of which consecutive characters
2175 ** Printing the "Space" character with escaping on now yields "#\\ ",
2176 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2177 ** Reading floating-point numbers with *READ-BASE* set to a number
2178 less than 10 works correctly.
2179 ** Reading floating-point numbers with *READ-BASE* set to a number
2180 more than 10 works correctly.
2181 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2182 the readtable currently in effect.
2184 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2185 * minor incompatible change: parsing of namestrings on a physical
2186 (Unix) host has changed; numbers after the final #\. in a
2187 namestring are no longer interpreted as a version field. This is
2188 intented to be largely invisible to the user, except that the
2189 meaning of the namestring "*.*.*" has changed: it now refers to a
2190 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2191 should usually be replaced by
2192 (make-pathname :name :wild :type :wild :version :wild)
2193 with the added benefit that this is more likely to be portable.
2194 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2195 option to OPEN now signals an error if the file being opened
2196 exists; this may have an impact on existing code.
2197 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2198 signals received, on the PowerPC platforms (both Linux and
2199 Darwin). (thanks to Pierre Mai for pointing out the location of
2201 * several fixes on OS X: The system now builds and runs cleanly on
2202 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2203 (thanks to Brian Mastenbrook)
2204 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2205 second-resolution: integer multiples of 1/3600 between -24 and 24.
2206 (thanks to Vincent Arkesteijn)
2207 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2208 type when called with 1 argument; PEEK-CHAR checked type of
2209 PEEK-TYPE only after having read first character from a
2210 stream. (reported by Peter Graves)
2211 * bug fix: the garbage collector now has much better locality
2212 behaviour, and in particular no longer treats all memory as being
2213 exhausted when many small objects point to each other in a deeply
2215 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2216 recognized as being TYPEP their class.
2217 * bug fix: the PUSHNEW documentation string has been corrected.
2218 (thanks to Vincent Arkesteijn)
2219 * bug fix: defaulting of the value for the last of an atypically
2220 large number of multiple values being bound was not being
2221 performed correctly on the Alpha or PPC platforms
2222 * optimization: implemented multiplication as a modular
2223 (UNSIGNED-BYTE 32) operation on the x86 backend.
2224 * optimization: SEARCH on simple-base-strings can now be open-coded.
2225 (see also contrib/compiler-extras.lisp for inspiration for
2226 teaching the compiler about the Boyer-Moore algorithm).
2227 * value, returned by MAX (and MIN) called with several EQUALP, but
2228 not EQL, arguments now does not depend on compiler settings.
2229 * fixed some bugs revealed by Paul Dietz' test suite:
2230 ** in stack analysis liveness information is propagated from
2231 non-local entry points.
2232 ** pathwise CAST removing failed when the CAST node did not start
2234 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2235 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2237 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2238 host is already defined.
2239 ** RENAME-FILE works on streams instead of signalling an internal
2241 ** PEEK-CHAR uses the current readtable when determining whether
2242 or not a character is whitespace.
2243 ** MERGE-PATHNAMES handles the case when the pathname does not
2244 specify a name while the default-pathname specifies a version.
2245 ** Pathnames now stand a better chance of respecting print/read
2247 ** Attempting to use standardized file system operators with a
2248 pathname with invalid :DIRECTORY components signals a
2250 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2252 ** DIRECTORY on logical pathnames is more correct.
2253 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2254 signal a TYPE-ERROR if their argument is not a stream.
2255 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2256 designator argument does not designate a stream.
2257 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2258 examining the synonym.
2259 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2261 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2262 element-type larger than ([UN]SIGNED-BYTE 32).
2264 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2265 * When built with the :SB-FUTEX feature, threaded builds now take
2266 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2267 for faster/more reliable mutex and condition variable support.
2268 * Incompatible change (but one you probably shouldn't have been using
2269 anyway): the interface and code for arbitrating between multiple
2270 threads in the same user session has been redesigned.
2271 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2272 error when called without an explicit environment argument.
2273 (thanks to Nikodemus Siivola)
2274 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2275 confused about their position. (thanks to Adam Warner and Gerd
2277 * bug fix: Pretty printing backquoted forms with unquotations in the
2278 argument list position of various code constructs such as LAMBDA
2279 now works correctly. (reported by Paul Dietz)
2280 * bug fix: Pretty printing unquotations no longer loses all
2281 stream position information.
2282 * optimization: performance of string output streams is now less
2283 poor for multiple small sequence writes.
2284 * optimization: performance of CSUBTYPEP in the presence of complex
2285 expressions involving CONS and NOT many times has been improved.
2286 (reported by Paul Dietz)
2287 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2289 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2290 streams for the REPL, for the debugger, and so on. (thanks to
2292 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2293 * fixed some bugs revealed by Paul Dietz' test suite:
2294 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2295 is ignored if it is too small, rather than propagating through
2296 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2297 ** extremely complex negations of CONS types were not being
2298 sufficiently canonicalized, leading to inconsistencies in
2300 ** VALUES tranformer lost derived type.
2302 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2303 * fixed a bootstrapping bug: the build process no longer assumes
2304 that the various BOOLE-related constants have the same value in
2305 host and target lisps. (noted by Paul Dietz' test suite on an
2306 SBCL binary built from CLISP)
2307 * The system can now be dynamically linked on the MIPS platform,
2308 which enables dynamic loading of foreign code from Lisp. (thanks
2309 to Ralf Baechle for discussions on the MIPS ABI)
2310 * The system now records debugging information for its own source
2311 files in a filesystem-position-independent manner, relative to
2312 the "SYS" logical host.
2313 * fixed a compiler bug: MV-LET convertion did not check references
2314 to the "max args" entry point. (reported by Brian Downing)
2315 * tweaked disassembly notes to be less confident about proclaiming
2316 some instruction as an LRA. (thanks to Brian Downing)
2317 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2318 now each have their own history, command character, and other
2319 characteristics. (thanks to David Lichteblau)
2320 * fixed some bugs revealed by Paul Dietz' test suite:
2321 ** compiler failure in compiling LOGAND expressions including a
2323 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2324 shift greater than 32.
2325 ** FLUSH-DEST did not mark blocks for type check regeneration.
2326 ** HANDLER-CASE failed to accept declarations in handler clauses
2327 in some circumstances.
2329 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2330 * New code in contrib/sb-introspect (still probably not entirely
2331 stable yet) provides some support for smart Lisp development
2332 environments like SLIME.
2333 * The conditions signalled for errors occurring when loading .fasl
2334 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2335 in a way which should help ASDF recover gracefully.
2336 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2337 stuff has been cleaned up. If you code contrib/ stuff, this might
2338 affect you, and you can look at contrib/README, contrib/STANDARDS,
2339 and/or the 0.8.4.27 diff to check.
2340 * In full calls the compiler now does not generate checks for declared
2341 argument types for all arguments.
2342 * various threading fixes
2343 ** and some experimental patches which didn't make it into
2344 the main tree for this release, but which are shipped in
2345 contrib/experimental-thread.patch as a possible fix for some
2346 failures (deadlock, spinning...) in GC-intensive multithreaded
2348 * fixed PPC build problem (source code incompatibility of different
2349 library versions): added offsetof() hackery which attempts to divine
2350 where glibc maintainers put uc_mcontext today
2351 * fixed bug 282: compiler does not trust type assertions while passing
2352 arguments to a full call.
2353 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2354 &OPTIONAL VALUES type parameter.
2355 * fix bug 214: algorithm for noting rejected templates is now more
2356 similar to that of template seletion. (also reported by rydis on
2358 * fixed bug 141b: printing backquoted information readably and prettily
2359 inserts a space where necessary.
2360 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2361 (CHARACTER 10) are now reported as errors, rather than propagated
2362 as unknown types. (reported by piso on #lisp)
2363 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2364 with values NIL and :ERROR. (thanks to Milan Zamazal)
2365 * fixed bug 191c: CLOS now does proper keyword argument checking as
2366 described in CLHS 7.6.5 and 7.6.5.1.
2367 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2368 longer raise an error; further, using a list as a for-as-arithmetic
2369 counter now raises a meaningful error.
2370 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2371 now understood by sequence creation functions such as MAKE-SEQUENCE
2373 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2374 a STRING-INPUT-STREAM.
2375 * compiler enhancement: SIGNUM is now better able to derive the type
2377 * type declarations inside WITH-SLOTS are checked. (reported by
2379 * fixed some bugs revealed by Paul Dietz' test suite:
2380 ** incorrect optimization of TRUNCATE for a positive first
2381 argument and negative second.
2382 ** compiler failure in let-convertion during flushing dead code.
2383 ** compiler failure while deriving type of TRUNCATE on an
2384 interval, containing 0.
2385 ** ASH of a negative bignum by a negative bignum count now returns
2387 ** intersection of CONS types now canonicalizes properly, fixing
2388 inconsistencies in SUBTYPEP.
2390 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2391 * incompatible change: The --disable-debugger command line
2392 option now clobbers the debugger at a more fundamental
2393 level, by redefining #'INVOKE-DEBUGGER instead of by
2394 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2395 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2396 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2397 enter the debugger REPL and then suffer recursive errors
2398 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2399 while under the new system, BREAK in --disable-debugger mode
2400 terminates the system just as an unhandled error would.
2401 * fixed compiler performance when processing loops with a step >1;
2402 * bug fix: DOCUMENTATION now retrieves generic function
2403 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2404 support has been systematized, and now supports the methods
2405 specified by ANSI, along with a default method and a method for
2406 slot documentation. (reported by Nathan Froyd)
2407 * bug fix: effective methods associated with a generic function are
2408 no longer cached over a change of that generic function's method
2409 combination. (reported by Andreas Fuchs)
2410 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2411 names a non-existent directory. (thanks to Andreas Fuchs)
2412 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2413 the PPC platform, lead to stack corruption; this has been fixed.
2414 (reported by Rainer Joswig)
2415 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2416 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2417 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2418 arithmetic in general, is now much more reliable on the PPC
2420 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2421 platform now returns the right answer.
2422 * optimization: restored some effective method precomputation in
2423 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2424 precomputation is now tunable.
2425 * optimization: compiler-internal data structure use has been
2426 reviewed, and changes have been made that should improve the
2427 performance of the compiler by about 20%.
2428 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2429 simple-base-strings and simple-bit-vectors is improved.
2430 * optimization: the optimization of 32-bit logical and arithmetic
2431 functions introduced in version 0.8.3 on the x86 has been
2432 implemented on the mips, ppc and sparc platforms; an
2433 implementation of the same facility, but for 64-bit arithmetic,
2434 has been added for the alpha.
2435 * microoptimization: the compiler is better able to make use of the
2436 x86 LEA instruction for multiplication by constants.
2437 * bug fix: in some situations compiler did not report usage of
2438 generic arithmetic in (SPEED 3) policy.
2439 * bug 145b fix: compiler used wrong type specifier while converting
2440 MEMBER-types to numeric.
2441 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2443 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2444 index argument now works properly on the Alpha platform.
2445 * bug fix: floating point exception treatment on the Alpha platform
2447 * bug fix: FILE-POSITION works much better on string input and
2448 output streams. (thanks to Nikodemus Siivola)
2449 * bug fix: many threading/garbage collection symptoms sorted.
2450 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2451 might be pseudo-atomic.
2452 * internal change: Stopping for GC is now done with signals not
2453 ptrace. GC is now done in whichever thread wanted it, instead of
2455 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2457 * bug fix: result form in DO is not contained in the implicit
2459 * incompatible change: ICR structure is changed; the value part of
2460 CONTINUATION is now called LVAR; corresponding functions are
2461 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2462 * added type deriver for ISQRT (thanks to Robert E. Brown).
2463 * bug fix: better support for loading from the command line when an
2464 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2466 * fixed some bugs revealed by Paul Dietz' test suite:
2467 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2468 ** ROUND and FROUND now give the right answer when given very
2469 small float arguments.
2470 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2472 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2473 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2474 ** LCM should return a non-negative integer.
2475 ** PARSE-INTEGER returned the index of a terminator instead of the
2476 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2477 ** PARSE-INTEGER returned an incorrect index being applied to a
2479 ** LCM with two arguments of 0 returns 0 rather than signalling
2481 ** unsigned addition of a 32-bit constant with the high bit set no
2482 longer causes an internal compiler error on the x86.
2483 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2484 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2485 with negative last argument.
2486 ** byte specifiers with zero size and position no longer cause
2487 an error during type derivation.
2488 ** bignum multiplication on the Alpha platform now returns the
2490 * porting: The system now builds on SuSE AMD64, although it still
2491 generates a 32-bit binary.
2492 * .fasl file incompatibility: The fasl file version number has
2493 been incremented (because of the changes to internal compiler
2494 data structures referred to above).
2496 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2497 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2498 more accurately, on the Darwin kernel running on PowerPC hardware.
2499 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2500 * Compiler code deletion notes now signal a condition of type
2501 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2502 an associated MUFFLE-WARNING restart.
2503 * The compiler now performs limited argument count validation of
2504 constant format strings in FORMAT, and where appropriate in ERROR,
2505 CERROR and WARN. (thanks to Gerd Moellmann)
2506 * New ASDF-INSTALL contrib can be used for automatic download and
2507 installation of third-party Lisp code from CCLAN or other sites
2509 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2510 which forces another thread to execute a function supplied by the
2512 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2513 now accept and act upon their :ELEMENT-TYPE keyword argument.
2514 (reported by Martin Atzmueller, Edi Weitz)
2515 * bug fix: FILE-POSITION now accepts position designators up to
2516 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2517 is the greater. (thanks to Patrik Nordebo)
2518 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2519 Kalle Olavi Niemitalo)
2520 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2521 printable. (reported by Eric Marsden)
2522 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2523 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2524 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2525 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2526 (thanks to Stig Sandoe)
2527 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2528 UNREAD-CHAR now correctly includes the unread character in the
2529 target sequence. (thanks to Gerd Moellmann)
2530 * bug fix in threaded builds: the system can now be suspended and
2531 resumed by shell job control with minimal disruption.
2532 * bug fixes in times and timezones >2038AD
2533 * better handling of "where is GNU make?" problem in build scripts
2534 (thanks to Nikodemus Siivola)
2535 * new optimization: inside a named function any reference to a
2536 function with the same name is considered to be a self-reference;
2537 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2538 optimization quality.
2539 * new optimization on x86: logical functions and + now have
2540 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2541 used when the result is truncated to 32 bits.
2542 * VALUES declaration is partially enabled.
2543 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2544 * fixed some bugs revealed by Paul Dietz' test suite:
2545 ** The system now obeys the constraint imposed by
2546 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2547 types form a lattice under type intersection.
2548 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2549 ** ASSOC now ignores NIL elements in an alist.
2550 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2551 and (1+ MOST-POSITIVE-FIXNUM) answers.
2552 ** The addition of a method with invalid qualifiers to a generic
2553 function does not cause an error to be signalled immediately;
2554 a warning is signalled, and the error is generated only on
2555 calling the generic function.
2556 * changed .fasl file version number, in part to add type codes for
2557 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2558 obscure ANSI requirements
2560 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2561 * fixed bug 148: failure to inline-expand a local function left
2562 garbage, confusing the compiler.
2563 * fixed bugs 3cd: structure slot readers perform type check if the
2564 slot can have an invalid value (i.e. it is either not initialized
2565 or can be written with a less specific slot writer).
2566 * bug fix: the compiler now traps array references to elements off
2567 the end of an array; previously, the bounds checking in some
2568 circumstances could go off-by-one.
2569 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2571 * type declarations for array element types now obey the description
2572 on the CLHS page "Declaration TYPE", as per discussions on
2573 sbcl-help around 2003-05-08. This means that a declaration
2574 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2575 declaration, all references to BAR will be asserted or assumed
2576 (with THE, so dependent on compiler policy) to involve objects of
2577 type FOO. Note that no such declaration is implied in
2578 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2579 * declared types of functions from the "Conditions"
2580 chapter. (reported by Paul Dietz)
2581 * bug fix: CERROR accepts a function as its first argument.
2582 * bug fix: NTH an NTHCDR accept a bignum as index
2583 arguments. (reported by Adam Warner)
2584 * optimization: character compare routines now optimize comparing
2585 against a constant character. (reported by Gilbert Baumann)
2586 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2587 argument now works properly on the MIPS platform.
2588 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2589 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2590 declared to have a functional type, does not cause an error
2592 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2593 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2594 * fixed bug: if last continuation of a deleted block has a
2595 destination, this destination should be deleted too. (reported by
2597 * fixed a bug in the bootstrap process: the host compiler's values
2598 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2599 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2600 test case from Patrik Nordebo)
2601 * improved the ability of the disassembler on the PPC platform to
2602 provide helpful disassembly notes.
2603 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2604 the class in more cases than previously.
2605 * bug fix: FILE-POSITION now understands :START and :END for
2606 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2607 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2608 8) data. (thanks to David Lichteblau)
2609 * bug fix: it is possible to add a method to a generic function
2610 without lambda list.
2611 * bug fix: reader failed to signal END-OF-FILE inside an
2612 object representation. (reported by Nikodemus Siivola)
2613 * fixed some bugs revealed by Paul Dietz' test suite:
2614 ** LAST and [N]BUTLAST should accept a bignum.
2615 ** condition slot accessors are methods.
2616 ** (VECTOR NIL) is a subtype of STRING.
2618 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2619 * minor incompatible change: some nonsensical specialized lambda
2620 lists (used in DEFMETHOD) which were previously ignored now signal
2622 * minor incompatible change: the system is now aware of the types of
2623 variables in the COMMON-LISP package, and will signal errors for
2624 most violations of these type constraints (where previously they
2625 were silently accepted).
2626 * minor incompatible change: COMPILE-FILE now uses the freedom
2627 afforded (ANSI 3.2.2.3) to use derived function types for
2628 functions defined in the same file. This also permits the system
2629 to warn on static type mismatches and function
2630 redefinition. (Currently it does not work with high DEBUG level.)
2631 * minor incompatible change: VALUES declaration is disabled.
2632 * When issuing notes, the compiler now signals a condition of type
2633 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2634 restart for use in user handlers. It is expected that the
2635 COMPILER-NOTE condition will eventually become a condition
2636 supertype to a hierarchy of note types, which will then be
2637 handleable in a similar fashion. However, at the moment, no such
2638 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2639 but it's an implementation detail, not a classification for the
2641 * Changes in type checking closed the following bugs:
2642 ** type checking of unused values (192b, 194d, 203);
2643 ** template selection based on unsafe type assertions (192c, 236);
2644 ** type checking in branches (194bc).
2645 * A short form of VALUES type specifier has ANSI meaning (it has
2646 increased the number of situations when SBCL cannot perform type
2648 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2649 keywords or constants is permissible.
2650 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2651 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2652 classes. (thanks to Antonio Martinez)
2653 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2654 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2655 argument to a pretty-print stream. Instead, it leaves any such
2656 operations to DESCRIBE-OBJECT methods.
2657 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2659 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2660 lambda lists. (thanks to David Lichteblau)
2661 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2662 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2663 (thanks to Gerd Moellmann)
2664 * fixed bug 246: increased compilation speed of long
2665 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2667 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2669 * DEFCONSTANT now throws a condition of type
2670 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2671 constant to a non-EQL value; CONTINUE and ABORT restarts
2672 respectively change and preserve the value.
2673 * fixed bug 63: The code walker, part of the implementation of CLOS,
2674 is now better at handling symbol macros.
2675 * bug fix: There is no longer an internal implementation type named
2676 CL:LENGTH. (reported by Raymond Toy)
2677 * bug fix: In macro-like defining macros/special operators the
2678 implicit block does not enclose the lambda list.
2679 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2680 enclosed in parentheses) are not suitable as type specifiers, and
2681 their use properly signals an error now.
2682 * bug fix: An argument count mismatch for a type specifier in code
2683 being compiled no longer causes an unhandled error at compile
2684 time, but signals a compile-time warning.
2685 * fixed simple vector readable printing
2686 * bug fix: DESCRIBE takes more care over whether the class
2687 precedence list slot of a class is bound before accessing it.
2688 (reported by Markus Krummenacker)
2689 * bug fix: FORMATTER can successfully compile pretty-printer format
2690 strings which use variants of the ~* directive inside.
2691 * bug fix: SEARCH now applies its TEST predicate to the elements of
2692 the arguments in the correct order. (thanks to Wolfhard Buss)
2693 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2694 in type inference. (thanks to Robert E. Brown)
2695 * bug fix: Reading in symbols with an explicit package name of ""
2696 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2697 * fixed some bugs revealed by Paul Dietz' test suite:
2698 ** NIL is now allowed as a structure slot name.
2699 ** Arbitrary numbers, not just REALs, are allowed in certain
2700 circumstances in LOOP for-as-arithmetic clauses.
2701 ** Multiple class redefinitions before slot access no longer
2702 causes a type error.
2703 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2704 association between the name and a class.
2705 ** Generic functions with non-standard method-combination and over
2706 five methods all of which return constants no longer return NIL
2707 after the first few invocations. (thanks to Gerd Moellmann)
2708 ** CALL-NEXT-METHOD with no arguments now passes the original
2709 values of the arguments, even in the presence of assignment.
2710 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2712 ** DISASSEMBLE works with closures and funcallable instances.
2713 ** ADD-METHOD now returns the generic function, not the new method.
2714 ** FIND-METHOD signals an error if the lengths of the specializers
2715 is incompatible with the generic function, even if the ERRORP
2717 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2718 which its argument is a member.
2719 ** DEFCLASS only redefines the class named by its class-name
2720 argument if that name is the proper name of the class;
2721 otherwise, it creates a new class.
2722 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2723 of the UNBOUND-SLOT condition to the name of the slot.
2724 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2725 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2726 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2727 SLOT-MAKUNBOUND in the specified fashion.
2729 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2730 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2731 cross-compilation host. As a consequence, we can now bootstrap our
2732 way up to SBCL starting with a bare gcc toolchain and human-readable
2733 source code (first the source to CLISP, then the source to SBCL).
2734 * A contributed module containing a partial implementation of the
2735 simple-streams interface has been included. (thanks to Rudi
2737 * A contributed module implementing the RFC1321 Message Digest
2738 Algorithm, known as MD5, has been included.
2739 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2740 no longer has any effect, as the code controlled by this feature
2741 has been deleted. (As far as we know, no-one has ever built using
2742 this feature, and its semantics were confused in any case).
2743 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2744 work on conditions (as required by the ANSI specification),
2745 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2746 expected behaviour on conditions. Users should note, however,
2747 that such behaviour is not required by the ANSI specification,
2748 and so use of this behaviour may render their code unportable.
2749 * fixed some bugs revealed by Paul Dietz' test suite:
2750 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2752 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2753 ** SXHASH on condition objects no longer returns NIL.
2754 ** :ALLOCATION :CLASS slots are better treated; their values are
2755 updated on class redefinition, and initforms inherited from
2756 superclasses are applied.
2757 ** REMOVE-METHOD returns its generic function argument even when
2758 no method was removed.
2759 ** SHARED-INITIALIZE now initializes the values of the requested
2760 slots, including those with :ALLOCATION :CLASS.
2761 ** ALLOCATE-INSTANCE now works on structure classes defined via
2762 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2764 ** SLOT-EXISTS-P now works on conditions, as well as structures
2766 ** MAKE-LOAD-FORM now has the required methods on
2767 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2768 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2769 keyword, and now implements the SLOT-NAMES argument.
2770 ** methods with &OPTIONAL arguments no longer allow too many
2771 arguments to be passed in the call without error.
2772 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2773 option is consistent with the required arguments of the generic
2774 function lambda list.
2775 * bug fix: REQUIRE accepts a string designator. (Thanks to
2777 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2778 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2779 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2780 * bug fix: The compiler now checks for duplicated variables in macro
2782 * bug fix: SETQ on globals returns the correct value.
2783 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2784 not a COMPILER-ERROR (followed by some other strange error on
2785 choosing the CONTINUE restart).
2786 * bug fix: make.sh and friends are now more consistent in the way that
2787 they look for GNU "make".
2789 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2790 * experimental native threads support (on x86 Linux >=2.4 only).
2791 This is not compiled in by default: you need to add :SB-THREAD to
2792 the target features. See the "Beyond ANSI" chapter of the manual
2794 * fix for longstanding nonANSIism: The old distinction between
2795 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2796 The return value from CL:FIND-CLASS is now a CLOS class, and
2797 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2798 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2799 * An interface to the MetaObject Protocol, as described in Kiczales,
2800 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2801 MIT Press, 1991, is available from the SB-MOP package.
2802 * incompatible change: the SB-PCL package should now be considered
2803 a private implementation detail, and no longer a semi-private MOP
2805 * minor incompatible change: due to rearrangement for threads, the
2806 control stack and binding stack are now allocated at arbitrary
2807 addresses instead of being hardcoded per-port. Users affected by
2808 this probably have to be doing advanced things with shared
2809 libraries, and will know who they are.
2810 * minor incompatible change: Previously, all --eval forms used were
2811 processed with READ before any of them were processed with EVAL.
2812 Now each --eval form is processed with both READ and EVAL before
2813 the next --eval form is processed. (Thus package operations like
2814 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2815 work as the user might reasonably expect.)
2816 * minor incompatible change: *STANDARD-INPUT* is now only an
2817 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2819 * minor incompatible change: Y-OR-N-P is now character-oriented,
2820 not line oriented. Also, YES-OR-NO-P now works without errors.
2821 (thanks to Antonio Martinez)
2822 * sb-aclrepl module improvements: an integrated inspector, added
2823 repl features, and a bug fix to :trace command.
2824 * Known functions, which cannot be open coded by the backend, are
2825 considered to be able to check types of their arguments. (fixing
2826 a bug report by Nathan J. Froyd)
2827 * fixed a bug in computing method discriminating functions: It is
2828 now possible to define methods specialized on classes which have
2829 forward-referenced superclasses. (thanks to Gerd Moellmann)
2830 * fixed evaluation order in optional entries (reported by Gilbert
2832 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2833 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2834 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2835 required-by-AMOP NAME argument, as well as CLASS and
2836 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2837 * fixed bug 20: DEFMETHOD can define methods using names that are
2838 not the proper names of classes to designate class specializers.
2839 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2840 Unix file descriptors, instead of blowing up. (thanks to Antonio
2842 * Garbage collection refactoring: user-visible change is that a
2843 call to the GC function during WITHOUT-GCING will not do garbage
2844 collection until the end of the WITHOUT-GCING. If you were doing
2845 this you were probably losing anyway.
2846 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2847 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2848 TYPEP the latter but not the former.
2849 * The compiler issues a full WARNING for calls to undefined functions
2850 with names from the CL package.
2851 * MAP-INTO for a vector destination is open coded. (reported by
2852 Brian Downing on c.l.l)
2853 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2854 documentation string.
2855 * fixed some bugs revealed by Paul Dietz' test suite:
2856 ** COPY-ALIST now signals an error if its argument is a dotted
2858 ** Condition slots are now accessed more correctly in the presence
2859 of multiple initargs for a given slot.
2860 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2861 exclude restarts of the same name associated with a different
2863 ** DEFCLASS of forward-referenced classes with another
2864 forward-referenced class in the superclasses list no longer
2866 ** Condition slots are now initialized once each, not multiple
2867 times. (thanks to Gerd Moellmann)
2868 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2870 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2871 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2872 arguments contain duplicated elements.
2873 ** RESTART-CASE understands local macros.
2874 ** RESTART-CASE associates exactly its own restarts with a condition.
2875 ** ENDP in safe mode checks its argument to be of type LIST.
2876 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2877 in question is unbound.
2878 ** Optimized MAKE-INSTANCE functions no longer cause internal
2879 assertion failures in the presence of duplicate initargs.
2880 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2881 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2882 propagated even in the ctor optimized implementation of
2884 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2886 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2887 * a better implementation of SXHASH on (simple) bit vectors,
2888 measured both in execution speed and in distribution of results
2889 over the positive fixnums, has been installed. Likewise, a better
2890 implementation of EQUAL for simple bit vectors is now available.
2891 * fixed CEILING optimization for a divisor of form 2^k.
2892 * fixed bug 240 (emitting extra style warnings "using the lexical
2893 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2894 by Antonio Martinez)
2895 * fixed SXHASH, giving different results for NIL depending on type
2896 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2897 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2898 variable. (found by Rolf Wester)
2899 * SBCL does not ignore type declarations for special
2900 variables. (reported by rif on c.l.l 2003-03-05)
2901 * some bug fixes in contrib/sb-aclrepl/
2902 * fixed some bugs revealed by Paul Dietz' test suite:
2903 ** a bug in the CONS type specifier, whereby the CAR and CDR
2904 types got intertwined, has been fixed;
2905 ** the type system is now able to reason about the interaction
2906 between INTEGER and RATIO types more completely;
2907 ** APPEND, [N]REVERSE and NRECONC check that those their
2908 arguments, which must be proper lists, are really so;
2909 ** An array specialized to be unable to hold elements has been
2910 implemented, as required -- yes, really -- by ANSI;
2911 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2912 on malformed property lists;
2914 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2915 * incompatible packaging change: in line with Unix convention,
2916 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2917 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2918 * REQUIRE and PROVIDE are now optionally capable of doing something
2919 useful. See the documentation string for REQUIRE.
2920 * infrastructure for a managed SBCL contrib system: contributed
2921 modules in this release include:
2922 ** the ASDF system definition facility;
2923 ** an interface to the BSD Sockets API;
2924 ** an ACL-like convenience interface to the repl;
2925 (thanks to Kevin Rosenberg)
2926 ** an implementation of ROTATE-BYTE, with efficient implementation
2928 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2929 gives a closer approximation to the right answer than previously.
2930 (thanks to Raymond Toy)
2931 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2932 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2933 optional environment arguments, as required by ANSI.
2934 * fixed bugs in other functions taking environment objects, allowing
2935 calls with an explicit NIL environment argument to be compiled
2937 * fixed bug 228: primary return values from
2938 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2939 COMPILE or FUNCTION.
2940 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2941 :TYPE will no longer signal errors on innocuous objects.
2942 * fixed bug 231b: SETQ is better at respecting type declarations in
2943 the lexical environment.
2944 * fixed a bug in DEFCLASS: classes named by symbols with no or
2945 unprintable packages can now be defined.
2946 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2947 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2948 * fixed compiler failure related to checking types of functions.
2949 (reported by Robert E. Brown)
2950 * the compiler is now much more consistent in its error-checking
2951 treatment of bounding index arguments to sequence functions: in
2952 (SAFETY 3) code, errors will be signalled in almost all cases if
2953 invalid sequence bounding indices are passed to functions defined
2954 by ANSI to operate on sequences.
2955 * fixed a bug in the build procedure: documentation of SBCL-specific
2956 packages is now preserved and available in the final Lisp image.
2957 * lifted FDEFINITION lookup out of loops in the implementation of
2958 many list operations. (thanks to Robert E. Brown)
2959 * fixed a bug in the reader: the #n# reader macro now works for
2960 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2961 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2962 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2963 E. Brown and Tony Martinez respectively)
2964 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2965 SLOT-BOUNDP: in particular, it is now able to optimize them much
2966 better, and is now not vulnerable to having packages renamed.
2967 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2968 always called when a slot is not present in an instance. (thanks
2970 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2971 longer signal an error on unreading a character following EOF on
2972 the previous constituent stream. (thanks to Tony Martinez)
2973 * fixed some bugs revealed by Paul Dietz' test suite:
2974 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2975 not just nonnegative fixnums;
2976 ** the logical bit-array operators such as BIT-AND now accept an
2977 explicit NIL for their "opt-arg" argument (to indicate a
2978 freshly-consed result bit-array);
2979 ** ELT now signals an error on an invalid sequence index in safe
2981 ** the type system is now cleverer about negations of numeric
2982 types, and consequently understands the BIGNUM and RATIO types
2984 ** the type system is now cleverer about the interaction between
2985 INTEGER and RATIO types: while bugs still remain, many more
2986 cases are accurately computed;
2987 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2988 if it is in the last clause;
2989 ** CONSTANTLY now correctly returns a side-effect-free function in
2991 ** DECLARE is no longer treated as a special-operator; in
2992 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2993 * incremented fasl file version number due to the change in the
2994 DEFSTRUCT-SLOT-DESCRIPTION structure.
2996 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2997 * minor incompatible change: code processed by the "interpreter" or
2998 EVAL now has a compilation optimization policy of (DEBUG 2)
2999 (changed from (DEBUG 1)) to improve debuggability of interactive
3000 development, and to allow the use of the debug RETURN command in
3002 * an experimental implementation of the RETURN command for the
3003 debugger has been included. (thanks to Frederik Kuivinen)
3004 * fixed bug 62: constraints were not propagated into a loop.
3005 * fixed bug in embedded calls of SORT (reported and investigated by
3007 * fixed some bugs revealed by Paul Dietz' test suite:
3008 ** printing and reading of arrays with some dimensions having
3009 length 0 (thanks to Gerd Moellmann);
3010 ** BOA constructor with &AUX argument without a default value does
3011 not cause a type error;
3012 ** CONSTANTP now returns true for all self-evaluating objects.
3014 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3015 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3016 accessors that are related by inheritance, as specified in the
3017 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3018 Valtteri Vuorikoski)
3019 * The compiler is now able to inline functions that were defined in
3020 a complex lexical environment (e.g. inside a MACROLET).
3021 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3023 * Support for the upcoming FreeBSD-5.0 release has been included.
3024 (thanks to Dag-Erling Smorgrav)
3025 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3026 effect when it is not in a toplevel context.
3027 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3028 stand a better chance of being correct. (thanks to Gerd
3030 * fixed bug in COERCE, which now signals an error on coercing a
3031 rational to a bounded real type which excludes the expected
3033 * The compiler is now able to derive types more accurately from the
3034 COERCE and COMPILE functions.
3035 * fixed bug 223: functional binding is considered to be constant
3036 only for symbols in the CL package.
3037 * fixed bug 231: SETQ did not check the type of a variable being set
3038 (reported by Robert E. Brown)
3039 * A new optimization for MAKE-INSTANCE has been included, fixing
3040 various bugs (including relating to :ALLOCATION :CLASS slots and
3041 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3043 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3044 ** As required by ANSI, LOOP now disallows anonymous collection
3045 clauses such as COLLECT I in conjunction with aggregate boolean
3046 clauses such as THEREIS (= I 1);
3047 ** LOOP now signals an error when any variable is reused in the
3048 same loop (including the potentially useful construct analogous
3049 to WITH A = 1 WITH A = (1+ A);
3050 ** IT is only a special loop symbol within the first clause of a
3051 conditional loop clause;
3052 ** LOOP with a typed iteration variable over a hashtable now
3053 signals a type error iff it should.
3054 * fixed some other bugs revealed by Paul Dietz' test suite:
3055 ** FILE-STREAM now names the class previously known as FD-STREAM;
3056 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3057 argument) no longer signals an error;
3058 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3059 of the slot symbol, rather than using the current package
3060 ((:CONC-NAME "") continues to intern the slot's name in the
3062 * incremented fasl file version number, because of the incompatible
3063 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3064 of the new implementation of DEFINE-COMPILER-MACRO.
3066 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3067 * Support for building SBCL for MIPS platforms running in
3068 little-endian mode has now been checked in, and basic
3069 functionality on said platforms verified.
3070 * minor incompatible change: PCL now records the pathname of a file
3071 in which methods and the like are defined, rather than its
3073 * minor incompatible change: TRUENAME now considers the truename of
3074 a file naming a directory to be the pathname with :DIRECTORY
3075 component indicating that directory.
3076 * minor incompatible change: a NAMED clause in the extended form of
3077 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3078 reason for the previous behaviour is unclear.
3079 * more systematization and improvement of CLOS and MOP conformance
3080 in PCL (thanks to Gerd Moellman and Pierre Mai):
3081 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3083 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3084 primary methods with no specializers;
3085 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3087 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3088 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3089 FORWARD-REFERENCED-CLASSes; error reporting on
3090 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3092 ** SXHASH on CLOS instances now uses a slot internal to the
3093 instance to return different numbers on distinct instances,
3094 while preserving the same return value through invocations of
3096 ** DEFMETHOD signals errors when methods with longer incongruent
3097 lambda lists are added to generic functions;
3098 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3099 CLASS, as specified in AMOP;
3100 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3101 on the order returned by the primary method for classes of
3102 class STANDARD-CLASS;
3103 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3104 * fixed some bugs shown by Paul Dietz' test suite:
3105 ** DOLIST puts its body in TAGBODY;
3106 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3108 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3109 value producing form;
3110 ** if more variables are given to PROGV than values, extra
3111 variables are bound and made to have no value;
3112 ** NSUBSTITUTE on list arguments gets the right answer with
3114 ** ELT signals an error of type TYPE-ERROR when the index argument
3115 is not a valid sequence index;
3116 ** LOOP signals (at macroexpansion time) an error of type
3117 PROGRAM-ERROR when duplicate variable names are found;
3118 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3119 ** FUNCALL of special-operators now cause an error of type
3121 ** PSETQ now works as required in the presence of side-effecting
3122 symbol-macro places;
3123 ** NCONC accepts any object as its last argument;
3124 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3126 ** loop-for-as-package does not require a package to be explicitely
3128 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3130 * fixed bug 166: compiler preserves "there is a way to go"
3131 invariant when deleting code.
3132 * fixed bug 172: macro lambda lists with required arguments after
3133 &REST arguments now cause an error to be signalled. (thanks to
3135 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3136 bug. (thanks to Matthew Danish)
3137 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3139 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3140 arguments are lexically rebound. (thanks to Gerd Moellmann and
3142 * fixed bug 194: error messages are now more informative when there
3143 is no primary method applicable in a call to a generic
3144 function. (thanks to Gerd Moellmann)
3145 * fixed bug in command line argument checking (thanks to Julian
3147 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3148 (thanks to Matthew Danish)
3149 * incremented fasl file version number, because of the
3150 SXHASH-related changes in the layout of CLOS data structures
3152 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3153 * minor incompatible change: The runtime (the Unix executable named
3154 "sbcl") is now much pickier about the .core files it will load.
3155 Essentially it now requires .core files to descend from the same
3156 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3157 as the runtime does. (The intent is to prevent the crashes which
3158 can occur, and which can even be reported as mysterious failures,
3159 when people patch the sources or change the build parameters
3160 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3161 sbcl and .core files.)
3162 * fixed bug: VALUES-LIST is no longer optimized away.
3163 * fixed bug 142: The FFI conversion of C string values to Lisp
3164 string values no longer conses excessively. (thanks to Nathan
3165 Froyd porting Raymond Toy's fix to CMU CL)
3166 * began to systematize and improve MOP conformance in PCL (thanks to
3167 Nathan Froyd, Gerd Moellman and Pierre Mai):
3168 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3170 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3171 ** FINALIZE-INHERITANCE is now called on class finalization;
3172 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3173 argument precedence order.
3174 * fixed bug 202: The compiler no longer fails on functions whose
3175 derived types contradict their declared type.
3176 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3177 so it can be non-toplevel.
3178 * The fasl file version number has changed (because of the new
3179 implementation of DEFMACRO).
3180 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3181 safe code, that any length requirement by their type-specifier
3182 argument is valid. The exceptions to this are described in bug
3184 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3185 functions now check the types of their inputs as required by ANSI.
3186 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3187 introduced symbol is DECLAREd to be SPECIAL.
3188 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3189 * fixed bug: :COUNT argument to sequence functions may be negative
3190 * fixed bug: body of DO-SYMBOLS may contain declarations
3191 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3192 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3193 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3194 symbol macro only once
3195 * fixed printing of call frame when argument list is unavailable
3196 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3197 * compiler no longer signals WARNING on unknown keyword
3200 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3201 * A beta-quality port to the mips architecture running Linux,
3202 based on the old CMUCL backend, has been made. It has been tested
3203 on a big-endian kernel, and works sufficiently well to be able to
3204 rebuild itself; it has not been tested in little-endian mode.
3205 * fixed an inconsistency between gencgc.c and purify.c which made
3206 dumping/loading .core files unreliable
3207 * fixed bug 120a: The compiler now deals correctly with IFs where
3208 the consequent is the same as the alternative, instead of
3209 misderiving the return type. (thanks to Alexey Dejneka)
3210 * fixed bug 113: Logical pathnames are now dumpable (the logical
3211 host is resolved at load-time, throwing an error if it is not
3213 * fixed bug 174: FORMAT's error message is slightly clearer when a
3214 non-printing character is used in a format directive.
3215 * fixed several bugs in compiler checking of type declarations, i.e.
3216 violations of the Python "declarations are assertions" principle
3217 (thanks to Alexey Dejneka)
3218 * fixed several bugs in PCL's error checking (thanks to Gerd
3220 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3222 * fixed bug in compilation of functions as first class values
3223 (thanks to Antonio Martinez-Shotton)
3224 * The compiler's handling TYPE-ERRORs which it can prove will
3225 inevitably happen at runtime has been cleaned up and corrected
3226 in several ways. (thanks to Alexey Dejneka)
3227 * improved argument type checking for various basic arithmetic
3228 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3229 lavished on them in the past that they can be compiled in many
3230 ways in different special cases
3231 * fixed bug 181: compiler checks validity of user supplied type
3233 * cleaned up code flushing in optimization: Function calls which
3234 should signal errors for safety purposes (e.g. which ANSI says
3235 should signal errors when their arguments are of incorrect type)
3236 are no longer optimized away.
3237 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3238 * incremented fasl file version number, because changes in the
3239 implementation of sequence functions like COERCE caused
3240 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3241 in old inline expansions) to become undefined. (Actually these
3242 changes were later undone, so we might very well be binary
3243 compatible with 0.7.7 after all, but leaving the version number
3244 incremented seemed like the simplest and most conservative
3247 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3248 * An alpha-quality port to the parisc architecture running Linux,
3249 based on the old CMUCL backend, has been made. This, even more so
3250 than the other backends, should be considered still a work in
3251 progress; known problems include that the Linux kernel in 64-bit
3252 mode does not propagate the correct sigcontext structure to
3253 userspace, and consequently SBCL on a parisc64 kernel will not
3255 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3256 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3257 suppressed.) Also now that the compiler is looking at declarations
3258 in the environment, it checks optimization declarations as well,
3259 and suppresses inlining when (> DEBUG SPEED).
3260 * More fixes have been made to treatment of floating point exception
3261 treatment and other Unix signals. In particular, floating point
3262 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3263 * The detection and handling of control stack exhaustion (infinite
3264 or very deeply nested recursion) has changed. Stack exhaustion
3265 detection is now done by write-protecting pages at the OS level
3266 and applies at all optimization settings; when found, a
3267 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3268 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3269 be caught using IGNORE-ERRORS.
3270 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3271 correctly when transforming recursive LABELS functions to LETs.
3272 (thanks to Alexey Dejneka)
3273 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3274 that are names of constants or global variables.
3275 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3276 alien routines with docstrings.
3277 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3278 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3280 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3281 (thanks to Eric Marsden)
3282 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3283 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3284 * Bug fix: The SPARC backend can now compile functions involving
3285 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3286 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3287 to a functions expecting a PATHNAME-DESIGNATOR.
3288 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3289 lambda lists. (thanks to Alexey Dejneka)
3290 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3291 functions so that EXPORTing the name of the class doesn't cause
3292 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3293 to Antonio Martinez for reporting this)
3294 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3295 constant-folded without failing an assertion. (thanks to Einar
3296 Floystad Dorum for reporting this)
3297 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3298 types are now dealt with more correctly by the compiler.
3299 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3300 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3301 behaviour analogous to RENAME-FILE. This puts its behaviour more
3302 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3304 * The fasl file version number has changed again. (because of the
3305 bug fix involving the names of PCL MAKE-INSTANCE functions)
3307 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3308 * bug fix: Floating point exceptions are treated much more
3309 consistently on the x86/Linux and PPC/Linux platforms.
3310 * Array initialization with :INITIAL-ELEMENT is now much faster for
3311 cases when the compiler cannot open code the array creation, but
3312 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3313 array accesses have also seen a speed increase.
3314 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3315 specified. (This was at the root of some bad interactions between
3316 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3318 * bug fix: Internal error arguments for undefined functions are now
3319 computed correctly on the PPC/Linux platform.
3320 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3321 Raymond Toy's patch for CMU CL)
3322 * Support for the Solaris 9 operating environment has been included
3323 (thanks to Daniel Merritt)
3324 * A very ugly but hopefully complete draft of the missing FFI chapter
3325 of the manual has been created by reformatting the corresponding
3326 CMU CL manual chapter into (currently very ugly and incoherent)
3327 DocBook and bringing it up to date for SBCL behavior. Thus, the
3328 manual is now essentially complete, at least by my extreme
3329 once-and-only-once standards, whereby it's acceptable to refer to
3330 the doc strings of SB-EXT functions as the primary documentation.
3331 * The fasl file version number has changed again, due to cleanup of
3332 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3333 is no longer a static symbol.)
3335 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3336 * SBCL now builds with OpenMCL (version 0.12) as the
3337 cross-compilation host; also, more progress has been made toward
3338 bootstrapping under CLISP.
3339 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3341 * bug 158 fixed: The compiler can now deal with integer loop
3342 increments different from 1; fixing this turned out also to fix
3344 * bug 169 fixed: no more bogus warnings about using lexical bindings
3345 despite the presence of perfectly good SPECIAL declarations (thanks
3346 to David Lichteblau)
3347 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3348 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3349 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3351 * bug 180 fixed: Method combination specifications no longer ignore
3352 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3353 * bug fix: Structure type predicate functions now check their argument
3354 count as they should.
3355 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3356 correctly. (thanks to Pierre Mai)
3357 * minor incompatible change: The --noprogrammer option is deprecated
3358 in favor of the new --disable-debugger option, which is very similar.
3359 (The major difference is that it takes effect at a slightly different
3360 time at startup, causing handling of errors in --sysinit and
3361 --userinit files will be affected differently.) The
3362 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3363 been added to allow this functionality to be controlled from ordinary
3364 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3365 maintainers, who might want to run non-interactive scripts to
3366 build SBCL cores which will later be used interactively.)
3367 * minor incompatible change: The LOAD function no longer, when given
3368 a wild pathname to load, loads all files matching that pathname.
3369 Instead, an error of type FILE-ERROR is signalled.
3371 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3372 * bug 147 fixed: The compiler preserves its block link/count
3373 invariants more correctly now so that it doesn't crash. (thanks
3375 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3377 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3378 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3379 no-value-for-this-character as NIL (as specified by ANSI).
3380 * HOST-NAMESTRING on physical pathnames now returns a string that is
3381 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3382 (thanks to Christophe Rhodes)
3383 * The Alpha port handles icache flushing more correctly. (thanks to
3385 * More progress has been made toward bootstrapping under CLISP. (thanks
3386 to Christophe Rhodes)
3387 * The fasl file format has changed again, because dynamic loading
3388 on OpenBSD (which has non-ELF object files) motivated some cleanups
3389 in the way that foreign symbols are transformed and passed around.
3390 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3391 is no longer treated as whitespace by the reader, but instead as
3392 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3393 |A
\x7fB|, instead of A as it used to.
3395 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3396 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3397 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3398 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3399 this as of 0.7.1.45, but was left out of the previous news section
3400 (thanks to Dan Barlow)
3401 * SBCL now runs on the Solaris operating system on SPARC architectures
3402 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3403 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3404 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3405 * SPARC backend cleanups, allowing builds of cores optimized for V8
3406 and V9 SPARCS, and also emission of code targeted to a particular
3407 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3409 * SBCL is closer to bootstrapping under CLISP, thanks to various
3410 fixes by Christophe Rhodes.
3411 * The fasl file format has changed again, to allow the compiler's
3412 INFO database to support symbol macros.
3413 * The user manual (in doc/) is formatted into HTML more nicely.
3414 (thanks to coreythomas)
3415 * The system is smarter about SUBTYPEP relationships, especially
3416 those involving NOT types (including types such as ATOM which are
3417 represented internally using NOT types). Thus SUBTYPEP is less
3418 likely to return (VALUES NIL NIL) in general, and in particular
3419 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3421 * The fasl file format has changed again, because the internal
3422 representation of types now includes a new slot to support the new
3423 SUBTYPEP-of-NOT-types logic.
3424 * (not a change in the main branch of SBCL, but a related prototype
3425 which can hopefully be merged into the main branch of SBCL in the
3426 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3427 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3428 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3429 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3430 dispatch macro character combinations. (thanks to Alexey Dejneka)
3431 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3432 that superficially look like logical namestrings correctly.
3433 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3435 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3436 default supplied-p) for &optional and &key arguments. (thanks to
3439 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3440 * incompatible change: The compiler is now less aggressive about
3441 tail call optimization, doing it only when (> SPACE DEBUG) or
3442 (> SPEED DEBUG). (This is an incompatible change because there are
3443 programs which relied on the old CMU-CL-style behavior to optimize
3444 away their unbounded recursion which will now die of stack overflow.)
3445 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3446 for non-GENCGC systems has been increased to 20M (since that
3447 seems much closer to the likely performance optimum for modern
3448 systems than the old 4M value was)
3449 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3450 and *DEBUG-PRINT-LEVEL*
3451 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3452 of CMU CL's support for SPARC, and various endianness and other
3453 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3454 * new syntactic sugar for the Unix command line: --load foo.bar is now
3455 an alternate notation for --eval '(load "foo.bar")'.
3457 ** The system now detects stack overflow and handles it gracefully,
3458 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3459 optimization settings. (This is a good thing in general, and
3460 its introduction in this version should be particularly timely
3461 for anyone whose code fails because of suppression of tail
3463 ** The system now hunts for the C variable "environ" in a more
3464 devious way, to avoid segfaults when the C library version differs
3465 between compile time and run time. (thanks to Christophe Rhodes)
3466 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3467 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3468 ** The compiler no longer issues bogus style warnings for undefined
3469 classes in the same source file as the DEFCLASSes which defined
3470 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3472 ** fixes in CONDITION class precedence list for undefined function
3473 errors (thanks to Alexei Dejneka)
3474 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3475 correctly. (thanks to Dan Barlow)
3476 ** portability fixes aiming at bootstrapping under CLISP (thanks
3477 to Dave McDonald and Christophe Rhodes)
3478 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3479 ** fixes in type translation and and type inference (thanks to
3481 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3482 ** various fixes in the new ports (thanks to Dan Barlow)
3483 * several changes related to debugging:
3484 ** suppression of tail recursion, as noted above
3485 ** stack overflow detection, as noted above
3486 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3487 is now the default. (For some time encapsulation has been more
3488 reliable than the breakpoint-based :ENCAPSULATE NIL
3489 implementation, at least on X86 systems; and I just noticed that
3490 encapsulation also seems closer to the spirit of the ANSI
3493 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3495 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3496 up properly again. (There was a packaging bug in 0.7.0 which
3497 left their definitions in SB-SYS::LOAD-FOREIGN and
3498 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3499 vital for most things which interface to C-level interfaces,
3500 like extensions working with sockets or databases or
3501 Perl-compatible regexes or whatever, and the need to fix
3502 this bug is the main reason that 0.7.1 was released so
3504 ** DEFGENERIC is now choosier about the methods it redefines, so that
3505 reLOADing a previously-LOADed file containing DEFGENERICs does
3506 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3507 cycle now works as it should. (thanks to Alexey Dejneka)
3508 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3509 fixed by Christophe Rhodes. (He actually submitted this patch
3510 months ago, and I delayed until after 0.7.0.)
3511 ** Bug 111 (internal compiler confusion about runtime checks on
3512 FUNCTION types) was fixed by Alexey Dejneka.
3513 * Some internal cleanups (getting rid of variables which aren't
3514 needed now that the byte interpreter is gone) caused the fasl
3515 file format number to change again.
3517 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3518 * major incompatible change: The default fasl file extension, i.e. the
3519 default extension for files produced by COMPILE-FILE, has changed
3520 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3522 ** There are many changes in the implementation of the compiler.
3523 SBCL is now essentially a compiler-only implementation of ANSI
3524 Common Lisp. EVAL still "interprets" a few special cases, but
3525 almost all the interesting cases are handled by creating
3526 a LAMBDA expression, calling COMPILE on it, then calling
3527 FUNCALL on the result.
3528 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3529 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3530 Since the code is newer, there might still be some new bugs
3531 (though not as many as before Martin Atzmueller's fixes:-). But
3532 the new code is substantially simpler and clearer, and hopefully
3533 any remaining bugs will be simpler, less fundamental, and more
3534 fixable then the bugs in the old code.
3535 ** The revised compiler is still a little unsteady on its feet.
3537 *** The debugging information it produces (particularly the names
3538 of FUNCTION objects) is sometimes much less useful than what
3539 the old compiler produced.
3540 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3541 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3542 within a MACROLET) has been temporarily weakened.
3543 ** There are new compiler optimizations for various functions:
3544 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3545 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3546 *** the math functions TRUNCATE, FLOOR, and CEILING
3547 *** the function-of-all-trades COERCE
3548 Mostly these should be transparent, but there's one
3549 potentially-annoying problem (bug 117): when the compiler
3550 inline-expands a function and does type analysis on the result,
3551 it can create control paths which have type mismatches, and
3552 when it can't prove that those control paths aren't taken,
3553 it will issue WARNINGs about the type mismatches. This is
3554 a particular problem in practice for the new sequence functions.
3555 It's not clear how this should be fixed, and for now, a
3556 workaround is given in the entry for 117 in the BUGS file.
3557 ** (Because of the interaction between the two previous items --
3558 occasional inlining problems and new inline expansions -- some
3559 of the new sequence function optimizations won't really kick in
3560 completely until debugging information, and then inlining, are
3561 straightened out in some future version.)
3562 * minor incompatible changes:
3563 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3564 DIRECTORY no longer implicitly promotes NIL slots of its
3565 pathname argument to :WILD. In particular, when you ask for the
3566 contents of a directory (which you used to be able to do without
3567 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3568 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3569 ** changes in behavior that ANSI explicitly defines to be
3570 implementation dependent:
3571 *** The new compiler-only implementation still conforms with ANSI,
3572 but acts a little different than before. Besides the obvious
3573 changes in performance tradeoffs (that the cost per form passed
3574 to EVAL has gone up, and the cost per form executed by EVAL
3575 has gone down), the behavior of the system changes a little
3576 because there are no longer any interpreted function objects.
3577 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3578 e.g. doing COMPILE on the output of interactive DEFUN is
3580 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3582 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3583 has changed from (:CL) to NIL.
3584 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3585 e.g. #\Newline and #\Tab, don't have names specified in the
3586 ANSI Common Lisp standard, is now based on their ASCII symbolic
3587 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3588 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3589 are no longer used for output.
3590 ** changes in internal implementation constants:
3591 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3592 4 million. (If your application spends a lot of time GCing and
3593 you have a lot of RAM, you might want to experiment with
3594 increasing it even more.)
3595 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3596 However, almost all old code should still continue to work without
3597 immediate update, as SB-C-CALL is now a (deprecated) nickname
3599 ** Old operator names in the style DEF-FOO are now deprecated in
3600 favor of new corresponding names DEFINE-FOO, for consistency with
3601 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3602 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3603 internal symbols, but a few supported extensions like
3604 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3605 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3606 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3607 etc. as you get deeper into recursive calls to the debugger
3608 command loop, instead of the old "5]", "5]]", "5]]]"
3609 sequence. (I was motivated to do this when squabbles between
3610 ILISP and SBCL left me very deeply nested in the debugger. In the
3611 short term, this change will probably provoke more ILISP/SBCL
3612 squabbles, but hopefully it will be an improvement in the long run.)
3613 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3614 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3615 function, since declaiming return types involving aliens is
3616 (1) annoyingly messy to do by hand and (2) vital to efficient
3617 compilation of code which calls such functions.
3618 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3619 longer reexported by the SB-EXT package. They're solely useful
3620 for alien code, so it seems more logical that you should get
3621 them from the SB-ALIEN package, not in SB-EXT.
3622 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3623 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3624 features. Instead, the code that they used to control is always
3625 built into the system.
3626 * many other bug fixes
3627 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3628 advantage of the new EVAL-WHEN stuff and to clean them up in
3629 general, and they are now more ANSI-compliant in a number of
3630 ways. Martin Atzmueller is responsible for a lot of this.
3631 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3633 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3634 *** correct ERROR type for various file operations
3635 *** some fixes for Lisp streams
3636 *** DEFMETHOD syntax checking
3637 *** changing old weird representation of debug information as
3638 strings (which, among their other deficiencies, don't transform
3639 correctly when you rename packages, and don't change their
3640 print representation when you change things like *PACKAGE*
3641 and *PRINT-LENGTH*) to symbols and lists of symbols
3642 He also made several improvements and fixed several bugs in DESCRIBE.
3643 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3645 *** misbehavior of WRITE-STRING/WRITE-LINE
3646 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3647 and several other LOOP problems as well
3648 *** DIRECTORY when similar filenames are present
3649 *** DEFGENERIC with :METHOD options
3650 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3651 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3652 *** argument ordering in FIND with :TEST option
3653 *** mishandled package designator argument in APROPOS-LIST
3654 *** various problems in the backquote readmacro
3655 *** a bug in APROPOS
3656 *** probably some others that I'm not describing very well here,
3657 since the CVS log documents them by reference to sbcl-devel
3658 messages, and the SourceForge archives aren't working well.:-(
3659 ** Dan Barlow improved the Alpha port (and is making progress on the
3660 PPC port, for those of you who think different).
3661 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3662 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3663 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3664 continues to work on the SPARC port (for those of you in a position
3665 to look down upon our little PC-compatible boxes from a great height).
3666 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3667 and uses that copy, rather than the current dynamic value, when
3668 it's trying to decide whether to truncate output. Thus e.g.
3669 (let ((*print-lines* 50))
3670 (pprint-logical-block (stream nil)
3672 (let ((*print-lines* 8))
3673 (print (aref possiblybigthings i) stream)))))
3674 should now truncate the logical block only at 50 lines, instead of
3675 often truncating it at 8 lines, as it did before.
3676 * The doc/cmucl/ directory, containing old CMU CL documentation
3677 from the time of the fork, is no longer part of the base system.
3678 SourceForge has shut down its anonymous FTP service, and with it
3679 my original plan for distributing the old CMU CL documentation
3680 there. For now, if you need these files you can download an old
3681 SBCL source release and extract them from it.
3682 * The fasl file version number changed again, for dozens of reasons,
3683 some of which are apparent above.
3685 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3686 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3687 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3688 patch, so that DISASSEMBLE output is much nicer.
3689 * The code in the SB-PROFILE package now seems reasonably stable.
3690 I still haven't decided what the final interface should look like
3691 (I'd like PROFILE to interact cleanly with TRACE, since both
3692 facilities use function encapsulation) but if you have a need
3693 for profiling now, you can probably use it successfully with
3694 the current CMU-CL-style interface.
3695 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3696 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3697 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3698 link now returns the dangling link itself, and for similar
3699 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3700 link itself. (In these cases the old code signalled an error and
3701 looped endlessly, respectively.) Thus, DIRECTORY now works even
3702 in the presence of dangling and cyclic symbolic links.
3703 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3704 is now a supported extension again, since the consensus on
3705 sbcl-devel was that it can be useful for ordinary development
3706 work, not just for debugging SBCL itself.
3707 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3708 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3709 that currently-defined functions might be redefined later with
3710 different return types.
3711 * Hash tables can be printed readably, as inspired by CMU CL code
3712 of Eric Marsden and SBCL code of Martin Atzmueller.
3713 * better error handling in CLOS method combination, thanks to
3714 Martin Atzmueller porting Pierre Mai's CMU CL patches
3715 * more overflow fixes for >16Mbyte I/O buffers
3716 * A bug in READ has been fixed, so that now a single Ctrl-D
3717 character suffices to cause end-of-file on character streams.
3718 In particular, now you only need one Ctrl-D at the command
3719 line (not two) to exit SBCL.
3720 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3722 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3723 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3724 does the right thing.
3725 * The implementation of some type tests, especially for CONDITION
3726 types, is now tidier and maybe faster, due to CMU CL code
3727 originally by Douglas Crosher, ported by Martin Atzmueller.
3728 * Some math functions have been fixed, and there are new
3729 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3730 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3731 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3732 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3733 likely to be built into sbcl-0.7.0.)
3734 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3735 CL, ported to SBCL by Martin Atzmueller)
3736 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3737 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3738 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3739 * There's a new slam.sh hack to shorten the edit/compile/debug
3740 cycle for low-level changes to SBCL itself, and a new
3741 :SB-AFTER-XC-CORE target feature to control the generation of
3742 the after-xc.core file needed by slam.sh.
3743 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3744 COMPILE-FILE is no longer supported, so that now every function
3745 gets an entry point, so that block compilation looks a little
3746 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3747 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3748 returns the number of bytes consed since the system started,
3749 rather than the number consed since the first time the function
3750 was called. (The new definition parallels ANSI functions like
3751 CL:GET-INTERNAL-RUN-TIME.)
3752 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3753 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3754 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3755 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3756 * Fasl file version numbers are now independent of the target CPU,
3757 since historically most system changes which required version
3758 number changes have affected all CPUs equally. Similarly,
3759 the byte fasl file version is now equal to the ordinary
3762 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3763 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3764 is no longer recognized. I apologize for this, because it was
3765 listed in SB-EXT as a supported extension, but I found that
3766 its existing behavior was poorly specified, as well as incorrectly
3767 specified, and it looked like too much of a mess to straighten it
3768 out. I have enough on my hands trying to get ANSI stuff to work..
3769 * many patches ported from CMU CL by Martin Atzmueller, with
3770 half a dozen bug fixes in pretty-printing and the debugger, and
3771 half a dozen others elsewhere
3772 * fixed bug 13: Floating point infinities are now supported again.
3773 They might still be a little bit flaky, but thanks to bug reports
3774 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3775 as flaky as they were.
3776 * The --noprogrammer command line option is now supported. (Its
3777 behavior is slightly different in detail from what the old man
3778 page claimed it would do, but it's still appropriate under the
3779 same circumstances that the man page talks about.)
3780 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3781 are now supported, and enabled by default. Thus, the compiler can
3782 handle many floating point and complex operations much less
3783 inefficiently. (Thus e.g. you can implement a complex FFT
3785 * The compiler now detects type mismatches between DECLAIM FTYPE
3786 and DEFUN better, and implements CHECK-TYPE more correctly, and
3787 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3788 thanks to patches from Martin Atzmueller.
3789 * various fixes to make the cross-compiler more portable to
3790 ANSI-conforming-but-different cross-compilation hosts (notably
3791 Lispworks for Windows, following bug reports from Arthur Lemmens)
3792 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3793 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3794 fixed thanks to Pierre Mai's CMU CL patches.
3795 * fixes to keep the system from overflowing internal counters when
3796 it tries to use i/o buffers larger than 16M bytes
3797 * fixed bug 45a: Various internal functions required to support
3798 complex special functions have been merged from CMU CL sources.
3799 (When I was first setting up SBCL, I misunderstood a compile-time
3800 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3801 * improved support for type intersection and union, fixing bug 12
3802 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3803 more obscure bugs as well
3804 * some steps toward byte-compiling non-performance-critical
3805 parts of the system, courtesy of patches from Martin Atzmueller
3806 * Christophe Rhodes has made some debian packages of sbcl at
3807 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3808 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3809 stable, but are nonetheless usable. When he's ready, I'd be happy
3810 to add them to the SourceForge "File Releases" section. (And if
3811 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3812 * new fasl file format version number (because of changes in
3813 internal representation of (OR ..) types to accommodate the new
3814 support for (AND ..) types, among other things)
3816 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3817 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3819 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3821 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3822 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3823 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3824 * DESCRIBE now works on structure objects again.
3825 * Most function call argument type mismatches are now handled as
3826 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3827 know whether the function will be redefined before the call is
3828 executed. (The compiler could flag local calls with full WARNINGs,
3829 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3830 it doesn't keep track of enough information to know whether calls
3831 are local in this sense.)
3832 * Compiler output is now more verbose, with messages truncated
3833 later than before. (There should be some supported way for users
3834 to override the default verbosity, but I haven't decided how to
3835 provide it yet, so this behavior is still controlled by the internal
3836 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3837 src/compiler/ir1util.lisp.)
3838 * Fasl file format version numbers have increased again, because
3839 support for the Gray streams extension changes the layout of the
3840 system's STREAM objects.
3841 * The Gray subclassable streams extension now works, thanks to a
3842 patch from Martin Atzmueller.
3843 * The full LOAD-FOREIGN extension (not just the primitive
3844 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3845 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3846 but like most other programs, it defaults to copying the Unix
3847 environment from the original process instead of starting the
3848 new process in an empty environment.
3849 * Extensions which manipulate the Unix environment now support
3850 an :ENVIRONMENT keyword option which doesn't smash case or
3851 do other bad things. The CMU-CL-style :ENV option is retained
3852 for porting convenience.
3853 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3854 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3856 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3858 * A patch from Martin Atzmueller seems to have solved the SIGINT
3859 problem, and as far as we know, signal-handling now works cleanly.
3860 (If you find any new bugs, please report them!)
3861 * The system no longer defaults Lisp source file names to types
3862 ".l", ".cl", or ".lsp", but only to ".lisp".
3863 * The compiler no longer uses special default file extensions for
3864 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3865 seems to expect a single default extension for all compiled code,
3866 and there's no compelling reason to try to stretch the standard
3867 to allow two different extensions.) Instead, byte-compiled files
3868 default to the same extension as native-compiled files.
3869 * Fasl file format version numbers have increased again, because
3870 a rearrangement of internal implementation packages made some
3871 dumped symbols in old fasl files unreadable in new cores.
3872 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3873 many fewer weird special cases.
3874 * Bug #17 (differing COMPILE-FILE behavior between logical and
3875 physical pathnames) has been fixed, and some related misbehavior too,
3876 thanks to a patch from Martin Atzmueller.
3877 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3878 by Tim Moore, ported to SBCL by Martin Atzmueller.
3879 * Martin Atzmueller fixed several filesystem-related problems,
3880 including bug #36, in part by porting CMU CL patches, which were
3881 written in part by Paul Werkowski.
3882 * More compiler warnings in src/runtime/ are gone, thanks to
3883 more patches from Martin Atzmueller.
3884 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3887 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3889 * DESCRIBE now works on CONDITION objects.
3890 * The debugger now handles errors which arise when trying to print
3891 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3893 * The build system now uses an additional file, customize-target-features.lisp,
3894 to allow local modifications to the target *FEATURES* list. (The point of
3895 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3896 debugging features enabled, without having to worry about propagating it
3897 into everyone's system when I do a "cvs update".) When no
3898 customize-target-features.lisp file exists, the target *FEATURES* list
3899 should be constructed the same way as before.
3900 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3901 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3902 warn about behavior which is undefined under the ANSI spec. Note: This
3903 is specified by ANSI, but it's not very popular with programmers.
3904 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3905 macro in the SBCL sources for an example of a workaround which you
3906 might use to make portable ANSI-standard code which does what you want.
3907 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3908 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3909 when it's used as a non-toplevel form. (This is required in order
3910 to implement the DEFCONSTANT-EQX macro.)
3911 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3912 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3913 funny things, due to bugs in the implementation of EVAL-WHEN
3914 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3915 non-toplevel DEFCONSTANTs are uncommon.)
3916 * The core file version number and fasl file version number have been
3917 incremented, because the old noncompliant DEFCONSTANT behavior involved
3918 calling functions which no longer exist, and because I also took the
3919 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3920 * fixed bug 1 (error handling before read-eval-print loop starts), and
3921 redid debugger restarts and related debugger commands somewhat while
3923 ** The QUIT debugger command is gone, since it did something
3924 rather different than the SB-EXT:QUIT command, and since it never
3925 worked properly outside the main toplevel read/eval/print loop.
3926 Invoking the new TOPLEVEL restart provides the same functionality.
3927 ** The GO debugger command is also gone, since you can just invoke
3928 the CONTINUE restart directly instead.
3929 ** The TOP debugger command is also gone, since it's redundant with the
3930 FRAME 0 command, and since it interfered with abbreviations for the
3932 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3934 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3935 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3936 some porting work by Martin Atzmueller)
3937 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3938 with a boatload of patches which he ported to SBCL. Now that those
3940 ** The system tries to make sure that its low-priority messages
3941 are prefixed by semicolons, to help people who like to use
3942 syntax highlighting in their ILISP buffer. (This patch
3943 was originally due to Raymond Toy.)
3944 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3945 patches originally written by Raymond Toy.
3946 ** The compiler understands coercion between single-value and
3947 multiple-VALUES type expressions better, getting rid of some very
3948 weird behavior, thanks to patches originally by Robert MacLachlan
3949 and Douglas Crosher.
3950 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3951 lambda lists, thanks to a patch originally by Pierre Mai.
3952 ** The system no longer bogusly warns about "abbreviated type
3954 ** The compiler gets less confused by inlining and RETURN-FROM,
3955 thanks to some patches originally by Tim Moore.
3956 ** The system no longer hangs when dumping circular lists to fasl
3957 files, thanks to a patch originally from Douglas Crosher.
3958 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3959 undefined function error.
3960 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3961 in structure slot initforms) for now, documented workaround instead:-|
3962 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3963 * fixed bug 5: added stubs for various Gray stream functions called
3964 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3965 installed, at least appropriate type errors are generated
3966 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3967 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3968 consistently in DEFMETHOD forms.
3969 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3970 it doesn't seem to affect SBCL after all
3971 * The C runtime system now builds with better optimization and many
3972 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3974 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3976 * The system is now under CVS at SourceForge (instead of the
3977 CVS repository on my home machine).
3978 * The new signal handling code has been tweaked to treat register
3979 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3980 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3981 downstream, e.g. in debug-int.lisp, has implicit dependencies
3982 on the unsignedness of integer representation of machine words,
3983 and that caused the system to bomb out with infinite regress
3984 when trying to recover from type errors involving signed values,
3985 e.g. (BUTLAST '(1 2 3) -1).)
3986 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3987 (This was one of the bugs Peter Van Eynde reported back in July.)
3988 * The system now uses code inspired by Colin Walters' O(N)
3989 implementation of MAP (from the cmucl-imp@cons.org mailing
3990 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3991 inline the MAP operation, and there is more than one
3992 sequence argument to the MAP call (so that it can't just
3993 do ETYPECASE once and for all based on the type of the
3994 single sequence argument). (The old non-inline implementation
3995 of the general M-argument sequence-of-length-N case required
3996 O(M*N*N) time when any of the sequence arguments were LISTs.)
3997 * The QUIT :UNIX-CODE keyword argument has been renamed to
3998 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3999 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4001 * Raymond Wiker's patches to port dynamic loading from Linux to
4002 FreeBSD have been added.
4003 * The BUGS file is now more nearly up to date, thanks in large part
4004 to Martin Atzmueller's review of it.
4005 * The debugger now flushes standard output streams before it begins
4006 its output ("debugger invoked" and so forth).
4007 * The core version number and fasl file version number have both
4008 been incremented, because of incompatible changes in the layout
4010 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4011 used to not be called for a saved Lisp image.)
4012 * Martin Atzmueller's version of a patch to fix a compiler crash,
4013 as posted on sbcl-devel 13 September 2000, has been installed.
4014 * Instead of installing Martin Atzmueller's patch for the
4015 compiler transform for SUBSEQ, I deleted the compiler transform,
4016 and transforms for some similar consing operations.
4017 * A bug in signal handling which kept TRACE from working on OpenBSD
4019 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4020 away by constant folding
4021 * The system now defines its address space constants in one place
4022 (in the Lisp sources), and propagates them automatically elsewhere
4023 (through GENESIS and the sbcl.h file). Therefore, patching the
4024 address map is less unnecessarily tedious and error-prone. The
4025 Lisp names of address space constants have also been systematized.
4026 * CVS tags like dollar-Header-dollar have been removed from
4027 the sources, because they have never saved me trouble and
4028 they've been source of trouble working with patches and other
4029 diff-related operations.
4030 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4031 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4033 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4035 * The system has been ported to OpenBSD.
4036 * The system now compiles with a simple "sh make.sh" on the systems
4037 that it's supported on. I.e., now you no longer need to tweak
4038 text in the target-features.lisp-expr and symlinks in src/runtime/
4039 by hand, the make.sh takes care of it for you.
4040 * The system is no longer so grossly inefficient when compiling code
4041 involving vectors implemented as general (not simple) vectors (VECTOR T),
4042 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4043 which dares to use the various sequence functions on non-simple
4044 vectors, takes less of a performance hit.
4045 * There is now a primitive type predicate VECTOR-T-P
4046 to test for the (VECTOR T) type, so that e.g.
4047 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4048 can now be compiled with some semblance of efficiency. (The old code
4049 turned the type declaration into a full call to %TYPEP at runtime!)
4050 * AREF on (VECTOR T) is still not fast, since it's still compiled
4051 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4052 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4053 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4054 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4055 without making full calls to SUBTYPEP at runtime.
4056 (Some analogous efficiency issues for non-simple vectors specialized to
4057 element types other than T, or for non-simple multidimensional arrays,
4058 have not been addressed. They could almost certainly be handled the
4059 same way if anyone is motivated to do so.)
4060 * The changes in array handling break binary compatibility, so
4061 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4062 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4063 instead of (VALUES T T).
4064 * By following the instructions that Dan Barlow posted to sbcl-devel
4065 on 2 July 2000, I was able to enable primitive dynamic object
4066 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4067 functionality is not implemented (since it calls ld to resolve
4068 library references automatically, requiring RUN-PROGRAM for its
4069 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4070 try to resolve library references) is now supported.
4071 * The system now flushes the standard output streams when it terminates,
4072 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4073 them at several other probably-convenient times, e.g. in each pass of
4074 the toplevel read-eval-print loop, and after evaluating a form given
4075 as an "--eval" command-line option. (These changes were motivated by a
4076 discussion of stream flushing issues on cmucl-imp in August 2000.)
4077 * The source transform for TYPEP of array types no longer assumes
4078 that an array whose element type is a not-yet-defined type
4079 is implemented as an array of T, but instead punts, so that the
4080 type will be interpreted at runtime.
4081 * There is now some support for cross-compiling in make.sh: each of
4082 the phases of make.sh has its own script. (This should be transparent
4083 to people doing ordinary, non-cross-compile builds.)
4084 * Since my laptop doesn't have hundreds of megabytes of memory like
4085 my desktop machine, I became more motivated to do some items on
4086 my to-do list in order to reduce the size of the system a little:
4087 ** Arrange for various needed-only-at-cold-init things to be
4088 uninterned after cold init. To support this, those things have
4089 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4090 symbols with such names are now uninterned after cold init).
4091 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4092 fasl files for cold load.
4093 ** Remove the old compiler structure pooling code (which used to
4094 be conditional on the target feature :SB-ALLOC) completely.
4095 ** Redo the representation of some data in cold init to be more compact.
4096 (I also looked into supporting byte compiled code at bootstrap time,
4097 which would probably reduce the size of the system a lot, but that
4098 looked too complicated, so I punted for now.)
4099 * The maximum signal nesting depth in the src/runtime/ support code has
4100 been reduced from 4096 to 256. (I don't know any reason for the very
4101 large old value. If the new smaller value turns out to break something,
4102 I'll probably just bump it back up.)
4103 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4105 * Many, many bugs reported by Peter Van Eynde have been added to
4106 the BUGS list; some have even been fixed.
4107 * While enabling dynamic object file loading, I tried to make the
4108 code easier to understand, renaming various functions and variables
4109 with less ambiguous names, and changing some function calling
4110 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4111 * While trying to figure out how to do the OpenBSD port, I tried to
4112 clean up some of the code in src/runtime/. In particular, I dropped
4113 support for non-POSIX signal handling, added various comments,
4114 tweaked the code to reduce the number of compilation warnings, and
4115 renamed some files to increase consistency.
4116 * To support the new automatic configuration functionality in make.sh,
4117 the source file target-features.lisp-expr has been replaced with the
4118 source file base-target-features.lisp-expr and the machine-generated
4119 file local-target-features.lisp-expr.
4120 * fixed a stupid quoting error in make.sh so that using CMU CL
4121 "lisp -batch" as cross-compilation host works again
4123 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4125 * DESCRIBE no longer tries to call itself recursively to describe
4126 bound/fbound values, so that it no longer fails on symbols which are
4127 bound to themselves (like keywords, T, and NIL).
4128 * DESCRIBE now works on generic functions.
4129 * The printer now prints less-screwed-up representations of closures
4130 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4131 * A private symbol is used instead of the :EMPTY keyword previously
4132 used to mark empty slots in hash tables. Thus
4133 (DEFVAR *HT* (MAKE-HASH-TABLE))
4134 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4135 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4136 now does what ANSI says that it should. (You can still get
4137 similar noncompliant behavior if bang on the hash table
4138 implementation with all the symbols you get back from
4139 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4140 This breaks binary compatibility, since tests for equality to
4141 :EMPTY are wired into things like the macroexpansion of
4142 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4144 * There's now a minimal placeholder implementation for CL:STEP,
4145 as required by ANSI.
4146 * An obscure bug in the interaction of the normal compiler, the byte
4147 compiler, inlining, and structure predicates has been patched
4148 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4149 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4151 * Missing ordinary arguments in a macro call are now detected even
4152 when the macro lambda list contains &KEY or &REST.
4153 * The debugger no longer complains about encountering the top of the
4154 stack when you type "FRAME 0" to explicitly instruct it to go to
4155 the top of the stack. And it now prints the frame you request even
4156 if it's the current frame (instead of saying "You are here.").
4157 * As specified by ANSI, the system now always prints keywords
4158 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4160 * The default initial SIZE of HASH-TABLEs is now smaller.
4161 * Type information from CLOS class dispatch is now propagated
4162 into DEFMETHOD bodies, so that e.g.
4163 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4165 is now basically equivalent to
4166 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4167 (DECLARE (TYPE SINGLE-FLOAT X))
4169 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4170 operation, without having to do run-time type dispatch.
4171 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4172 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4173 * Since I don't seem to be making big file reorganizations very often
4174 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4175 is overflowing my ability to conveniently back them up), I've finally
4176 checked the system into CVS. (The CVS repository is on my home system,
4177 not at SourceForge -- putting it on SourceForge might come later.)
4178 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4179 high-level GC-NOTIFY-FOO functions send their output. (There's
4180 still very little control of where low-level verbose GC functions
4181 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4182 less than it used to -- the GC-NOTIFY-FOO functions are now under
4183 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4184 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4185 in only one place in the source code, and propagates it automatically
4186 everywhere that it's needed. Thus e.g. when I bump the version from
4187 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4188 * The C source files now include boilerplate legalese and documentation
4189 at the head of each file (just as the Lisp source files already did).
4190 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4191 to his page will be replaced with a link to his new CLiki service.
4193 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4195 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4196 * The build process now looks for GNU make under the default name "gmake",
4197 instead of "make" as it used to. If GNU make is not available as "gmake"
4198 on your system, you can change this default behavior by setting the
4199 GNUMAKE environment variable.
4200 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4201 can build without error under CMU CL.
4203 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4205 * There is now a partial SBCL user manual (with some new text and some
4206 text cribbed from the CMU CL manual).
4207 * The beginnings of a profiler have been added (starting with the
4208 CMU CL profiler and simplifying and cleaning up). Eventually the
4209 main interface should be through the TRACE macro, but for now,
4210 it's still accessed through vaguely CMU-CL-style functions and macros
4211 exported from the package SB-PROFILE.
4212 * Some problems left over from porting CMU CL to the new
4213 cross-compilation bootstrap process have been cleaned up:
4214 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4215 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4216 part when getting rid of the compiler *BACKEND* stuff.)
4217 ** The value of *NULL-TYPE* was screwed up, because it was
4218 being initialized before the type system knew the final
4219 definition of the 'NULL type. This screwed up several key
4220 optimizations in the compiler, causing inefficiency in all sorts
4221 of places. (I found it because I wanted to understand why
4222 GET-INTERNAL-RUN-TIME was consing.)
4223 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4224 PROCLAIM FTYPE information. Unfortunately this broke binary
4225 compatibility again, since now the forms output by DEFGENERIC
4226 to refer to functions which didn't exist in 0.6.3.
4227 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4228 can use the new (as of 0.6.3) transform for SOME into MAP into
4230 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4231 values to (INTEGER 0 1000000), so that the time code will no longer
4232 occasionally get blown up by Linux returning 1000000 microseconds
4233 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4234 its output conform to the ANSI spec. (Alas, this makes its output
4235 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4236 * A full call to MAP NIL with a single sequence argument no longer conses.
4237 * fixes to problems pointed out by Martin Atzmueller:
4238 * The manual page no longer talks about multiprocessing as though
4239 it were currently supported.
4240 * The ILISP support patches have been removed from the distribution,
4241 because as of version 5.10.1, ILISP now supports SBCL without us
4242 having to maintain patches.
4243 * added a modified version of Raymond Toy's recent CMU CL patch for
4244 EQUALP comparison of HASH-TABLE
4246 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4248 * The system still can't cross-compile itself with
4249 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4250 entails), but at least it can compile more of itself that way
4251 than it used to be able to, and various buglets which were uncovered
4252 by trying to cross-compile itself that way have now been fixed.
4253 * This release breaks binary compatibility again. This time
4254 at least I've incremented the FASL file format version to 2, so that the
4255 problem can be detected reliably instead of just causing weird errors.
4256 * various new style warnings:
4257 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4258 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4259 ** using the lexical binding of a variable named in the *FOO* style
4260 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4261 as specified by ANSI.
4262 * *RANDOM-STATE* is no longer automatically initialized from
4263 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4264 default behavior of the system is to repeat its behavior every time
4265 it's run. If you'd like to change this behavior, you can always
4266 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4267 old convention there was no comparably easy way to get the system to
4268 repeat its behavior every time it was run.
4269 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4270 FTYPE declarations has been removed, in favor of their ANSI
4271 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4272 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4273 the types of their sequence arguments can be determined at compile time.
4274 This is done through a new open code expansion for MAP which eliminates
4275 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4276 argument types can be determined at compile time.
4277 * The optimizer now transforms COERCE into an identity operation when it
4278 can prove that the coerced object is already of the correct type. (This
4279 can be a win for machine generated code, including the output of other
4280 optimization transforms, such as the MAP transform above.)
4281 * Credit information has been moved from source file headers into CREDITS.
4282 * Source file headers have been made more standard.
4283 * The CASE macro now compiles without complaining even when it has
4286 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4288 * (Note that the way that the PCL macroexpansions were rewritten
4289 to accommodate the change in DEFGENERIC below breaks binary
4290 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4291 not run under sbcl-0.6.2. Once we get out of alpha releases,
4292 i.e. hit release 1.0.0, we'll probably try to maintain binary
4293 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4294 and sbcl-1.4.4. Until then, however, it might be fairly common
4295 for maintenance releases to break binary compatibility.)
4296 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4298 * The reporting of unhandled signals has been changed to print some
4299 explanatory text as well as the report form. (Previously only
4300 the report form was printed.)
4301 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4302 it defines, so that the compiler no longer issues undefined function
4303 warnings for compiled-but-not-yet-loaded generic functions.
4304 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4305 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4306 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4307 section of the ANSI spec are supported.
4308 * The "" nickname for the "KEYWORD" package has been removed.
4309 The reader still handles symbol tokens which begin with a package marker
4310 as keywords, but it doesn't expose its mechanism for doing so in the
4311 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4312 * The system now issues STYLE-WARNINGs for contradictory TYPE
4313 proclamations. (Warnings for contradictory FTYPE proclamations would
4314 be nice too, but those can't be done usefully unless the type system
4315 is made smarter about FUNCTION types.)
4316 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4317 systematized, so that "*target-*.lisp is supposed to exist only on the
4318 target and imply that there's a related file which exists on the
4319 host, and *host-*.lisp is supposed to exist only on the host and imply
4320 that there's a related file which exists on the target. This involves a
4321 lot of renaming. Hopefully the acute confusion caused by the renaming
4322 will be justified by the reduction in chronic confusion..
4323 ** runtime-type.lisp -> early-target-type.lisp
4324 ** target-type.lisp -> late-target-type.lisp
4325 ** early-host-format.lisp -> early-format.lisp
4326 ** late-host-format.lisp -> late-format.lisp
4327 ** host-error.lisp -> misc-error.lisp
4328 ** early-error.lisp -> early-target-error.lisp
4329 ** late-error.lisp -> late-target-error.lisp
4330 ** host-defboot.lisp -> early-defboot.lisp
4331 ** code/misc.lisp -> code/target-misc.lisp
4332 ** code/host-misc.lisp -> code/misc.lisp
4333 ** code/numbers.lisp -> code/target-numbers.lisp
4334 ** code/early-numbers.lisp -> numbers.lisp
4335 ** early-host-type.lisp -> early-type.lisp
4336 ** late-host-type.lisp -> late-type.lisp
4337 ** host-typep.lisp -> typep.lisp
4338 ** load.lisp -> target-load.lisp
4339 ** host-load.lisp -> load.lisp
4340 ** host-disassem.lisp -> disassem.lisp
4341 ** host-insts.lisp -> insts.lisp
4342 ** byte-comp.lisp -> target-byte-comp.lisp
4343 ** host-byte-comp.lisp -> byte-comp.lisp
4344 ** host-signal.lisp -> signal.lisp
4345 ** host-defstruct.lisp -> defstruct.lisp
4346 ** late-target-type.lisp -> deftypes-for-target.lisp
4347 Furthermore, several other previously target-only files foo.lisp (e.g.
4348 hash-table.lisp and random.lisp) have been split into a target-and-host
4349 foo.lisp file and a target-only target-foo.lisp file, with their key type
4350 definitions in the target-and-host part, so that the cross-compiler will
4351 know more about target types.
4352 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4353 gone away. In their place are various *BACKEND-FOO* variables
4354 corresponding to the slots of the old structure.
4355 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4356 into the target SBCL has been fixed.
4357 * The chill.lisp system for loading cold code into a running SBCL
4359 * Support for the CMU CL "scavenger hook" extension has been removed.
4360 (It was undocumented and unused in the CMU CL sources that SBCL was
4361 derived from, and stale in sbcl-0.6.1.)
4362 * Various errors in the cross-compiler type system were detected
4363 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4364 (enabling various consistency checks). Many of them were fixed,
4365 but some hard problems remain, so the compiler is back to
4366 running without *TYPE-SYSTEM-INITIALIZED* for now.
4367 * As part of the cross-compiler type system cleanup, I implemented
4368 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4369 * I have started adding UNCROSS calls throughout the type system
4370 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4371 on cross-compiler input files will be able to go away, and various
4373 * CONSTANTP now returns true for quoted forms (as explicitly required
4376 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4378 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4379 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4380 (MAKE-LIST -1). In the longer term, it ought to become true
4381 that declarations are assertions even at SAFETY 1. For now, it's not
4382 quite true even at SAFETY 3, but it's at least more nearly true..
4383 (Note that this change seems to increases the size of the system by
4384 O(5%) and to decrease the speed of the compiler by 20% or more.)
4385 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4386 for the problem of printing dozens of lines of distracting information
4387 about low-level system machinery as part of the top stack frame
4388 on entry to the debugger when an undefined function was called.
4389 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4390 is not reset to COMMON-LISP-USER.
4391 * Compilation of stuff related to dyncount.lisp has been made conditional
4392 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4393 smaller. The various dyncount-related symbols have been moved into
4394 a new "SB-DYNCOUNT" package.
4395 * tty-inspect.lisp has been renamed to inspect.lisp.
4396 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4397 feature has gone away. (When we eventually port to other flavors of
4398 libc and/or Unix, we'll try to make the differences between flavors
4399 invisible at the user level.)
4400 * Various other *FEATURES* tags, and/or their associated conditionals,
4401 have been removed if obsolescent, or given better documentation, or
4402 sometimes given more-mnemonic names.
4404 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4406 * tidied up "make.sh" script
4407 * tidied up system directory structure
4408 * better "clean.sh" behavior
4409 * added doc/FOR-CMUCL-DEVELOPERS
4410 * many many small tweaks to output format, e.g. removing possibly-confusing
4411 trailing #\. character in DESCRIBE-INSTANCE
4412 * (EQUALP #\A 'A) no longer signals an error.
4413 * new hashing code, including EQUALP hashing
4414 * tidied up Lisp initialization and toplevel
4415 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4416 * command line argument processing
4417 * added POSIX-GETENV function to deal with Unix-ish environment variables
4418 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4419 terminating SBCL on EOF
4420 * non-verbose GC by default
4421 * There is no more "sbcl" shell script; the sbcl file is now the C
4422 runtime executable (just like CMU CL).
4423 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4425 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4426 symbol and package information as Lisp native symbol and package objects
4427 instead of strings naming symbols and strings naming packages. This way,
4428 whenever packages are renamed (as in warm init), debug information is
4429 transformed along with everything else.
4430 * tweaked the optimization policy declarations which control the building
4431 of SBCL itself. Now, among other things, the system no longer saves
4432 source location debugging information. (This helps two problems at once
4433 by reducing SBCL size and by keeping SBCL from trying to look for its
4434 sources -- which may not exist -- when reporting errors.)
4435 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4436 debugging and testing purposes
4437 * cleaned up printing, making the printer call PRINT-OBJECT for
4438 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4439 methods, giving nearly-ANSI behavior
4440 * converted almost all special variables to use *FOO* naming convention
4441 * deleted PARSE-TIME functionality, since it can be done portably
4442 * moved some files out of cold init into warm init
4443 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4445 * regularized formatting of source files
4446 * added an install.sh script
4447 * fixed ridiculous memory usage of cross-compiler by making
4448 compiler/alloc.lisp not try to do pooling unless it can hook
4449 itself into the GC of the cross-compilation host. Now the system
4450 builds nicely on my old laptop.
4451 * added :SB-ALLOC in target-features.lisp-expr
4452 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4453 was not implemented)
4454 * re-did condition handling and note reporting in the compiler. Notes
4455 are no longer handled by signalling conditions. Style warnings
4456 and warnings are handled more correctly and reported in such a way
4457 that it's easy to find one or the other in your output (so that you
4458 can e.g. figure out which of many problems caused COMPILE-FILE to
4460 * changed the severity of several compiler warnings from full WARNING
4461 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4462 compiler note reporting so that it doesn't use the condition system
4463 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4464 COMPILE-FILE command)
4465 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4466 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4467 no longer supported, since their implementation was deeply intertwingled
4468 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4469 * removed bogus "support" for compiler macros named (SETF FOO), and
4470 removed the compiler macro for SETF INFO (but only after making a fool
4471 of myself on the cmucl-imp mailing list by posting a bogus patch for
4472 DEFINE-COMPILER-MACRO..)
4473 * Compiled files containing forms which have side effects on the Lisp
4474 reader (such as DEFPACKAGE forms) are now handled more correctly.
4475 (Compiler queuing of top level lambdas has been suppressed by setting
4476 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4477 * deleted various currently-unused source files, e.g. gengc.lisp. They
4478 may be added back at some point e.g. when porting to other architectures,
4479 but until they are it's distracting to distribute them and to try to
4481 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4482 there were so many of them they're just distractions, and UNCROSS is
4483 known to be able to handle the current sources
4484 * moved PROFILE functionality into TRACE, so that it will be clear
4485 how the wrapping and unwrapping of functions when you profile them
4486 interacts with the wrapping and unwrapping of functions when you
4487 trace them. (Actually, the functionality isn't there yet, but at least
4488 the interface specification is there. Hopefully, the functionality will
4489 arrive with some future maintenance release.)
4490 * removed host-oops.lisp
4491 * changed signature of QUIT function to allow UNIX-CODE argument
4492 * fixed READ-SEQUENCE bug
4493 * tweaked verbose GC output so that it looks more like the progress
4494 output that ANSI specifies for functions like LOAD
4495 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4496 * added <http://sbcl.sourceforge.com> to the banner information printed by