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 * optimization: CONCATENATE on strings is an order of magnitue faster
9 in code compiled with (> SPEED SPACE).
10 * optimization: SUBSEQ is ~50% faster on lists.
11 * bug fix: bug 417 fixed -- source location reporting is now more robust.
12 * bug fix: SUBSEQ on a list will now correctly signal an error if if
13 END is smaller then START.
14 * bug fix: SB-PROFILE will no longer report extra consing for nested
15 calls to profiled functions.
16 * bug fix: ROOM implementation had bogus fixnum declarations which
17 could cause type-errors when calling ROOM on large images.
18 * bug fix: if file compilation is aborted, the partial fasl is now
19 deleted, and COMPILE-FILE returns NIL as the primary value.
20 * bug fix: number of thread safety issues relating to SBCL's internal
21 hash-table usage have been fixed.
22 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
23 be returned from its body when the values were being returned
24 using unknown-values return convection and the W-P-O was wrapped
25 inside an UNWIND-PROTECT.
26 * bug fix: sb-posix should now compile again under Windows, enabling
29 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
30 * incompatible change: hash-table accessor functions are no longer
31 automatically protected by locks. Concurrent accesses on the same hash-table
32 from multiple threads can give inconsistent results or even corrupt the
33 hash-table completely. Multi-threaded applications should do their own
34 locking at the correct granularity. In the current implementation it is
35 still safe to have multiple readers access the same table, but it's not
36 guaranteed that this property will be maintained in future releases.
37 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
38 and will signal an error at runtime.
39 * enhancement: SB-THREAD package now exports a semaphore interface.
40 * enhancement: CONS can now stack-allocate on x86 and
41 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
43 * enhancement: nested lists can now be stack allocated on
44 platforms providing stack allocation support.
45 * enhancement: dynamic-extent support has been extended to support
46 cases where there are multiple possible sources for the stack
48 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
49 if the mutex is uncontested on Linux.
50 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
51 as specified by the CLHS. (thanks to Tobias Rittweiler)
52 * bug fix: NaN comparison now works on x86-64.
53 * bug fix: CLOSE :ABORT T on a stream with pending output now
55 * bug fix: instances of non-standard metaclasses using standard
56 instance structure protocol sometimes missed the slot type checks
58 * bug fix: known functions can be inlined in high-debug code.
60 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
61 * minor incompatible change: the MSI installer on Windows no longer
62 associates .lisp and .fasl files with the installed SBCL.
63 * minor incompatible change: :UNIX is no longer present in *FEATURES*
64 on Windows. (thanks to Luis Oliviera)
65 * new feature: SB-CLTL2 contrib module now implements
66 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
67 * optimization: scavenging weak pointers is now more efficient,
68 requiring O(1) instead of O(N) per weak pointer to identify
69 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
70 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
71 method bodies using constant slot-names when the first argument is
72 a specializer parameter for the method.
73 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
74 names, when the class of the instance is a direct instance
75 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
76 as slow as the constant slot-name case.)
77 * optimization: member type construction is now O(N) instead
79 * optimization: UNION and NUNION are now O(N+M) for large
80 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
82 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
83 now more readable in environments like Slime which display it.
84 (thanks to Tobias C. Rittweiler)
85 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
86 was able to infer the correct argument type for the object on which
87 the CAS operation was being performed.
88 * bug fix: copy propagation interfered with parallel assignment
89 semantics in local calls. (reported by Paul Khuong)
90 * bug fix: the signed modular fixnum shift compiled to wrong code on
91 x86 and x86-64. (spotted by a slight modification to some of
94 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
95 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
96 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
97 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
99 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
100 and generic function now print the object identity for anonymous
101 instances. (thanks to Attila Lendvai)
102 * enhancement: as an extension to MOP, SBCL now supports SETF of
103 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
104 (thanks to Attila Lendvai)
105 * enhancement: sb-cover has an alternate annotation mode which puts
106 the form annotations at the car of the form, instead of annotating
107 it completely. The benefit of this mode is that it shows better
108 which forms were instrumented by the compiler.
109 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
110 by printer control variables. (reported by Dan Corkill)
111 * bug fix: system leaked memory when delayed output was performed by
112 the OS in smaller chunks then expected. (thanks to David Smith)
113 * bug fix: system leaked memory when file streams were not closed
115 * bug fix: large objects written to slow streams that were modified
116 after the write could end up with the modified state written to
117 the underlying file descriptor.
118 * bug fix: multiple threads operating in parallel on the same stream
119 could cause buffer-overflows.
120 * bug fix: source location information is stored correctly
121 (broken since 1.0.6). This bug would generally show up as the
122 Slime debugger higlighting the wrong form when the sldb-show-source
124 * bug fix: Branch forms are again annotated as branches in the sb-cover
126 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
127 on x86-64. (reported by Gregory Vanuxem)
128 * bug fix: EQUALP could return wrong results for structures with raw
129 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
130 word). (reported by Vjacheslav Fyodorov)
131 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
134 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
135 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
136 atomic compare-and-swap operations on threaded platforms.
137 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
138 allows assining a global minimum value to optimization qualities
139 (overriding proclamations and declarations).
140 * enhancement: closed over variables can be stack-allocated on x86
142 * performance bug fix: GETHASH and (SETF GETHASH) are once again
144 * optimization: slot definition lookup is now O(1). This speeds up
145 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
146 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
147 * optimization: ASSOC and MEMBER can now be open-coded for all
148 combinations of keyword arguments when second argument is constant
149 and SPEED >= SPACE. In other cases a specialized version is
151 * bug fix: using obsoleted structure instances with TYPEP and
152 generic functions now signals a sensible error.
153 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
154 (reported by Kristoffer Kvello)
155 * bug fix: backtrace construction is now more careful when making
156 lisp-objects from pointers on the stack, to avoid creating bogus
157 objects that can be seen by the GC.
158 * bug fix: defaulting of values in contexts expecting more than 7
159 variables now works on x86-64. (reported by Christopher Laux)
160 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
162 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
163 as the property-list of a symbol.
164 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
165 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
166 situations requested, are once again file-compileable. (reported
169 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
170 * MOP improvement: support for user-defined subclasses of
171 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
172 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
173 the expansion of DEFMETHOD, is responsible for generating a form
174 which creates a list of specializers when evaluated. Additional
175 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
176 debugging and introspective support.
177 * minor incompatible change: the (unsupported) spinlock interface
178 has changed: free spinlock now has the value NIL, and a held spinlock
179 has the owning thread as its value.
180 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
181 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
183 * enhancement: name of a socket-stream is now "a socket" instead of
185 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
186 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
187 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
188 for generic functions. (Thanks to Tobias C. Rittweiler)
189 * optimization: bignum printing speed has been improved by 20-40%
190 (depending on the bignum size.)
191 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
193 * bug fix: the cache used by the CLOS to store precomputed effective
194 methods, slot offsets, and constant return values is now thread and
196 * bug fix: generic function dispatch function updating is now thread
197 and interrupt safe (in the sense that the known issues have been
199 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
200 * bug fix: interning EQL-specializers is now thread and interrupt safe.
201 * bug fix: asdf systems with dependencies to the SB-POSIX or
202 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
203 * bug fix: interrupt safety of applicable method computation has been
206 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
207 * new contrib: sb-cover, an experimental code coverage tool, is included
209 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
210 significantly faster.
211 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
212 produce more efficient code on x86-64 if the compiler can determine
213 that (+ INDEX <constant>) does not require a bounds check and FOO
214 has an element type at least 8 bits wide.
215 * enhancement: a new, experimental synchronous timeout facility is
216 provided. Refer to SB-SYS:WITH-DEADLINE for details.
217 * enhancement: when a symbol name conflict error arises, the
218 conflicting symbols are always printed with a package prefix.
219 (thanks to Kevin Reid)
220 * enhancement: stepping is now once again supported on the SPARC and
221 MIPS platforms. (It is also now more likely to work on CheneyGC
223 * enhancement: sb-sprof can now also track and report accurate call
225 * bug fixes: the treatment of non-standard subclasses of
226 SB-MOP:SPECIALIZER is more correct.
227 * incompatible change: PURIFY no longer copies the data from the
228 dynamic space into the static and read-only spaces on platforms
229 that use the generational garbage collector
230 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
232 * bug fix: GC race condition occasionally resulting in crashes with
233 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
235 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
237 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
238 a STYLE-WARNING to be signalled (regression from 1.0.4.)
239 * bug fix: an asynchronous interrupt could previously leave the
240 system running with GC inhibited.
241 * bug fix: a DECLARE form evaluated at top-level now causes an error
242 rather than silently (or verbosely) returning NIL.
243 * bug fix: trying to dissassemble functions compiled at high DEBUG could
244 result in a error being signalled due to source form lookup errors.
245 (reported by Peter Graves)
247 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
248 * incompatible change: removed writer methods for host-ent-name,
249 host-ent-addresses -- changing the values did not update the DNS
251 * minor incompatible change: changed experimental JOIN-THREAD interface
252 * documentation: the manual now lists reader and writer methods
253 in class slot documentation sections. (thanks to Richard M Kreuter)
254 * documentation: unwinding from asyncronous events has been
255 documented as unsafe.
256 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
257 in multithreaded application code.
258 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
259 platforms. (thanks to James Anderson for the optimization hint)
260 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
262 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
263 variants no longer cons.
264 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
265 their NOT- variants no longer cons.
266 * optimization: EQUAL hash tables no longer use SXHASH for objects
267 of all data types, but instead use an EQL hash for types for which
268 EQUAL is the same as EQL.
269 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
270 are significantly faster.
271 * optimization: new STRING-OUTPUT-STREAM implementation is much
272 faster for large outputs and conses 30% less on average.
273 * enhancement: XREF information is now collected for references made
274 to global variables using SYMBOL-VALUE and a constant argument.
275 * enhancement: SIGINT now causes a specific condition
276 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
277 * bug fix: FILE-POSITION used to signal an error in some cases where
278 ANSI requires it to return NIL.
279 * bug fix: ADJUST-ARRAY is now interrupt-safe.
280 * bug fix: adding and removing fd-handlers is now interrupt-safe.
281 * bug fix: inlined calls to C now ensure 16byte stack alignment on
283 * bug fix: bad type declaration in the CLOS implementation has
284 been fixed. (reported by James Anderson)
285 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
286 (thanks to Lutz Euler)
287 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
288 required by the CLHS. (reported by Samium Gromoff)
289 * bug fix: dead unbound variable references now signal an error.
290 (reported by Marco Monteiro)
291 * bug fix: / with an unused value was being deleted in safe code.
292 (thanks to Marco Monteiro and Kevin Reid)
293 * bug fix: number of characters that can be written onto a single
294 line in a file is unlimited.
295 * bug fix: some GC deadlocks caused by asynchronous interrupts have
296 been fixed by inhibiting interrupts for when GC is disbled.
297 * bug fix: some interrupt safety issues with GETHASH, (SETF
298 GETHASH), CLRHASH and REMHASH have been fixed.
299 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
300 specifier no longer causes infinite recursion.
301 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
302 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
303 * bug fix: an &environment argument with the correct variable information
304 is passed to macros that are expanded during byte compilation.
305 (reported by Samium Gromoff)
306 * bug fix: a logic error was causing the hash function for bignums to
307 have excessive amounts of collisions. (reported by Faré Rideau)
308 * bug fix: modifying the contents of an array could change the return
309 value of SXHASH on that array, which is only allowed for strings
310 and bit vectors. (bug introduced in 0.9.16)
311 * improvement: the x86-64/darwin port now passes all tests (except
312 for the debugger tests) but should still be considered
313 experimental until this is fixed.
314 * improvement: a style-warning is signaled for CASE (etc) clauses with
315 duplicate keys. (thanks to Kevin Reid)
316 * improvement: macos/x86 and macos/x86-64 now use mach exception
317 handlers for dealing with illegal instructions (for trapping and
318 error handling) and memory protection violations (for GC).
320 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
321 * new platform: experimental support for x86-64/darwin (MacOS).
322 * incompatible change: the thread-safe (on most platforms) getaddrinfo
323 and getnameinfo sockets functions are used instead of gethostbyaddr
324 and gethostbyname, on platforms where the newer functions are available.
325 As a result, the ALIASES field of HOST-ENT will always be NIL on these
327 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
328 don't print the startup banner, but behave as if --noinform was passed
329 as a command line argument. (thanks to Kevin Reid)
330 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
331 * optimization: code using alien values with undeclared types is much faster.
332 * optimization: the compiler is now able to open code SEARCH in more cases.
333 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
334 * bug fix: using standardized COMMON-LISP special variables as loop
335 variables no longer signals bogus package lock violations. (reported
337 * bug fix: declaring local loop variables to be of a range-limited type
338 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
339 (reported by Andras Simon)
340 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
341 bugs remain on x86-64.)
342 * bug fix: the #= and ## reader macros now interact reasonably with
343 funcallable instances.
344 * bug fix: type-checks for function arguments were compiled using the
345 compiler policy settings of the wrong lexical environment (in 1.0.2
347 * bug fix: SHADOW accepts characters as string designators, as required
348 by the spec (thanks to Eric Marsden)
349 * bug fix: fixed GC safety issues when foreign functions are called with
350 non-base strings as arguments
351 * bug fix: more consistent error output for fatal-compiler-warnings, like
353 * bug fix: functions with proper names were showing up as NIL in
354 backtraces (reported by Edi Weitz, regression in 1.0.2)
356 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
357 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
358 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
359 produce more efficient code on the x86 if the compiler can determine
360 that (+ INDEX <constant>) does not require a bounds check and FOO
361 has an element type at least 8 bits wide.
362 * bug fix: references to undefined variables are handled the same way
363 in toplevel forms as in the normal compiler
364 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
366 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
367 (thanks to Jon Buller)
368 * improvement: faster compilation times for complex functions
369 * improvement: added readlink support to SB-POSIX (thanks to Richard
372 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
373 * improvement: experimental support for mach exception handling on
374 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
376 * improvement: support for GBK external format.
377 (thanks to Chun Tian (binghe))
378 * improvement: the debugger now displays variables that have been closed
379 over, in code compiled with (DEBUG 2) or higher
380 * improvement: support for executable cores on NetBSD (thanks to
382 * new feature: added a RESTART-FRAME debugger command
383 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
384 be used to provide an implementation for FILE-POSITION on Gray streams
385 (thanks to Eric Marsden)
386 * new feature: add syslog, openlog and closelog support to SB-POSIX
387 (thanks to Richard Kreuter)
388 * optimization: the function call overhead in code compiled with
389 a high DEBUG optimization setting is significantly
390 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
391 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
392 (thanks to Tony Martinez)
393 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
394 evaluated once (reported by Kevin Reid)
395 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
396 works again (thanks to Kilian Sprotte)
397 * bug fix: an error is signaled for tagbodies with duplicate tags
398 (thanks to Stephen Wilson)
399 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
400 * bug fix: Win32 port can now handle foreign code unwinding Lisp
401 stack frames from alien callbacks.
402 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
403 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
404 65k pages (thanks to David Woodhouse)
405 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
407 changes in sbcl-1.0.1 relative to sbcl-1.0:
408 * new platform: FreeBSD/x86-64, including support for threading.
409 * new feature: the compiler stores cross-referencing information
410 abount function calls (who-calls), macroexpansion (who-macroexpands)
411 and special variables (who-binds, who-sets, who-references) for code
412 compiled with (< SPACE 3). This information is available through the
413 sb-introspect contrib.
414 * new feature: users may subclass SEQUENCE, and have instances of
415 these classes interoperate with standard Common Lisp functions if
416 a number of methods are defined. (This feature is experimental
417 and the interface subject to change based on feedback from SBCL
418 users and the general community)
419 * improvement: sb-sprof traces call stacks to an arbitrary depth on
420 x86/x86-64, rather than the previous fixed depth of 8
421 * improvement: another pthread back-end of mutex "pthread-futex".
422 * bug fix: non-ascii command-line arguments are processed correctly
423 (thanks to Yaroslav Kavenchuk)
424 * bug fix: non-required arguments were not passed correctly when a method
425 defined using DEFMETHOD was called from a mop-generated method using
426 CALL-NEXT-METHOD (reported by Pascal Costanza)
427 * bug fix: recursion is now permitted in accessors through
428 SLOT-UNBOUND. (reported by Pascal Costanza)
429 * bug fix: an error was signaled at startup if the HOME environment
430 variable was defined, but had an empty value (reported by Peter Van Eynde)
431 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
432 signal a PROGRAM-ERROR, not a BUG.
433 * bug fix: SB-POSIX:READDIR works when built with large file support.
434 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
436 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
437 for better automatic indentation support. (thanks to Matt Pillsbury)
438 * optimization: loading generic functions no longer takes O(n^2) time,
439 proportional to the amount of methods in the generic function
440 (reported by Todd Sabin and Jeremy Brown)
441 * optimization: the FIND and POSITION family of sequence functions
442 are significantly faster on arrays whose element types have been
444 * improvements to the Windows port:
445 ** Intermittent heap corruption problems have been fixed. (thanks
446 to Alastair Bridgewater)
447 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
449 ** Lisp is able to unwind foreign exception frames from alien
450 callbacks. (thanks to Alastair Bridgewater)
452 changes in sbcl-1.0 relative to sbcl-0.9.18:
453 * improvement: experimental support for threading on FreeBSD/x86.
454 (thanks to NIIMI Satoshi)
455 * improvement: runtime option --dynamic-space-size can be used
456 to set the size of the dynamic space reserved on startup.
457 * improvement: floating point modes in effect are now saved in
458 core, and restored on startup.
459 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
460 startup, not time since first call to GET-INTERNAL-REAL-TIME.
461 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
462 threads are not running after *SAVE-HOOKS* have run.
463 * improvement: writes to CLOS instance slots are type-checked in code
464 compiled with (SAFETY 3)
465 * improvement: floating-point exception handling on FreeBSD (thanks to
467 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
468 (thanks to Zach Beane)
469 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
471 * improvement: added support for the Shift-JIS external format.
472 (contributed by NIIMI Satoshi)
473 * improvement: callbacks are supported on Linux/PPC. (thanks to
475 * bug fix: compiler bug triggered by a (non-standard) VALUES
476 declaration in a LET* was fixed. (reported by Kaersten Poeck)
477 * bug fix: file compiler no longer confuses validated and already
478 dumped structurres. (reported by Kaersten Poeck)
479 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
480 fill-pointer signals a type-error as required. (thanks to
482 * bug fix: disassemly of funcallable instances works.
483 * bug fix: single stepping on PPC.
484 * bug fix: fix thread-safety problems in the type system (generally
485 manifesting as nonsensical errors like "STRING is a bad type specifier
486 for sequences" or "The value 1 is not of type FIXNUM")
487 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
488 (reported by Josip Gracin)
489 * bug fix: an error is signaled for attempts to displace arrays with
490 incompatible element types (thanks to Mario Mommer)
491 * bug fix: more correct handling of wide characters in debug info
492 (bug reported by Attila Lendvai and fixed by Juho Snellman)
493 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
494 and don't cause extra consing
495 * optimization: MAP and MAP-INTO are significantly faster on vectors
496 whose elements types have been declared.
497 * Improvements to SB-SPROF:
498 ** Support for allocation profiling
499 ** Reduced profiling overhead, especially for long profiling runs
500 * Improvements to the Windows port:
501 ** floating point exceptions are now reported correctly.
502 ** stack exhaustion detection works partially.
503 ** more accurate GET-INTERNAL-REAL-TIME.
504 ** better RUN-PROGRAM behaviour in case of errors in the child
506 ** PROBE-FILE now simplifies pathnames correctly.
507 ** DIRECTORY now works correctly with :WILD-INFERIORS.
508 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
509 1st 1970 (although time timezone data will be unreliable).
510 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
512 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
513 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
514 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
516 * bug fix: two potential GC deadlocks affecting threaded builds.
517 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
518 returns T (reported by Anton Kazennikov)
519 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
520 with non-variable places
521 * bug fix: remove a race condition in the setting of
522 funcallable-instance functions, this should make threaded CLOS
523 code more stable against memory faults.
524 * bug fix: corruption of specials when unbinding is interrupted by an
525 asynchronous unwind (reported by Hannu Koivisto)
526 * improvement: the debugger will now also display local variables that
527 are only used once, for code compiled with a DEBUG optimization quality
530 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
531 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
532 * incompatible change: External-format support for FFI calls. The
533 SB-ALIEN:C-STRING no longer implies an ASCII
534 external-format. Instead the string is subject to external-format
535 conversion. Additionally return values of type SB-ALIEN:C-STRING
536 are by default of type (SIMPLE-ARRAY CHARACTER), not
537 SIMPLE-BASE-STRING. For an alien type that behaves like the old
538 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
539 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
540 * incompatible change: SB-EXT package no longer contains the
541 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
542 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
544 * incompatible change: the single-stepper is no longer available
545 on Alpha, Mips and Sparc platforms.
546 * minor incompatible change: the direct superclasses of
547 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
548 not (STANDARD-OBJECT FUNCTION). This makes the
549 class-precedence-lists of GENERIC-FUNCTION and
550 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
552 * minor incompatible change: the default stream external format on
553 non-unicode SBCL is no longer detected from the locale, but is always
555 * new feature: Add a version of evaluator that uses an interpreter instead
556 of the compiler. EVAL still uses the compiler by default, to switch it
557 to use the interpreter, set the value of the variable
558 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
559 * minor incompatible change: the single-stepper REPL has been merged
560 with the normal debugger (see the "Stepping" heading of the debugger help
561 for more details). The debugger command STEP will no longer switch
562 to the single-stepper REPL.
563 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
564 for a type now works.
565 * bug fix: loading of default sysinit file works. (thanks to Leonid
567 * bug fix: better detection of circularities in the file-compiler.
568 (reported by Marco Monteiro)
569 * bug fix: the CL pathname functions now work with files that have
570 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
571 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
572 systems with Unix98 pty semantics.
573 * bug fix: ASDF-INSTALL will now work with bsd tar.
574 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
576 * bug fix: timers expiring in dead threads no longer cause a
577 type-error (reported by Paul "Nonny Mouse").
578 * bug fix: thanks to more lightweight single-stepper instrumentation,
579 code compiled with (DEBUG 3) will compile and execute significantly faster,
580 and will have more accurate type-inferencing than before
581 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
582 whose bindings are modified
583 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
584 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
585 ** replace dummy implementations of CL:MACHINE-INSTANCE and
586 CL:SOFTWARE-VERSION with proper version
588 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
589 * feature: implemented the READER-METHOD-CLASS and
590 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
591 as specified by AMOP.
592 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
594 * optimization: faster LOGCOUNT implementation on x86 and x86-64
595 (thanks to Lutz Euler)
596 * optimization: hashing of general arrays and vectors has been
597 improved. (reported by Any Fingerhut)
598 * enhancement: SB-INTROSPECT is now able to find definitions of
599 profiled functions. (thanks to Troels Henriksen)
600 * enhancement: compiler-macro expansion applies now to FUNCALL forms
601 as well, allowing compiler-macros for SETF-functions to expand.
602 * enhancement: step-instrumentation no longer wraps known
603 single-value functions with multiple-value context, allowing
604 better type inference.
605 * fixed bug #337: use of MAKE-METHOD in method combination now works
606 even in the presence of user-defined method classes. (reported by
607 Bruno Haible and Pascal Costanza)
608 * fixed bug #339(c): if there are applicable methods not part of any
609 long-form method-combination group, call INVALID-METHOD-ERROR.
610 (reported by Bruno Haible)
611 * fixed bug #361: the :FUNCTION initarg in the protocol for
612 initialization of methods can now be used to override
613 internally-produced optimized functions. (reported by Bruno
615 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
616 system-provided lambda expression no longer cause warnings about
617 unbound #:|pv-table| symbols.
618 * bug fix: improved the handling of type declarations and the
619 detection of violations for keyword arguments with non-constant
621 * bug fix: potentially erronous calls to PATHNAME and
622 MERGE-PATHNAMES were being flushed in some cases.
623 (reported by Richard Kreuter)
624 * bug fix: compiled calls to TYPEP were mishandling obsolete
625 instances. (reported by James Bielman and Attila Lendvai)
626 * bug fix: format strings with ~> without matching ~< no longer
627 trigger an AVER, but signal an understandable error instead.
628 (reported by Antonio Martinez)
629 * bug fix: specifying an output-file without a pathname-type for
630 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
631 of input-file instead of "fasl". (reported by Robert Dodier)
632 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
633 (reported by James Y Knight).
634 * bug fix: compiler-macros-function did not consider the environment
635 argument for shadowing by local functions.
636 * bug fix: compiler-macros expansion was inhibited by local INLINE
638 * bug fix: inline expansions of known functions were subject to
639 step-instrumentation in high debug policies, leading to problems
641 * bug fix: compiler failed to differentiate between different CONS
643 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
644 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
645 * bug fix: #368: incorrect use of expressed vs. upgraded array
647 * bug fix: #306a: more precise unions of array types.
648 * thread-safety improvements:
649 ** CONDITION-WAIT could return early on Linux, if the thread was
650 interrupted and subsequently continued with SIGCONT.
651 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
653 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
654 * added support for the ucs-2 external format. (contributed by Ivan
656 * minor incompatible change: pretty printing of objects of type
657 (cons symbol) is, in the default pprint-dispatch-table, now
658 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
660 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
661 called later in a class's lifetime, possibly as late as when the
662 first instance of the class is created. Previously,
663 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
664 class became finalizeable.
665 * fixed bug: FILE-POSITION sometimes returned inconsistent results
666 for multibyte external-format streams. (thanks to "vbzoli")
667 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
668 with :ALLOCATION :CLASS inherited from superclasses of the
670 * fixed bug: anonymous classes can now be created using the :NAME
671 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
672 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
673 * fixed bug: core-files saved with :EXECUTABLE T can again be
674 executed when SBCL_HOME isn't set. (reported by James Knight)
675 * fixed bug: toplevel LOCALLY forms with declarations could
676 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
677 * fixed bug: printing from several different threads using different
678 values of *print-case* could cause invalid output, due to
679 some internal special variables of the printer not being bound
680 thread-locally (reported by Max Mikhanosha)
681 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
682 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
683 (reported by Pascal Costanza)
684 * fixed bug: FINALIZE-INHERITANCE is called from
685 REINITIALIZE-INSTANCE on classes when the class has previously
686 been finalized, as required by AMOP.
687 * minor code generation optimizations:
688 ** better register allocation in CLOS dispatching functions
689 ** overflow detection when coercing signed bytes to fixnums on x86-64
690 ** is now implemented with one IMUL instruction instead of three shifts
691 ** more efficient bit-vector access on x86 and x86-64
692 ** more efficient access to raw structure slots on x86 and x86-64
693 * fixed some bugs revealed by Paul Dietz' test suite:
694 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
697 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
698 * feature: thread support on Solaris/x86, and experimental thread support
700 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
702 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
703 default method for SLOT-UNBOUND.
704 * minor incompatible change: prevent the user from specializing the
705 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
706 somewhat counter to the intent of the protocol, I (CSR) think, and
707 additionally it just doesn't work in SBCL as currently
708 implemented, thanks to optimizations (that are always valid for
709 the other three SLOT-VALUEish functions, but not for the setter).
710 * minor incompatible change: the :SB-LDB feature is now enabled by
711 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
712 the low-level debugger.
713 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
714 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
715 down to an SB-SYS:FD-STREAM, allowing direct communication in
716 more cases. Temporary files and pipes are still used as a fallback
718 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
719 on the non-reentrant C functions localtime(3) and gmtime(3).
720 * bug fix: global optimization policy was not visible in LOCALLY and
722 * bug fix: class objects can be used as specializers in methods.
723 (reported by Pascal Costanza)
724 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
725 with a type error. (reported by blitz_ on #lisp)
726 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
727 failed with a type error. (reported by Pascal Bourguignon)
728 * bug fix: merging pathnames against defaults with :DIRECTORY
729 starting with '(:RELATIVE :BACK) should preserve the :BACK.
730 (reported by James Y Knight)
731 * bug fix: saving large (>2GB) cores on x86-64 now works
732 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
733 constant index and a (simple-array (signed-byte 32)) array
734 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
735 error (patch by Robert J. Macomber)
736 * feature: TIME now displays the amount of run-time spent in GC
737 * bug fix: The debugger now does a better job of respecting
738 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
739 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
740 workaround for bug 403.)
741 * bug fix: floating point exception handling now works on FreeBSD
742 (thanks to NIIMI Satoshi)
743 * fixed some bugs revealed by Paul Dietz' test suite:
744 ** MISC.641: LET-conversion were not supposed to work in late
746 * improvements to the Win32/x86 port:
747 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
748 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
750 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
751 * new feature: source path information is generated for macro-expansion
752 errors for use in IDE's like Slime (thanks to Helmut Eller)
753 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
754 * bug fix: compiler does not loop forever on an invalid type in
756 * improvement: compilation of most CLOS applications is significantly
758 * optimization: added a limited bytecode compiler for simple toplevel
759 forms, speeding up compilation and FASL loading
760 * bug fix: the statistical profiler now properly distinguishes anonymous
763 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
764 * minor incompatible change: in sbcl-0.9.11 (but not earlier
765 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
766 into subclasses of the finalized class. Now user calls to
767 FINALIZE-INHERITANCE finalize just the one class, and calls by the
768 system return before any subclasses are finalized.
769 * minor incompatible change: The reader no longer ignores errors
770 regarding non-existent packages in #+ and #- feature tests.
771 * new feature: command line options --no-sysinit, --no-userinit to
772 inhibit loading the corresponding init files
773 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
774 for sending data through UDP sockets (thanks to François-René Rideau)
775 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
776 error is signalled instead (thanks to François-René Rideau)
777 * minor incompatible change: Error signalling behaviour of lexical
778 operations violating package locks has changed slightly. Refer to
779 documentation on package locks for details.
780 * enhancement: EVAL can process IF-expressions without resorting to the
782 * bug fix: Some operations involving SIN, COS, or TAN weren't being
783 constant-folded properly due to missing out-of-line functions.
784 (reported by Mika Pihlajamäki)
785 * bug fix: LISTEN sometimes returned T even in cases where no data was
786 immediately available from the stream
787 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
788 were reversed. (reported by Levente Mészáros)
789 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
790 when given a vector displaced to an adjustable array. (reported
792 * bug fix: derivation of float boundaries from numbers outside the
793 appropriate float range (reported by John Wiseman)
794 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
796 * bug fix: precomputing cache entries for generic functions with
797 some subclasses of specializers as yet invalid does not attempt to
798 fill a cache line with a negative offset. (reported by Levente
800 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
801 allow condition class objects as arguments to DOCUMENTATION and
802 (SETF DOCUMENTATION); only find and set documentation for
803 structure names for the STRUCTURE doc-type. (suggested by Gary
805 * improvements to the Win32/x86 port:
806 ** (user-homedir-pathname) and default initialization file
807 locations now know about the user's "Documents and Settings"
808 directory (thanks to Yaroslav Kavenchuk)
809 ** run-program is implemented (thanks to Mike Thomas)
810 ** sockets support (thanks to Timothy Ritchey)
811 ** better backtrace support (thanks to Alastair Bridgewater)
812 ** sb-grovel supported
813 ** asdf-install and sb-posix work somewhat
814 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
815 * improvements to the Solaris/x86 port:
816 ** works on Solaris 11/Solaris Express
817 ** floating-point exception handling support
818 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
819 * fixed some bugs revealed by Paul Dietz' test suite:
820 ** REMOVE-METHOD returns its generic function argument even when
821 the method is not one of the generic functions' methods.
822 ** objects declared as MEMBER types can be admissible arguments to
824 ** printing characters should simply be printed by the FORMAT ~:C
826 ** compiler failure when compiling functions with hairy constant
827 defaults for optional parameters.
828 ** compiler produces wrong code when MAYBE-INLINE-expanding a
829 function, which is already optimized.
831 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
832 * new platform: experimental support for SBCL x86/Darwin, including
833 MacOS X 10.4.5 on Intel.
834 * new feature: Unicode character names are now known to the system
835 (through CHAR-NAME and NAME-CHAR).
836 * new feature: the filesystem location of SBCL's core file is
837 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
838 * minor incompatible change: the contrib modules SB-POSIX and
839 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
840 this change is to make it easier to distribute
841 location-independent binaries.
842 * bug fix: as implied by AMOP, standardized classes no longer have
843 slots named by external symbols of public packages. (reported by
845 * optimization: calling functions via constant symbols -- as in
846 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
847 via the function object as in (FUNCALL #'FOO).
848 * optimization: on x86 and x86-64, the calling convention for the
849 case of dealing with an unknown number of values has been altered
850 to be friendlier to the prediction heuristics implemented,
851 particularly with reference to CALL and RET pairing. (thanks to
852 Alastair Bridgewater)
853 * optimization: on x86, the code for access to thread-local symbol
854 values and binding/unbinding of thread-local symbols is smaller.
855 (thanks to Alastair Bridgewater)
856 * enhancement: CONSTANTP is now able to determine constantness of
857 more complex forms, including calls to constant-foldable standardized
858 functions and some special forms beyond QUOTE.
859 * fixed bug: occasional GC crashes on Solaris/x86
860 * optimization: x86-64 supports stack allocation of results of simple
861 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
862 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
863 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
864 (thanks to James Knight)
865 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
866 by the standard (thanks to Stephen Compall)
868 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
869 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
870 be used for bundling the runtime and the core file into one
871 executable binary. This feature is not currently supported on all SBCL
872 platforms. (thanks to James Bielman and NIIMI Satoshi)
873 * new feature: a generational or ephemeral garbage collector is now
874 the default on the PowerPC platform (both Linux and Darwin). The
875 old Cheney (stop and copy) collector is a build-time option.
876 (thanks to Cyrus Harmon, after Raymond Toy)
877 * minor incompatible change: the method by which SBCL finds its
878 contributed modules has changed; it no longer relies on symbolic
879 links from an $SBCL_HOME/systems directory, but searches directly
880 in the subdirectories of $SBCL_HOME.
881 * enhancement: the dynamic heap size on the Linux/PPC platform is
882 markedly larger, even using the older Cheney garbage collector.
883 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
884 certain complicated string types. (reported by Gary King)
885 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
886 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
887 * fixed bug: DOCUMENTATION on structures no longer signals an
888 error if no structure type of the right name exists. (reported by
890 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
891 applicable non-standard primary or :AROUND method on
892 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
893 no longer cause unbound variable errors. (reported by Kevin Reid)
894 * fixed bug: metacircle resolution in cases where methods have slots
895 added before the slots from STANDARD-METHOD. (reported by Jean
897 * fixed bug: the Power PC assembly code for calling into lisp
898 sometimes computed its return address wrongly (depending on the
899 15th bit of the address, fixed at link time). This used to
900 manifest itself in a segmentation violation while building PCL.
901 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
902 many others over the years)
903 * fixed bug: ROOM no longer reports silly numbers for stack usage if
904 the stack pointer (treated as a signed integer) is negative.
905 (thanks to Peter van Eynde)
907 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
908 * new platform: experimental support for the Windows operating
909 system has been added. (thanks to Alastair Bridgewater)
910 * fixed several bugs in and robustified the PPC FFI (including
911 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
912 * bug fix: allow non-simple string symbol names (reported by Paul
914 * bug fix: interrupt handling on NetBSD (thanks to Richard M
916 * bug fix: saving a core corrupted callbacks on x86/x86-64
917 * bug fix: closed a loophole in metacircularity detection and
918 grounding in the PCL implementation of CLOS.
919 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
920 (reported by tomppa on #lisp)
921 * bug fix: clear the direction flag on Lisp -> C transitions, as
922 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
923 (reported by Andrej Grozin and Hendrik Maryns)
924 * optimization: major improvements to GC efficiency on GENCGC platforms
925 * optimization: faster implementation of EQUAL
926 * optimization: emit more efficient opcodes for some common
927 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
928 * optimization: several other minor code-generation improvements on x86-64
929 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
931 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
932 * minor incompatible change: (SETF CLASS-NAME) and (SETF
933 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
934 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
935 by ANSI as a generic function, and for consistency (SETF
936 GENERIC-FUNCTION-NAME) is treated likewise.
937 * fixed bug #233.b: make constraint propagation notice when a variable
938 value is changed after it is referenced but before it is used
939 * fixed bug #296: no more arbitrary behaviour from filenames with
940 odd characters as --load arguments. (reported by Adam Warner)
941 * fixed bug #390: :CHARACTER-SET pathname components now work as
942 expected. (reported by Tim Daly Jr)
943 * fixed bug #391: complicated :TYPE intersections in slot
944 definitions no longer cause an error in PCL internals.
945 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
946 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
947 returning the number of octets which would be written to the
948 file-stream. (thanks to Robert J. Macomber)
949 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
950 arguments to the internal stream functions. (thanks to David
952 * bug fix: stack exhaustion is now handled on the x86/SunOS
953 platform. (thanks to Daisuke Homma)
954 * bug fix: more accurate ROOM results on GENCGC platforms
955 * optimization: improved type inference for arithmetic-for
956 index variables in LOOP
957 * optimization: faster floating-point SQRT on x86-64
958 * fixed some bugs revealed by Paul Dietz' test suite:
959 ** DOCUMENTATION returns NIL instead of "" for method combinations
960 that don't have a docstring
962 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
963 * minor incompatible change: (SETF CLASS-NAME) and (SETF
964 GENERIC-FUNCTION-NAME) are no longer generic functions, and
965 therefore are not customizeable by user code (as seems to be at
966 least permitted and maybe required by AMOP). As a consolation,
967 however, the SBCL implementation of these functions now calls
968 REINITIALIZE-INSTANCE as specified by AMOP.
969 * bug fix: slot-definition documentation is propagated as per ANSI
970 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
971 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
972 argument, as do eslotd initialization methods. (from Pascal
973 Costanza's "Closer" project)
974 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
975 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
977 * bug fix: it is now possible to have more than one subclass of
978 STANDARD-GENERIC-FUNCTION without causing stack overflow.
979 (reported by Bruno Haible, Pascal Costanza and others)
980 * bug fix: the dependent update protocol now works for generic
981 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
983 * bug fix: condition-class instances corresponding to
984 DEFINE-CONDITION forms are now created eagerly. (reported by
985 Kalle Olavi Niemitalo on comp.lang.lisp)
986 * bug fix: floating point printing is more accurate in some
987 circumstances. (thanks to Simon Alexander)
988 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
989 merged with *DEFAULT-PATHNAME-DEFAULTS*.
990 * bug fix: callbacks on OS X now preserve stack-alignment.
991 (thanks to Cyrus Harmon)
992 * enhancement: the x86-64 disassembler is much better at
993 disassembling SSE instructions. (thanks to Lutz Euler)
994 * enhancement: CHECK-TYPE on a variable now allows type-inference
995 (thanks to Vincent Arkesteijn)
996 * optimization: improved performance of EUC-JP external format.
997 (thanks to NIIMI Satoshi)
998 * optimization: performance improvements to IO on file streams of
999 :ELEMENT-TYPE CHARACTER
1000 * optimization: much faster memory allocation on x86-64
1001 * optimization: faster garbage collections (latency and throughput)
1003 * optimization: faster bignum subtraction and division on x86 and x86-64
1004 (thanks to Lutz Euler)
1005 * optimization: more accurate type-inference for loops that use a
1006 floating point index variable or a negative step.
1008 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1009 * bug fix: add a workaround to SBCL looping infinitely at startup on
1010 Linux kernels with apparently buggy implementations of personality().
1011 (thanks to Svein Ove Aas)
1012 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1013 (thanks to David Lichteblau)
1014 * bug fix: local bindings shadow global symbol macros for
1015 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1016 * optimization: non-open coded uses of numeric comparison operators
1017 (e.g. >) no longer cons when called with more than one parameter
1018 on platforms supporting dynamic-extent allocation.
1019 * enhancement: saving cores with foreign code loaded is now supported
1020 on MIPS/Linux in addition to the previously supported platforms.
1021 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1022 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1023 * bug fix: division by zero in sb-sprof when no samples were collected
1024 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1025 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1026 (thanks to Alexander Kjeldaas)
1027 * new feature: ignoring the return values of destructive functions
1028 like NREVERSE signals a compile-time style-warning.
1029 (thanks to Kevin Reid)
1030 * new feature: the alignment of alien structure fields can be
1031 explicitly specified. (thanks to Cyrus Harmon)
1032 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1033 and dump core on SIGQUIT
1035 ** incompatible change: threads do not inherit values of specials
1036 from their parents (see manual)
1037 ** bug fix: threads stacks belonging to dead threads are freed by the
1038 next exiting thread, no need to gc to collect thread stacks anymore
1039 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1040 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1041 ** bug fix: dynamic variable and thread start related gc lossage
1042 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1044 * fixed some bugs revealed by Paul Dietz' test suite:
1045 ** SUBTYPEP is slightly more accurate on heinously complicated
1046 CONS types where some of the members have uncertain (in the
1047 NIL, NIL sense) type relationships to each other.
1049 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1050 no longer purified unless :PURIFY T is explicitly specified.
1051 ** Non-purified cores are significantly smaller than before
1053 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1054 * new feature: timers based on Zach Beane's excellent timer package
1055 * added support for the following external formats: koi8-u,
1056 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1057 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1058 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1059 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1060 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1061 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1062 * incompatible change: a threaded SBCL will no longer revert to
1063 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1064 * bug fix: interrupts are disabled until startup is complete; no
1065 more sigsegvs when receiving a signal to soon
1066 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1068 * bug fix: add a workaround for the memory randomization features in
1069 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1070 workaround will only be in effect on systems with the proc filesystem
1072 * bug fix: printing objects of type HASH-TABLE signals a
1073 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1075 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1076 have been running for over 50 days. (reported by Gilbert Baumann)
1077 * bug fix: the logic for getting names of functions gets less
1078 confused when confronded with alternate-metaclass
1079 funcallable-instances. (reported by Cyrus Harmon)
1080 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1081 consistent, even on internal alternate-metaclass objects.
1082 * bug fix: flush closure information collected by physical
1083 environment analysis prepass before the main pass. (bug reported
1085 * bug fix: compiler pack phase does not modify a hashtable, which is
1086 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1087 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1088 may be expanded. (reported by Alexander <malishev> on c.l.l)
1089 * bug fix: no more style-warnings when compiling code using LAST
1090 under high-SPEED low-SPACE settings. (reported by David Wallin
1092 * fixed some incompatibilities between SBCL's MOP and the MOP
1094 ** the METAOBJECT class is now implemented;
1095 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1096 STANDARD-OBJECT, as required;
1097 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1098 now compatible (as required by VALIDATE-SUPERCLASS); there
1099 remains a consistency requirement over the presence or absence
1100 of the FUNCTION class in the superclass list of finalized
1101 classes; see the manual for more details;
1102 ** the :AROUND method for COMPUTE-SLOTS on
1103 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1104 requested slot ordering.
1106 ** bug fix: parent thread now can be gc'ed even with a live
1108 ** bug fix: binding a special with PROGV to no value is not
1109 the same as that symbol not having been bound (thanks to
1111 ** bug fix: binding specials is thread safe (thanks to
1113 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1114 and SB-SPROF do not die with 'no handler for signal XX in
1115 interrupt_handle_now(..)' anymore
1116 ** bug fix: WITH-TIMEOUT works with multiple running threads
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1119 the :method-class keyword argument.
1121 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1122 * new port: the Solaris operating system on x86 processors is now
1123 mostly supported, though some rough edges in the environment
1124 remain. (thanks to Daisuke Homma)
1125 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1126 space for its heap. (thanks to Thiemo Seufer)
1127 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1128 closures. (thanks to Thiemo Seufer)
1129 * minor incompatible change: eof selects abort in the debugger.
1130 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1131 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1133 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1134 deprecated, and will go away in a future revision of SBCL.
1135 * minor incompatible change: GC-ON and GC-OFF are no longer
1136 implemented with a counter, it does not matter how many times gc
1137 is switched on or off
1138 * bug fix: discriminating functions for generic function classes
1139 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1140 make invalid assumptions about method precedence order. (reported
1142 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1143 platforms. (thanks to Thiemo Seufer)
1144 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1145 * bug fix: correct pathname printing with printer escaping is on.
1146 (thanks to Kevin Reid)
1147 * bug fix: complex VOP definitions in "user-space" no longer trigger
1148 package locks. (reported by Zach Beane)
1149 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1150 for funcallable-instance functions, and is no different from
1152 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1153 characters in the current readtable. (reported by Nicholas Neuss)
1154 * bug fix: SUBTYPEP on various CONS types returns more a more
1155 accurate acknowledgment of its certainty.
1156 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1157 lists in some cases. This partially fixes bug 384.
1158 * flush all standard streams before prompting in the REPL and the
1160 * bug fix: signal handling and triggering gc do not conflict
1161 directly anymore, in particular a high frequency sb-sprof does
1162 not prevent gc from running
1163 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1164 approximation for timezone and DST information between the
1165 universal time epoch and the smallest negative 32-bit time_t.
1166 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1167 year 1899 so long as the encoded time is non-negative.
1168 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1169 the correct answer for non-constant rotate amounts when performing
1170 an inline 32-bit rotation.
1172 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1173 there is only one thread in the session
1174 ** bug fix: memory leak for streams created in one thread and
1175 written to in another
1176 ** bug fix: lockup when compiled with gcc4
1177 ** bug fix: race that allows the gc to be triggered when gc is
1179 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1180 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1182 ** bug fix: finished threads are gc'ed properly
1183 * fixed some bugs revealed by Paul Dietz' test suite:
1184 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1185 has arguments with a different set of applicable methods from
1186 the orignal arguments.
1187 ** The type error thrown by MAP now has a correct expected-type
1189 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1190 name a compiled function.
1191 ** Three MISC tests where a large bignum was improperly coerced to
1192 a float that couldn't represent that bignum during type
1193 derivation were fixed.
1194 ** SUBTYPEP can now handle types involving the negation of a
1195 list-form FUNCTION type.
1196 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1197 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1198 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1200 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1201 * New feature: Experimental support for bivalent streams: streams
1202 opened with :element-type :default now allow character and binary
1203 (unsigned-byte 8) I/O
1204 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1205 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1207 * bug fix: no more highly sporadic "couldn't check whether ~S is
1208 readable" when reading a stream and an interrupt hits in the middle
1209 of a select system call
1210 * compiler better recognizes complex arrays (reported by Eduardo
1212 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1213 for variables on SBCL built with linkage-tables. (reported by Luis
1215 * various error reporting improvements.
1216 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1217 (Thanks to Thiemo Seufer)
1218 * bug fix: nested reader invokations work correctly
1219 * bug fix: it is possible to have simultaneous references to foreign
1220 code and foreign data with the same name.
1222 ** added x86-64 support
1223 ** incompatible change: the threading api now works with thread
1224 objects instead of thread ids
1225 ** bug fix: threads are protected from signals and interruption when
1226 starting up or going down
1227 ** bug fix: a race where an exiting thread could lose its stack to gc
1228 ** bug fix: don't halt on infinite error in threads if possible
1229 ** fixed numerous gc deadlocks introduced in the pthread merge
1230 ** bug fix: fixed thread safety issues in read and print
1231 ** bug fix: debugger doesn't hang on session lock if interrupted at
1232 an inappropriate moment
1233 ** bug fix: run-program is now thread safe(r)
1234 ** bug fix: inner with-recursive-lock no longer releases the mutex
1235 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1238 EXPECTED-TYPE slots filled.
1239 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1241 ** MAKE-CONDITION accepts classes as type-designators.
1242 ** COMPILE may never return NIL.
1243 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1244 range before calling Unix time functions
1246 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1247 * numerous signal handling fixes to increase stability
1248 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1249 * minor incompatible change: we now correctly canonize default
1250 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1251 as per the MOP, rather than the historical (INITARG INITFUNCTION
1252 INITFORM). (reported by Bruno Haible)
1253 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1254 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1255 for more information.
1256 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1257 pathname is a directory pathname.
1258 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1259 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1261 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1262 :METACLASS argument in addition to a class name. (reported by
1263 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1264 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1265 (reported by Bruno Haible; patch from Gerd Moellmann)
1266 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1268 * bug fix: more cleanups to the floating point exception handling on
1269 x86-64 (thanks to James Knight)
1270 * bug fix: the compiler does not try to constant fold calls of
1271 COERCE and BIT-* functions when they return freshly constructed
1272 objects. (reported by Nikodemus Siivola and Paul Dietz)
1273 * optimization: improved the MIPS versions of generic (in the
1274 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1275 * optimization: direct conversion of (unsigned-byte 32) to floats on
1276 the PowerPC platform.
1277 * optimization: structure instances with raw slots now use less
1278 memory, and probably show better memory locality. (thanks to
1280 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1281 are treated as requests for stack allocation on the x86-64,
1282 Alpha-32, PPC and SPARC platforms.
1283 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1284 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1285 (thanks to Tony Martinez)
1287 ** gcing a dead thread can no longer lead to lockups
1288 ** threads block signals until they are set up properly
1289 ** errno is no longer shared by threads
1290 ** interrupt-thread restores the eflags register on x86
1291 ** fixed some lockups due to gc/thread interaction
1292 * fixed some bugs revealed by Paul Dietz' test suite:
1293 ** invalid dotted lists no longer raise a read error when
1294 *READ-SUPPRESS* is T
1295 ** signal an error if a symbol that names a declaration is used
1296 as the name of a type, or vice versa
1297 ** allow using the (declare (typespec var*)) abbreviation for
1298 (declare (type typespec var*)) with all type specifiers
1299 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1300 ** always return NIL from PPRINT-POP when OBJECT is NIL
1301 ** don't signal errors when pretty-printing malformed LABELS,
1302 FLET or MACROLET forms
1303 ** declarations in a DOLIST body are also in scope for the
1305 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1307 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1310 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1311 * fixed cross-compiler leakages that prevented building a 32-bit
1312 target with a 64-bit host compiler.
1313 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1314 opened with :IF-EXISTS :OVERWRITE.
1315 * fixed bug 281: error for an invalid qualifier in a short-form method
1316 combination method is not signalled until the faulty method is called.
1317 * bug fix: iteration variable type inferrer failed to deal with open
1318 intervals. (reported by Alan Shields)
1319 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1320 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1321 or more runtime options were provided to the sbcl binary.
1322 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1323 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1324 * optimization: there's now a fast-path for fixnum arguments in the
1325 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1326 * optimization: the code generated on x86-64 is more compact thanks
1327 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1328 * bug fix: floating-point exception handling now partly works on
1329 x86-64. (Thanks to James Knight)
1330 * improvement to the MIPS backend from Thiemo Seufer: C-style
1331 64-bit long long arguments and return values to alien functions
1333 * fixed some bugs revealed by Paul Dietz' test suite:
1334 ** the type-error signalled from WARN has a filled-in DATUM slot.
1335 ** the type-error required when a stream is not associated with
1336 a file has the stream as its datum.
1337 ** type-errors on single-floats on x86-64 no longer have
1338 :INVALID-OBJECT as the datum
1339 ** the type-errors signalled for invalid function names now have
1340 a correct expected type
1341 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1342 for structures defined with a :TYPE. Documentation strings for
1343 typed structures are no longer immediately discarded
1344 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1345 broadcast streams and synonym streams. FILE-LENGTH now also works
1346 on broadcast streams.
1348 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1349 * incompatible change: the --noprogrammer option, deprecated since
1350 version 0.7.5, has been removed. Please use the equivalent
1351 --disable-debugger option instead.
1352 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1354 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1355 inoperational for a while now) has been completely removed.
1356 * null lexical environments are now printed as #<NULL-LEXENV>,
1357 significantly reducing the amount of clutter in typical backtraces.
1358 * documentation on weak pointers, finalization, and after GC hooks
1359 has been added to the manual.
1360 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1361 as other specialized array types, is much faster. SUBSEQ and
1362 COPY-SEQ on such arrays have also been sped up.
1363 * optimization: EQL is now more efficient when at least other argument
1364 is known to be of type (OR FIXNUM (NOT NUMBER)).
1365 * fixed bug: the runtime is now less vulnerable to changes in the
1366 size of the SBCL object on OS X, and virtual memory is reserved for
1367 all spaces that need to be at a fixed address.
1368 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1369 * fixed bug: finalizers and after GC hooks that cause consing are now
1371 * fixed bug: compiler error messages and summaries are now printed to
1372 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1373 * fixed inference of the upper bound of an iteration variable.
1374 (reported by Rajat Datta).
1375 * fixed bug 211e: calling local functions with duplicated constant
1376 keyword argument no longer causes a bogus style warning about an
1378 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1379 ftype declaration to be disregarded. (reported by Dave Roberts)
1380 * fixed bug 373: caused by erronous compilation of references to alien
1381 variables in the runtime on ppc/darwin.
1382 * fixed bug 376: CONJUGATE type deriver.
1383 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1384 (reported by Baughn on #lisp)
1385 * a cross-compiler bug on non-x86 platforms has been identified and
1386 fixed. (thanks to Bruno Haible)
1387 * improvements to the MIPS runtime code for increased stability.
1388 (thanks to Thiemo Seufer)
1389 * increased the maximimum compact environment size to allow
1390 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1391 * partial workaround for bug 135: don't name the function we're
1392 calling for hairy cases of EVAL, so as not to accumulate one environment
1393 entry per call to EVAL. (reported by Kevin Reid)
1394 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1395 * optimization: type testing for non-vector arrays should be faster.
1396 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1398 * bug fix: setting 31st element of a bit vector to zero did not work
1400 * bug fix: redefining a class definition which failed due to a
1401 previous accessor / function clash now works (but see BUGS entry
1402 #380 for more problems in this area). (thanks to Zach Beane)
1403 * the long form of DEFINE-METHOD-COMBINATION disables method group
1404 checking when given a single method group with pattern *. (thanks
1406 * on x86 compiler supports stack allocation of results of simple
1407 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1408 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1409 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1410 * fixed some bugs related to Unicode integration:
1411 ** the restarts for recovering from input and output encoding
1412 errors only appear when there is in fact such an error to
1414 * fixed some bugs revealed by Paul Dietz' test suite:
1415 ** MISC.549 and similar: late transformation of unsafe type
1416 assertions into derived types caused unexpected code
1418 ** SCALE-FLOAT type deriver is less wrong.
1419 ** type derivers for EXP, LOG and similar functions compute result
1420 types for complex arguments better.
1421 ** (MISC.563) CONJUGATE type deriver works for very restricted
1423 ** out-of-line type testers for character strings are available.
1424 ** EQUAL compiler transform understands specialness of objects
1426 ** accessing double-floats stored on the stack now works on x86-64.
1427 ** debugger internals could sometimes create invalid lispobjs,
1428 resulting in GC crashes.
1429 ** MISC.548: type check weakening can convert required type into
1431 ** initialization forms for bindings are not in scope of free special
1434 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1435 * incompatible change: thread support for non-NPTL systems has
1436 been removed - locking is buggy and unreliable. A threaded
1437 SBCL build will now warn at startup and refuse to create
1438 new threads, unless futex support is detected in the kernel
1439 * incompatible change: the top level REPL now has only an ABORT
1440 restart associated with it, not TOPLEVEL and ABORT as it used to.
1441 TOP and TOPLEVEL are now available as debugger commands for
1442 returning to the top level.
1443 * incompatible change: forms evaluated in the REPL now use the
1444 global optimization policy.
1445 * incompatible change: user- and system-initialization files are
1446 no longer processed with LOAD, but by READ and EVAL; hence the
1447 global optimization policy, startup package, readtable, etc,
1449 * The .fasl file format number has been incremented because of
1450 various incompatible changes.
1451 * internal entry point details and argument counts no longer appear
1452 in backtraces unless explicitly requested by setting
1453 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1454 * built-in and standard functions no longer have names like "top
1455 level local call to FOO".
1456 * fixed bug 32: functions defined in non-null lexical environments
1457 now have more legible printed representation
1458 * fixed bug 33: functions defined in non-null lexical environemnts
1459 are now more amenable to inspection by INSPECT.
1460 * workaround for bug 354: XEPs no longer appear in backtraces unless
1461 explicitly requested.
1462 * fixed bug: receiving the signal which results from attempting to
1463 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1464 SIGBUS on other *BSDs) on architectures where the C stack is also
1465 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1466 to debugger confusion or wild instability. (reported by Cheuksan
1468 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1469 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1470 (reported by Lutz Euler)
1471 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1472 compiled code. (reported by Paul Dietz)
1473 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1474 specialized parameter to a method to something that is not TYPEP
1475 the specializer is now possible.
1476 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1477 face of package deletion.
1478 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1479 pathname. (reported by David Tolpin on c.l.l)
1480 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1481 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1482 STRING-STREAM and FILE-STREAM, respectively.
1483 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1484 than previously; in addition, its test suite is now run on build.
1485 (thanks to Luke Gorrie)
1486 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1487 * optimization: added a immediate representation for single-floats
1489 * fixed some bugs revealed by Paul Dietz' test suite:
1490 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1491 ** attempting to create a package with a colliding nickname causes
1492 correctable errors to be signalled.
1493 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1494 ** several x86-64 backend bugs related to sign-extension of immediate
1497 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1498 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1499 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1501 * fixed loading of multiply forward-referenced layouts.
1502 (thanks to Cheuksan Wang)
1503 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1504 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1505 (see COMPILE-FILE documentation for details of :PRINT option)
1506 * fixed bugs 19 and 317: fixed-format floating point printing is
1507 more accurate. This also fixes a bug reported by Adam Warner
1508 related to the ~@F format directive.
1509 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1511 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1512 dispatch table if the from-char is a dispatch macro character.
1513 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1514 by the wordsize no longer ignore the last word. (reported by Lutz
1516 * fixed bug in type checking of dynamic-extent variables. (reported
1518 * optimization: sequence traversal functions use their freedom to
1519 coerce function designators to functions.
1520 * optimization: code with many calls to CLOS methods specialized on
1521 CLOS classes has had redundant type checks removed.
1522 * optimization: added declarations to speed up operations that access
1523 the internal character database (for example STRING-UPCASE)
1524 * optimization: comparison operations between floats and sufficiently small
1525 fixnums no longer create extra rationals
1526 * fixed some bugs related to Unicode integration:
1527 ** portions of multibyte characters at the end of buffers for
1528 character-based file input are correctly transferred to the
1529 start of the buffer at the next read.
1530 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1531 passing it through to OPEN.
1532 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1533 argument given to internal calls to OPEN.
1534 ** the built-in comment readers (introduced by character sequences
1535 ";" and "#|") are more forgiving to encoding errors; they will
1536 STYLE-WARN and then attempt to resync the stream at a character
1537 boundary. (thanks to Teemu Kalvas)
1538 * fixed some bugs in the x86-64 port:
1539 ** Negative short int return values from c-calls are sign-extended
1541 ** The stack is aligned to 16-bytes for c-calls, as required by
1542 the ABI. (thanks to Cheuksan Wang)
1543 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1544 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1545 for enums. (thanks to Vincent Arkesteijn)
1546 ** Multiple small optimizations and bugfixes for floating point
1548 * fixed some bugs revealed by Paul Dietz' test suite:
1549 ** Space, Tab, Linefeed, Return and Page have the invalid
1550 secondary constituent character trait.
1551 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1553 ** WITH-INPUT-FROM-STRING should only update the index place on
1555 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1556 works more reliably.
1557 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1558 with one-dimensional arrays with fill pointers.
1559 ** TYPE-OF failed on a complex with an integer realpart and a
1561 ** compiler failure during type inference for the code of form
1562 (IF (EQL X (THE ...)) ...) (MISC.535).
1564 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1565 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1566 platform. The port supports 61-bit fixnums, large memory spaces
1567 and reloading shared object files.
1568 * enhancement: saving cores with foreign code loaded is now
1569 supported on ppc/Darwin in addition to the previously supported
1571 * enhancement: the statistical profiler now walks deeper into the
1572 call stack for more meaningful call-graphs and accrued time
1573 reports (x86/x86-64 only). It also now reports time spent in
1575 * enhancement: it is now possible to trace most individual methods
1576 of a generic function in addition to tracing the generic function
1578 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1580 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1581 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1582 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1583 produces an error. (thanks to Vincent Arkesteijn)
1584 * bug fix: NAMESTRING on pathnames with :WILD components in their
1585 directories works correctly. (thanks to Artem V. Andreev)
1586 * fixed bug 125: compiler preserves identity of closures. (reported
1588 * bug fixed: functions with &REST arguments sometimes failed with
1589 "Undefined function" when compiled with (DEBUG 3). (reported by
1591 * bug fix: overflow during compiling of setting element of a bit
1592 vector with constant index and value. (reported by Timmy Douglas)
1593 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1594 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1595 * on x86 compiler partially supports stack allocation of dynamic-extent
1597 * GO and RETURN-FROM do not check the extent of their exit points
1598 when compiled with SAFETY 0.
1599 * fixed some bugs related to Unicode integration:
1600 ** encoding and decoding errors are now much more robustly
1601 handled; it should now be possible to recover even from invalid
1602 input or output to the terminal. (thanks to Teemu Kalvas)
1603 ** provided a first cut at implementing STRING-TO-OCTETS and
1604 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1605 ** altered the SB-MD5 contributed module slightly, changing the
1606 interface just enough for it to be supportable for builds where
1607 lisp characters are not eight bits.
1608 * fixed some bugs revealed by Paul Dietz' test suite:
1609 ** the FORMATTER-generated functions for ~V[ conditionals require
1610 the correct number of arguments.
1611 ** READ-FROM-STRING returns the mandated second value when applied
1612 to displaced strings.
1613 ** the #\Rubout and #\Backspace characters are treated as invalid
1614 constituent characters by the tokenizer.
1616 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1617 * new feature: reloading changed shared object files with
1618 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1619 * new feature: references to foreign variables and functions
1620 can now be compiled and loaded before the corresponding shared
1621 object file is loaded, as long as the foreign definitions are
1622 available at runtime.
1623 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1624 supported. (thanks to Dan Debertin)
1625 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1626 just keywords. (thanks to Vincent Arkesteijn)
1627 * fixed bug #331: structure-class instances corresponding to
1628 DEFSTRUCT forms are now created eagerly.
1629 * fixed bug #345: backtraces from calls to undefined functions work
1630 on x86 as well. Related bug #61 is now also partially fixed on x86
1631 (backtraces from throws to unknown catch tags.)
1632 * bug fix: lambda-list parsing is now stricter vrt. order and number
1633 of lambda-list keywords.
1634 * bug fix: as specified by AMOP, an error is signalled if a
1635 class-option appears multiple times in a DEFCLASS form. (reported
1637 * bug fix: RANDOM can be compiled when the compiler derives the type
1638 of its numeric argument as a disjoint set of small integers.
1639 (reported by Paul Dietz)
1640 * bug fix: starting a core saved with shared objects loaded when
1641 those objects are not available no longer causes threaded SBCL to
1642 hang. (reported by Sean Ross)
1643 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1645 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1646 argument, which must always be NIL. (reported by Kalle Niemitalo)
1647 * bug fix: printing 1.0d+23 no longer results in an error.
1648 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1649 * bug fix: structure slot setters preserve evaluation order. (thanks
1651 * fixed some bugs related to Unicode integration:
1652 ** RUN-PROGRAM can allow its child to take input from a Lisp
1653 stream. (reported by Stefan Scholl)
1654 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1655 locales. (reported by Ken Causey)
1656 * fixed some bugs revealed by Paul Dietz' test suite:
1657 ** INCF, DECF and REMF evaluate their place form as specified in
1659 ** FORMATTER expands ~{ ~} iteration directives with V or #
1660 parameters correctly.
1661 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1662 consequent uses no arguments correctly.
1663 ** the system has a partial understanding of the (COMPLEX RATIO)
1665 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1668 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1669 * new feature: a build-time option (controlled by the :SB-UNICODE
1670 keyword feature, enabled by default) for building the system with
1671 support for the entire 21-bit character space defined by the
1673 * new feature: the system now has rudimentary external-format
1674 support; the primary user-visible change at this time is that
1675 characters with the high bit set (such as lower-case-e-acute) will
1676 print correctly to a terminal in a UTF-8 environment.
1677 * minor incompatible change: BASE-CHAR no longer names a class;
1678 however, CHARACTER continues to do so, as required by ANSI.
1679 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1680 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1681 has been moved to the SB-EXT package (temporarily re-exported from
1683 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1684 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1686 * fixed bug #351: better error-handlind and reporting for malformed
1688 * fixed bug #350: bignum-printing is now more memory-efficient,
1689 allowing printing of very large bignums, eg. (expt 2 10000000).
1690 (reported by Bruno Haible)
1691 * fixed bug #302: better primitive-type selection for intersection
1693 * fixed bug #308: non-graphic characters now all have names, as
1694 required. (reported by Bruno Haible)
1695 * bug fix: redefining a class with different superclasses now correctly
1696 removes it from the direct-subclasses of its previous superclasses.
1697 (reported by David Morse)
1698 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1699 new value now works. (reported by Bruno Haible)
1700 * bug fix: correct canonicalization of multiple non-standard slot
1701 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1702 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1703 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1704 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1705 now exists, an signals an error.
1706 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1707 during macroexpansion for non-iteration variables as well. (reported
1708 by Bruno Haible for CMUCL)
1709 * bug fix: Cyclic structures and unprintable objects in compiler
1710 messages no longer cause errors. (reported by Bruno Haible)
1711 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1712 and READ-FROM-STRING all now return a primary value of NIL if
1713 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1714 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1715 (reported by Bruno Haible for CMUCL)
1716 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1717 do not designate a condition. (reported by Bruno Haible for
1719 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1720 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1721 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1722 permitting resolution in favour of any of the conflicting symbols.
1723 (reported by Bruno Haible for CMUCL)
1724 * bug fix: EQUAL compiler optimizations is less aggressive on
1725 strings which can potentially compare true despite having distinct
1726 specialized array element types.
1727 * bug fix: unit enumerations can be defined without dividing by
1728 zero. (thanks to Vincent Arkesteijn)
1729 * FORMAT compile-time argument count checking has been enhanced.
1730 (report from Bruno Haible for CMUCL)
1731 * a partial workaround for the bug 262: the compiler does not try to
1732 inline-expand a local function doing RETURN-FROM from a deleted
1733 BLOCK. (thanks to Peter Denno for the bug report and to David
1734 Wragg for the simple test case)
1735 * fixed some bugs revealed by Paul Dietz' test suite:
1736 ** RENAME-PACKAGE allows all package designators as new package
1738 ** constraint propagation and lambda variable substitution are
1739 more cautious in dealing with partially deleted code.
1740 ** compiler sometimes lost reoptimization passes.
1741 ** CERROR, given a condition as condition designator, passes its
1742 remaining arguments to the continue format control without
1744 ** Case-altering FORMAT directives work correctly on non-ASCII
1746 ** The REAL type specifier handles bounds outside the floating
1747 point ranges without signalling FLOATING-POINT-OVERFLOW.
1748 ** Functions with IR1-transformations can create intercomponent
1749 references to global functions.
1750 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1752 ** FORMAT ~R treats a nil value for its first parameter correctly.
1754 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1755 * enhancement: saving cores with foreign code loaded is now
1756 supported on x86/NetBSD and sparc/Linux in addition to the previously
1757 supported platforms.
1758 * bug fix: on some platforms repeated installations caused multiple
1759 copies of HTML documentation to be installed -- should not happen
1760 any more. (reported by Stefan Scholl)
1761 * bug fix: parsing self-recursive alien record types multiple times
1762 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1763 original patch by Helmut Eller for CMUCL)
1764 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1765 (thanks to Richard Kreuter)
1766 * bug fix: defining classes whose accessors are methods on existing
1767 generic functions in other (locked) packages no longer signals
1768 bogus package lock violations. (reported by François-René Rideau)
1769 * bug fix: special variables as DEFMETHOD parameters no longer have
1770 associated bogus type declarations. (reported by David Wragg and
1772 * bug fix: read-write consistency on streams of element-type
1773 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1774 * bug fix: redefiniton of the only method of a generic function with
1775 no DEFGENERIC no longer emits a full WARNING. In addition,
1776 redefinition of generic functions with no DEFGENERIC to an
1777 incompatible lambda list now signals an error. (thanks to Zach
1779 * bug fix: DEFGENERIC now works even when there's a function of the
1780 same name in an enclosing lexical environment. (thanks to Zach
1782 * fixed compiler failure, caused by instrumenting code during
1783 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1784 * optimization: added loop analysis and improved register allocation
1785 * fixed some bugs revealed by Paul Dietz' test suite:
1786 ** POSITION on displaced vectors with non-zero displacement
1787 returns the right answer.
1788 ** (SIMPLE-STRING) is a valid type specifier for sequence
1790 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1792 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1793 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1795 ** The justification version of the FORMAT ~< directive treats
1796 non-zero minpad parameter correctly.
1798 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1799 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1800 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1801 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1802 the supported interface.
1803 * new feature: Single-stepping of code compiled with DEBUG 2 or
1804 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1805 * new feature: saving cores with foreign code loaded is now
1806 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1807 Timothy Moore's work for CMUCL)
1808 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1809 parameters to * instead of NIL if no initform is supplied.
1810 (reported by Johan Bockgaard)
1811 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1812 correctly when FUNCALL appears as the car of the form. Note:
1813 despite this FUNCALL forms are not currently subject to
1814 compiler-macro expansion. (port of Raymond Toy's fix for the
1815 same from CMUCL, reported by Johan Bockgaard)
1816 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1817 (thanks for Teemu Kalvas)
1818 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1819 case instead of terminating the iteration (thanks for Julian
1820 Squires, Sean Champ and Raymond Toy)
1821 * bug fix: incorrect expansion of defgeneric that caused a style
1822 warning. (thanks for Zach Beane)
1823 * bug fix: slot accessor effective method computation works properly
1824 for classes with multiple non-standard applicable methods on
1825 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1826 * on x86 compiler supports stack allocation of results of LIST and
1827 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1828 CMUCL implementation by Gerd Moellmann)
1829 * fixed some bugs revealed by Paul Dietz' test suite:
1830 ** FORMAT strings with both the ~<~:;~> form of the justification
1831 directive and pretty-printing directives cause an error.
1833 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1834 * incompatible change: the internal functions
1835 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1836 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1837 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1838 instead of the old functions.
1839 * new feature: on platforms where "dladdr" is available foreign
1840 function names now appear in backtraces. (based on Helmut Eller's
1842 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1843 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1845 * documentation: ASDF manual is now distributed with SBCL.
1846 * bug fix: SBCL can now load its contributed modules with REQUIRE
1847 even if the system-provided entries have been removed from
1848 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1850 * bug fix: backtraces involving undefined functions or assembly
1851 routines are more informative. (thanks to Brian Downing)
1852 * bug fix: mutually referent alien structures now work correctly.
1853 (reported by Rick Taube)
1854 * bug fix: structures defined by WITH-ALIEN can be referred to
1855 within other definitions in the same WITH-ALIEN.
1856 * bug fix: division operators (MOD, TRUNCATE and the like) with
1857 constant zero divisors and integer dividends no longer generate
1859 * bug fix: provide default methods for INPUT-STREAM-P and
1860 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1861 * bug fix: improve the detection and resolution of MOP metacycles
1862 (where effective method computation depends on the generic function
1863 having its effective method computed). (reported by Bruno Haible)
1864 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1865 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1866 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1867 * bug fix: pathnames with relative directory components can be
1868 represented relative to default pathnames.
1869 * optimization: in taking the GCD of bignums, reduce the two bignums
1870 to approximately the same size (using Euclid's algorithm) before
1871 applying the more sophisticated binary GCD. (thanks to Juho
1873 * optimization: COUNT on bitvectors now operates word-at-a-time.
1874 * optimization: ASH with a positive, but not necessarily constant,
1875 (leftwards) shift, when in a modular context, is compiled to a
1877 * fixed some bugs revealed by Paul Dietz' test suite:
1878 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1879 if the corresponding argument is NIL.
1880 ** FORMAT directives accepting a minpad parameter treat negative
1882 ** literal commas as character parameters to FORMAT directives are
1884 ** literal spaces directly after ~<Newline> directives within a
1885 format-logical-block (~:< ~@:>) do not induce :FILL-style
1886 conditional newlines.
1887 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1888 as specified: it no longer includes conditional newlines).
1889 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1891 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1892 :PER-LINE-PREFIX argument does not evaluate to a string.
1894 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1895 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1896 the manual for details; for now, package locks can be disabled by
1897 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1898 if no major problems are found then it is likely that they will be
1899 compiled in unconditionally.
1900 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1901 unsupported operators on all platforms. To load a shared library
1902 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1903 object file, link it into a shared library outside of SBCL and
1904 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1906 * bug workaround: ROOM T and the GC were not working together
1907 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1908 reporting facility aren't preserved. That reporting has been
1909 disabled, so now until and unless someone figures out how to make it
1910 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1911 * minor incompatible change: as threatened around sbcl-0.8.0, the
1912 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1913 The symbols are also exported from SB-PCL for backwards
1914 compatibility, but more so than before SB-PCL should be treated as
1915 an implementation-internal package.
1916 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1918 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1919 It is known as of this release not to work on the Alpha, however.
1920 * fixed bug #167: errors signalled due to illegal syntax in method
1921 bodies are now more legible.
1922 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1923 designators and can hence be used with TYPEP.
1924 * fixed bug #333: CHECK-TYPE now ensures that the type error
1925 signalled, if any, has the right object to be accessed by
1926 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1927 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1928 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1930 * fixed bug #334: programmatic addition of slots using specialized
1931 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1932 and :ALLOCATION :CLASS slots.
1933 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1934 float exponents. (rereported by Peter Seibel)
1935 * fixed a bug: #\Space (and other whitespace characters) are no
1936 longer considered to be macro characters in standard syntax by
1937 GET-MACRO-CHARACTER.
1938 * fixed bug: initialization of condition class metaobjects no longer
1939 causes an instance of the condition to be created. (reported by
1941 * fixed bug: it is now possible to have slots such that
1942 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1943 description is neither :INSTANCE nor :CLASS.
1944 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1945 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1946 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1947 non-standard SLOT-DEFINITION-ALLOCATION values.
1948 * fixed bugs: various race conditions handling exiting threads.
1949 CL-PPCRE's thread torture test now passes.
1950 * fixed arguably-a-bug: GC time proportional to number of threads in
1951 system even when most of them are idle
1952 * optimization: improved performance of BIT and SBIT on bit-vectors.
1953 * .fasl file incompatibility: The fasl file version number has
1954 been incremented because of changes associated with package locks.
1956 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1957 * minor incompatible change: the system no longer provides
1958 optimization hints (conditions which are TYPEP
1959 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1960 mode; these hints are still emitted when higher SPEED optimization
1962 * new contrib module: a sampling profiler (profiling by statistical
1963 sampling, rather than by instrumenting functions) is available as
1964 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1965 * the behaviour of the standard function ED is now customizeable by
1966 third parties through a hook variable: see ED's documentation
1967 string for information on the protocol.
1968 * the compiler no longer emits efficiency notes for (FUNCALL X)
1969 when the type of X is uncertain under default optimization
1971 * fixed bug 276: mutating a binding of a specialized parameter to a
1972 method to something that is not TYPEP the specializer is now
1974 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1975 exhibit the required behaviour on the x86 platform. (thanks to
1976 Peter van Eynde, Eric Marsden and Bruno Haible)
1977 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1978 even for difficult arguments. (reported by Peter Graves)
1979 * fixed bug 141a: the backquote printer now descends quoted
1981 * fixed another bug in backquote printing: no more destructive
1982 modification of the form's list structure. (reported by Brian
1984 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1985 move between its address being taken and the call to
1986 interrupt_thread, fixing a crashing race condition.
1987 * the SB-POSIX contrib implementation has been adjusted so that it
1988 no longer exhibits ridiculously poor performance when constructing
1989 instances corresponding to C structs.
1991 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1992 * minor incompatible change: the sb-grovel contrib now treats C
1993 structures as alien (in the sense of SB-ALIEN) objects rather than
1994 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1995 has implications for memory management of client code
1996 (sb-grovel-returned objects must now be manually managed) and for
1997 type safety (alien objects now have full types).
1998 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1999 used to control emission of compiler diagnostics, rather than the
2000 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2001 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2002 quality should be considered deprecated.
2003 * install.sh now installs the user manual as well
2004 * (not quite a new documentable feature, but worth considering in
2005 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2006 beginnings of a semantically meaningful condition hierarchy is
2007 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2009 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2010 designator as the defaults argument.
2011 * fixed bug: Displaced arrays whose displaced-to array has become
2012 too small now cause ARRAY-DIMENSION to signal an error, providing
2013 for safer bounds-checking. (reported by Bruno Haible)
2014 * fixed bug: DEFCLASS slot definitions with identical :READER and
2015 :WRITER names now signal a reasonable error. (reported by Thomas
2017 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2019 * fixed bug: Invalid slot specification errors now print correctly.
2020 (thanks to Zach Beane)
2021 * fixed bug 320: Shared to local slot value transfers in class
2022 redefinitions now happen corrently from superclasses as
2023 well. (reported by Bruno Haible)
2024 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2026 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2027 improper lists correctly. (reported by Bruno Haible)
2028 * fixed bug 313: source-transform for <fun-name> was erroneously
2029 applied to a call of a value of a variable with name <fun-name>.
2030 (reported by Antonio Menezes Leitao)
2031 * fixed bug 307: The obsolete instance protocol ensures that
2032 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2033 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2034 remove dead unknown-values globs from the middle of the stack.
2035 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2036 easier to resume long computations after using *BREAK-ON-SIGNALS*
2037 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2038 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2039 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2040 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2041 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2042 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2044 * fixed a (fairly theoretical) bug in string printing: if
2045 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2046 does not have array-element-type equal to the most general string
2048 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2049 function name. (thanks to Nikodemus Siivola)
2050 * fixed bug: compile-time format string checker failed on
2051 non-closed ~{. (reported by Thomas F Burdick)
2052 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2053 don't warn on legal loop code involving more than one aggregate
2054 boolean. (thanks to Nikodemus Siivola)
2055 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2056 in keyword position to MAKE-PACKAGE was not being reported as
2058 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2059 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2061 * fixed bugs 280 and 312: the checking for multiple definitions in a
2062 file is less likely to become confused by uses of inline
2064 * fixed bug: the #S reader macro performs the keyword coercion
2065 specified for slot names. (reported by Kalle Niemitalo)
2066 * fixed bug: lambda lists may contain symbols whose names start with
2067 & but are not lambda-list-keywords; their occurrence triggers a
2069 * fixed bug 321: define-method-combination argument lambda lists do
2070 not cause invalid code to be generated when &optional or &aux
2071 variables have default values. (reported by Bruno Haible)
2072 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2073 have CLOS classes; this appears also to have fixed reported
2074 instability in stack exhaustion detection.
2075 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2076 MUFFLE-WARNING when no associated restart is present is now
2078 * optimization: rearranged the expansion of various defining macros
2079 so that each expands into only one top-level form in a
2080 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2082 * optimization: used a previously little-used slot in symbols to
2083 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2085 * fixed some bugs revealed by Paul Dietz' test suite:
2086 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2088 ** Fixed an optimization bug related to inheritance of initforms
2089 from local to shared slots.
2090 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2091 ** CAST optimizer forgot to flush argument derived type.
2092 ** print/read consistency on floats is now orders of magnitude
2093 more likely. (thanks also to Bruno Haible for a similar report
2095 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2096 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2097 for objects of type REAL. Make it so.
2098 ** SXHASH is specified (infelicitously) to respect similarity,
2099 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2100 it so. (thanks to Markus Ziegler)
2101 ** on the Alpha, the compiler succeeds in compiling functions
2102 returning a known number of arguments greater than 63.
2103 ** fixed handling of invalid NIL arguments in keyword position
2105 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2107 ** provided a readably-printable representation for RANDOM-STATE
2109 ** ensured that pathnames, where they have a namestring, always
2110 print using #P"..." syntax.
2112 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2113 * Support for the forthcoming 2.0 version of the NetBSD kernel
2114 running on x86 hardware has been added. (thanks to Perry
2115 E. Metzger most immediately, and others for their past work)
2116 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2117 current release version; SBCL's OpenBSD support had been broken
2118 since about the time of OpenBSD's switch to ELF binary format.)
2119 * [placeholder for DX summary]
2120 ** user code with &REST lists declared dynamic-extent, under high
2121 speed or space and low safety and debug optimization policy.
2122 * The manual has been converted to Texinfo format and the debugger
2123 chapter from the cmucl manual has been added.
2124 * A facility has been added to extract documentation strings from
2125 sbcl and store them as Texinfo-formatted snippets for inclusion in
2126 the manual (via Texinfo's @include directive)
2127 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2128 the test case to Dave Roberts)
2129 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2130 pointers. (reported by Sean Ross)
2131 * bug fix: PROFILE output is printed nicely even for large numerical
2132 values. (thanks to Zach Beane)
2133 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2134 greater than 32 handle EOF correctly.
2135 * bug fix: on X86 an immediate argument of the IMUL instruction is
2136 correctly printed in disassembly. (thanks to Lutz Euler)
2137 * bug fix: class slots in redefined classes preserve their old
2138 values. (thanks to Bruno Haible and Nikodemus Siivola)
2139 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2140 succeeds. (reported by Marco Baringer)
2141 * fixed some bugs revealed by Paul Dietz' test suite:
2142 ** READ-SEQUENCE now works on ECHO-STREAMs.
2143 ** RATIONALIZE works more according to its specification. (thanks
2146 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2147 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2148 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2149 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2150 should matter to you only if you rebind the printer control
2151 variables and then find you want different bindings in the
2152 debugger than in the ordinary execution of your program.)
2153 * The runtime build system has been tweaked to support building
2154 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2155 assembler and linker. (thanks to Nikodemus Siivola)
2156 * Unbound, undefined, undeclared variables now trigger full
2157 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2158 more often programmer error than deliberate exploitation of undefined
2160 * optimization: the hash algorithm for strings has changed to one
2161 that is less vulnerable to spurious collisions. (thanks to Juho
2163 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2164 needless bounds checking. (thanks to Juho Snellman)
2165 * optimization: implemented multiplication as a modular
2166 (UNSIGNED-BYTE 32) operation on the PPC backend.
2167 * fixed some bugs revealed by Paul Dietz' test suite:
2168 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2170 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2171 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2172 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2174 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2176 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2178 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2179 their output stream on EOF from read.
2180 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2181 have been read to end-of-file.
2182 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2184 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2185 description of determination of which consecutive characters
2187 ** Printing the "Space" character with escaping on now yields "#\\ ",
2188 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2189 ** Reading floating-point numbers with *READ-BASE* set to a number
2190 less than 10 works correctly.
2191 ** Reading floating-point numbers with *READ-BASE* set to a number
2192 more than 10 works correctly.
2193 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2194 the readtable currently in effect.
2196 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2197 * minor incompatible change: parsing of namestrings on a physical
2198 (Unix) host has changed; numbers after the final #\. in a
2199 namestring are no longer interpreted as a version field. This is
2200 intented to be largely invisible to the user, except that the
2201 meaning of the namestring "*.*.*" has changed: it now refers to a
2202 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2203 should usually be replaced by
2204 (make-pathname :name :wild :type :wild :version :wild)
2205 with the added benefit that this is more likely to be portable.
2206 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2207 option to OPEN now signals an error if the file being opened
2208 exists; this may have an impact on existing code.
2209 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2210 signals received, on the PowerPC platforms (both Linux and
2211 Darwin). (thanks to Pierre Mai for pointing out the location of
2213 * several fixes on OS X: The system now builds and runs cleanly on
2214 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2215 (thanks to Brian Mastenbrook)
2216 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2217 second-resolution: integer multiples of 1/3600 between -24 and 24.
2218 (thanks to Vincent Arkesteijn)
2219 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2220 type when called with 1 argument; PEEK-CHAR checked type of
2221 PEEK-TYPE only after having read first character from a
2222 stream. (reported by Peter Graves)
2223 * bug fix: the garbage collector now has much better locality
2224 behaviour, and in particular no longer treats all memory as being
2225 exhausted when many small objects point to each other in a deeply
2227 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2228 recognized as being TYPEP their class.
2229 * bug fix: the PUSHNEW documentation string has been corrected.
2230 (thanks to Vincent Arkesteijn)
2231 * bug fix: defaulting of the value for the last of an atypically
2232 large number of multiple values being bound was not being
2233 performed correctly on the Alpha or PPC platforms
2234 * optimization: implemented multiplication as a modular
2235 (UNSIGNED-BYTE 32) operation on the x86 backend.
2236 * optimization: SEARCH on simple-base-strings can now be open-coded.
2237 (see also contrib/compiler-extras.lisp for inspiration for
2238 teaching the compiler about the Boyer-Moore algorithm).
2239 * value, returned by MAX (and MIN) called with several EQUALP, but
2240 not EQL, arguments now does not depend on compiler settings.
2241 * fixed some bugs revealed by Paul Dietz' test suite:
2242 ** in stack analysis liveness information is propagated from
2243 non-local entry points.
2244 ** pathwise CAST removing failed when the CAST node did not start
2246 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2247 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2249 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2250 host is already defined.
2251 ** RENAME-FILE works on streams instead of signalling an internal
2253 ** PEEK-CHAR uses the current readtable when determining whether
2254 or not a character is whitespace.
2255 ** MERGE-PATHNAMES handles the case when the pathname does not
2256 specify a name while the default-pathname specifies a version.
2257 ** Pathnames now stand a better chance of respecting print/read
2259 ** Attempting to use standardized file system operators with a
2260 pathname with invalid :DIRECTORY components signals a
2262 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2264 ** DIRECTORY on logical pathnames is more correct.
2265 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2266 signal a TYPE-ERROR if their argument is not a stream.
2267 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2268 designator argument does not designate a stream.
2269 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2270 examining the synonym.
2271 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2273 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2274 element-type larger than ([UN]SIGNED-BYTE 32).
2276 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2277 * When built with the :SB-FUTEX feature, threaded builds now take
2278 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2279 for faster/more reliable mutex and condition variable support.
2280 * Incompatible change (but one you probably shouldn't have been using
2281 anyway): the interface and code for arbitrating between multiple
2282 threads in the same user session has been redesigned.
2283 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2284 error when called without an explicit environment argument.
2285 (thanks to Nikodemus Siivola)
2286 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2287 confused about their position. (thanks to Adam Warner and Gerd
2289 * bug fix: Pretty printing backquoted forms with unquotations in the
2290 argument list position of various code constructs such as LAMBDA
2291 now works correctly. (reported by Paul Dietz)
2292 * bug fix: Pretty printing unquotations no longer loses all
2293 stream position information.
2294 * optimization: performance of string output streams is now less
2295 poor for multiple small sequence writes.
2296 * optimization: performance of CSUBTYPEP in the presence of complex
2297 expressions involving CONS and NOT many times has been improved.
2298 (reported by Paul Dietz)
2299 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2301 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2302 streams for the REPL, for the debugger, and so on. (thanks to
2304 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2305 * fixed some bugs revealed by Paul Dietz' test suite:
2306 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2307 is ignored if it is too small, rather than propagating through
2308 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2309 ** extremely complex negations of CONS types were not being
2310 sufficiently canonicalized, leading to inconsistencies in
2312 ** VALUES tranformer lost derived type.
2314 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2315 * fixed a bootstrapping bug: the build process no longer assumes
2316 that the various BOOLE-related constants have the same value in
2317 host and target lisps. (noted by Paul Dietz' test suite on an
2318 SBCL binary built from CLISP)
2319 * The system can now be dynamically linked on the MIPS platform,
2320 which enables dynamic loading of foreign code from Lisp. (thanks
2321 to Ralf Baechle for discussions on the MIPS ABI)
2322 * The system now records debugging information for its own source
2323 files in a filesystem-position-independent manner, relative to
2324 the "SYS" logical host.
2325 * fixed a compiler bug: MV-LET convertion did not check references
2326 to the "max args" entry point. (reported by Brian Downing)
2327 * tweaked disassembly notes to be less confident about proclaiming
2328 some instruction as an LRA. (thanks to Brian Downing)
2329 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2330 now each have their own history, command character, and other
2331 characteristics. (thanks to David Lichteblau)
2332 * fixed some bugs revealed by Paul Dietz' test suite:
2333 ** compiler failure in compiling LOGAND expressions including a
2335 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2336 shift greater than 32.
2337 ** FLUSH-DEST did not mark blocks for type check regeneration.
2338 ** HANDLER-CASE failed to accept declarations in handler clauses
2339 in some circumstances.
2341 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2342 * New code in contrib/sb-introspect (still probably not entirely
2343 stable yet) provides some support for smart Lisp development
2344 environments like SLIME.
2345 * The conditions signalled for errors occurring when loading .fasl
2346 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2347 in a way which should help ASDF recover gracefully.
2348 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2349 stuff has been cleaned up. If you code contrib/ stuff, this might
2350 affect you, and you can look at contrib/README, contrib/STANDARDS,
2351 and/or the 0.8.4.27 diff to check.
2352 * In full calls the compiler now does not generate checks for declared
2353 argument types for all arguments.
2354 * various threading fixes
2355 ** and some experimental patches which didn't make it into
2356 the main tree for this release, but which are shipped in
2357 contrib/experimental-thread.patch as a possible fix for some
2358 failures (deadlock, spinning...) in GC-intensive multithreaded
2360 * fixed PPC build problem (source code incompatibility of different
2361 library versions): added offsetof() hackery which attempts to divine
2362 where glibc maintainers put uc_mcontext today
2363 * fixed bug 282: compiler does not trust type assertions while passing
2364 arguments to a full call.
2365 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2366 &OPTIONAL VALUES type parameter.
2367 * fix bug 214: algorithm for noting rejected templates is now more
2368 similar to that of template seletion. (also reported by rydis on
2370 * fixed bug 141b: printing backquoted information readably and prettily
2371 inserts a space where necessary.
2372 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2373 (CHARACTER 10) are now reported as errors, rather than propagated
2374 as unknown types. (reported by piso on #lisp)
2375 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2376 with values NIL and :ERROR. (thanks to Milan Zamazal)
2377 * fixed bug 191c: CLOS now does proper keyword argument checking as
2378 described in CLHS 7.6.5 and 7.6.5.1.
2379 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2380 longer raise an error; further, using a list as a for-as-arithmetic
2381 counter now raises a meaningful error.
2382 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2383 now understood by sequence creation functions such as MAKE-SEQUENCE
2385 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2386 a STRING-INPUT-STREAM.
2387 * compiler enhancement: SIGNUM is now better able to derive the type
2389 * type declarations inside WITH-SLOTS are checked. (reported by
2391 * fixed some bugs revealed by Paul Dietz' test suite:
2392 ** incorrect optimization of TRUNCATE for a positive first
2393 argument and negative second.
2394 ** compiler failure in let-convertion during flushing dead code.
2395 ** compiler failure while deriving type of TRUNCATE on an
2396 interval, containing 0.
2397 ** ASH of a negative bignum by a negative bignum count now returns
2399 ** intersection of CONS types now canonicalizes properly, fixing
2400 inconsistencies in SUBTYPEP.
2402 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2403 * incompatible change: The --disable-debugger command line
2404 option now clobbers the debugger at a more fundamental
2405 level, by redefining #'INVOKE-DEBUGGER instead of by
2406 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2407 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2408 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2409 enter the debugger REPL and then suffer recursive errors
2410 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2411 while under the new system, BREAK in --disable-debugger mode
2412 terminates the system just as an unhandled error would.
2413 * fixed compiler performance when processing loops with a step >1;
2414 * bug fix: DOCUMENTATION now retrieves generic function
2415 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2416 support has been systematized, and now supports the methods
2417 specified by ANSI, along with a default method and a method for
2418 slot documentation. (reported by Nathan Froyd)
2419 * bug fix: effective methods associated with a generic function are
2420 no longer cached over a change of that generic function's method
2421 combination. (reported by Andreas Fuchs)
2422 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2423 names a non-existent directory. (thanks to Andreas Fuchs)
2424 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2425 the PPC platform, lead to stack corruption; this has been fixed.
2426 (reported by Rainer Joswig)
2427 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2428 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2429 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2430 arithmetic in general, is now much more reliable on the PPC
2432 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2433 platform now returns the right answer.
2434 * optimization: restored some effective method precomputation in
2435 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2436 precomputation is now tunable.
2437 * optimization: compiler-internal data structure use has been
2438 reviewed, and changes have been made that should improve the
2439 performance of the compiler by about 20%.
2440 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2441 simple-base-strings and simple-bit-vectors is improved.
2442 * optimization: the optimization of 32-bit logical and arithmetic
2443 functions introduced in version 0.8.3 on the x86 has been
2444 implemented on the mips, ppc and sparc platforms; an
2445 implementation of the same facility, but for 64-bit arithmetic,
2446 has been added for the alpha.
2447 * microoptimization: the compiler is better able to make use of the
2448 x86 LEA instruction for multiplication by constants.
2449 * bug fix: in some situations compiler did not report usage of
2450 generic arithmetic in (SPEED 3) policy.
2451 * bug 145b fix: compiler used wrong type specifier while converting
2452 MEMBER-types to numeric.
2453 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2455 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2456 index argument now works properly on the Alpha platform.
2457 * bug fix: floating point exception treatment on the Alpha platform
2459 * bug fix: FILE-POSITION works much better on string input and
2460 output streams. (thanks to Nikodemus Siivola)
2461 * bug fix: many threading/garbage collection symptoms sorted.
2462 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2463 might be pseudo-atomic.
2464 * internal change: Stopping for GC is now done with signals not
2465 ptrace. GC is now done in whichever thread wanted it, instead of
2467 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2469 * bug fix: result form in DO is not contained in the implicit
2471 * incompatible change: ICR structure is changed; the value part of
2472 CONTINUATION is now called LVAR; corresponding functions are
2473 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2474 * added type deriver for ISQRT (thanks to Robert E. Brown).
2475 * bug fix: better support for loading from the command line when an
2476 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2478 * fixed some bugs revealed by Paul Dietz' test suite:
2479 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2480 ** ROUND and FROUND now give the right answer when given very
2481 small float arguments.
2482 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2484 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2485 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2486 ** LCM should return a non-negative integer.
2487 ** PARSE-INTEGER returned the index of a terminator instead of the
2488 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2489 ** PARSE-INTEGER returned an incorrect index being applied to a
2491 ** LCM with two arguments of 0 returns 0 rather than signalling
2493 ** unsigned addition of a 32-bit constant with the high bit set no
2494 longer causes an internal compiler error on the x86.
2495 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2496 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2497 with negative last argument.
2498 ** byte specifiers with zero size and position no longer cause
2499 an error during type derivation.
2500 ** bignum multiplication on the Alpha platform now returns the
2502 * porting: The system now builds on SuSE AMD64, although it still
2503 generates a 32-bit binary.
2504 * .fasl file incompatibility: The fasl file version number has
2505 been incremented (because of the changes to internal compiler
2506 data structures referred to above).
2508 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2509 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2510 more accurately, on the Darwin kernel running on PowerPC hardware.
2511 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2512 * Compiler code deletion notes now signal a condition of type
2513 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2514 an associated MUFFLE-WARNING restart.
2515 * The compiler now performs limited argument count validation of
2516 constant format strings in FORMAT, and where appropriate in ERROR,
2517 CERROR and WARN. (thanks to Gerd Moellmann)
2518 * New ASDF-INSTALL contrib can be used for automatic download and
2519 installation of third-party Lisp code from CCLAN or other sites
2521 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2522 which forces another thread to execute a function supplied by the
2524 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2525 now accept and act upon their :ELEMENT-TYPE keyword argument.
2526 (reported by Martin Atzmueller, Edi Weitz)
2527 * bug fix: FILE-POSITION now accepts position designators up to
2528 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2529 is the greater. (thanks to Patrik Nordebo)
2530 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2531 Kalle Olavi Niemitalo)
2532 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2533 printable. (reported by Eric Marsden)
2534 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2535 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2536 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2537 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2538 (thanks to Stig Sandoe)
2539 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2540 UNREAD-CHAR now correctly includes the unread character in the
2541 target sequence. (thanks to Gerd Moellmann)
2542 * bug fix in threaded builds: the system can now be suspended and
2543 resumed by shell job control with minimal disruption.
2544 * bug fixes in times and timezones >2038AD
2545 * better handling of "where is GNU make?" problem in build scripts
2546 (thanks to Nikodemus Siivola)
2547 * new optimization: inside a named function any reference to a
2548 function with the same name is considered to be a self-reference;
2549 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2550 optimization quality.
2551 * new optimization on x86: logical functions and + now have
2552 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2553 used when the result is truncated to 32 bits.
2554 * VALUES declaration is partially enabled.
2555 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2556 * fixed some bugs revealed by Paul Dietz' test suite:
2557 ** The system now obeys the constraint imposed by
2558 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2559 types form a lattice under type intersection.
2560 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2561 ** ASSOC now ignores NIL elements in an alist.
2562 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2563 and (1+ MOST-POSITIVE-FIXNUM) answers.
2564 ** The addition of a method with invalid qualifiers to a generic
2565 function does not cause an error to be signalled immediately;
2566 a warning is signalled, and the error is generated only on
2567 calling the generic function.
2568 * changed .fasl file version number, in part to add type codes for
2569 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2570 obscure ANSI requirements
2572 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2573 * fixed bug 148: failure to inline-expand a local function left
2574 garbage, confusing the compiler.
2575 * fixed bugs 3cd: structure slot readers perform type check if the
2576 slot can have an invalid value (i.e. it is either not initialized
2577 or can be written with a less specific slot writer).
2578 * bug fix: the compiler now traps array references to elements off
2579 the end of an array; previously, the bounds checking in some
2580 circumstances could go off-by-one.
2581 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2583 * type declarations for array element types now obey the description
2584 on the CLHS page "Declaration TYPE", as per discussions on
2585 sbcl-help around 2003-05-08. This means that a declaration
2586 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2587 declaration, all references to BAR will be asserted or assumed
2588 (with THE, so dependent on compiler policy) to involve objects of
2589 type FOO. Note that no such declaration is implied in
2590 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2591 * declared types of functions from the "Conditions"
2592 chapter. (reported by Paul Dietz)
2593 * bug fix: CERROR accepts a function as its first argument.
2594 * bug fix: NTH an NTHCDR accept a bignum as index
2595 arguments. (reported by Adam Warner)
2596 * optimization: character compare routines now optimize comparing
2597 against a constant character. (reported by Gilbert Baumann)
2598 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2599 argument now works properly on the MIPS platform.
2600 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2601 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2602 declared to have a functional type, does not cause an error
2604 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2605 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2606 * fixed bug: if last continuation of a deleted block has a
2607 destination, this destination should be deleted too. (reported by
2609 * fixed a bug in the bootstrap process: the host compiler's values
2610 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2611 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2612 test case from Patrik Nordebo)
2613 * improved the ability of the disassembler on the PPC platform to
2614 provide helpful disassembly notes.
2615 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2616 the class in more cases than previously.
2617 * bug fix: FILE-POSITION now understands :START and :END for
2618 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2619 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2620 8) data. (thanks to David Lichteblau)
2621 * bug fix: it is possible to add a method to a generic function
2622 without lambda list.
2623 * bug fix: reader failed to signal END-OF-FILE inside an
2624 object representation. (reported by Nikodemus Siivola)
2625 * fixed some bugs revealed by Paul Dietz' test suite:
2626 ** LAST and [N]BUTLAST should accept a bignum.
2627 ** condition slot accessors are methods.
2628 ** (VECTOR NIL) is a subtype of STRING.
2630 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2631 * minor incompatible change: some nonsensical specialized lambda
2632 lists (used in DEFMETHOD) which were previously ignored now signal
2634 * minor incompatible change: the system is now aware of the types of
2635 variables in the COMMON-LISP package, and will signal errors for
2636 most violations of these type constraints (where previously they
2637 were silently accepted).
2638 * minor incompatible change: COMPILE-FILE now uses the freedom
2639 afforded (ANSI 3.2.2.3) to use derived function types for
2640 functions defined in the same file. This also permits the system
2641 to warn on static type mismatches and function
2642 redefinition. (Currently it does not work with high DEBUG level.)
2643 * minor incompatible change: VALUES declaration is disabled.
2644 * When issuing notes, the compiler now signals a condition of type
2645 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2646 restart for use in user handlers. It is expected that the
2647 COMPILER-NOTE condition will eventually become a condition
2648 supertype to a hierarchy of note types, which will then be
2649 handleable in a similar fashion. However, at the moment, no such
2650 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2651 but it's an implementation detail, not a classification for the
2653 * Changes in type checking closed the following bugs:
2654 ** type checking of unused values (192b, 194d, 203);
2655 ** template selection based on unsafe type assertions (192c, 236);
2656 ** type checking in branches (194bc).
2657 * A short form of VALUES type specifier has ANSI meaning (it has
2658 increased the number of situations when SBCL cannot perform type
2660 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2661 keywords or constants is permissible.
2662 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2663 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2664 classes. (thanks to Antonio Martinez)
2665 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2666 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2667 argument to a pretty-print stream. Instead, it leaves any such
2668 operations to DESCRIBE-OBJECT methods.
2669 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2671 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2672 lambda lists. (thanks to David Lichteblau)
2673 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2674 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2675 (thanks to Gerd Moellmann)
2676 * fixed bug 246: increased compilation speed of long
2677 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2679 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2681 * DEFCONSTANT now throws a condition of type
2682 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2683 constant to a non-EQL value; CONTINUE and ABORT restarts
2684 respectively change and preserve the value.
2685 * fixed bug 63: The code walker, part of the implementation of CLOS,
2686 is now better at handling symbol macros.
2687 * bug fix: There is no longer an internal implementation type named
2688 CL:LENGTH. (reported by Raymond Toy)
2689 * bug fix: In macro-like defining macros/special operators the
2690 implicit block does not enclose the lambda list.
2691 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2692 enclosed in parentheses) are not suitable as type specifiers, and
2693 their use properly signals an error now.
2694 * bug fix: An argument count mismatch for a type specifier in code
2695 being compiled no longer causes an unhandled error at compile
2696 time, but signals a compile-time warning.
2697 * fixed simple vector readable printing
2698 * bug fix: DESCRIBE takes more care over whether the class
2699 precedence list slot of a class is bound before accessing it.
2700 (reported by Markus Krummenacker)
2701 * bug fix: FORMATTER can successfully compile pretty-printer format
2702 strings which use variants of the ~* directive inside.
2703 * bug fix: SEARCH now applies its TEST predicate to the elements of
2704 the arguments in the correct order. (thanks to Wolfhard Buss)
2705 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2706 in type inference. (thanks to Robert E. Brown)
2707 * bug fix: Reading in symbols with an explicit package name of ""
2708 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2709 * fixed some bugs revealed by Paul Dietz' test suite:
2710 ** NIL is now allowed as a structure slot name.
2711 ** Arbitrary numbers, not just REALs, are allowed in certain
2712 circumstances in LOOP for-as-arithmetic clauses.
2713 ** Multiple class redefinitions before slot access no longer
2714 causes a type error.
2715 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2716 association between the name and a class.
2717 ** Generic functions with non-standard method-combination and over
2718 five methods all of which return constants no longer return NIL
2719 after the first few invocations. (thanks to Gerd Moellmann)
2720 ** CALL-NEXT-METHOD with no arguments now passes the original
2721 values of the arguments, even in the presence of assignment.
2722 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2724 ** DISASSEMBLE works with closures and funcallable instances.
2725 ** ADD-METHOD now returns the generic function, not the new method.
2726 ** FIND-METHOD signals an error if the lengths of the specializers
2727 is incompatible with the generic function, even if the ERRORP
2729 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2730 which its argument is a member.
2731 ** DEFCLASS only redefines the class named by its class-name
2732 argument if that name is the proper name of the class;
2733 otherwise, it creates a new class.
2734 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2735 of the UNBOUND-SLOT condition to the name of the slot.
2736 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2737 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2738 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2739 SLOT-MAKUNBOUND in the specified fashion.
2741 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2742 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2743 cross-compilation host. As a consequence, we can now bootstrap our
2744 way up to SBCL starting with a bare gcc toolchain and human-readable
2745 source code (first the source to CLISP, then the source to SBCL).
2746 * A contributed module containing a partial implementation of the
2747 simple-streams interface has been included. (thanks to Rudi
2749 * A contributed module implementing the RFC1321 Message Digest
2750 Algorithm, known as MD5, has been included.
2751 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2752 no longer has any effect, as the code controlled by this feature
2753 has been deleted. (As far as we know, no-one has ever built using
2754 this feature, and its semantics were confused in any case).
2755 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2756 work on conditions (as required by the ANSI specification),
2757 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2758 expected behaviour on conditions. Users should note, however,
2759 that such behaviour is not required by the ANSI specification,
2760 and so use of this behaviour may render their code unportable.
2761 * fixed some bugs revealed by Paul Dietz' test suite:
2762 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2764 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2765 ** SXHASH on condition objects no longer returns NIL.
2766 ** :ALLOCATION :CLASS slots are better treated; their values are
2767 updated on class redefinition, and initforms inherited from
2768 superclasses are applied.
2769 ** REMOVE-METHOD returns its generic function argument even when
2770 no method was removed.
2771 ** SHARED-INITIALIZE now initializes the values of the requested
2772 slots, including those with :ALLOCATION :CLASS.
2773 ** ALLOCATE-INSTANCE now works on structure classes defined via
2774 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2776 ** SLOT-EXISTS-P now works on conditions, as well as structures
2778 ** MAKE-LOAD-FORM now has the required methods on
2779 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2780 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2781 keyword, and now implements the SLOT-NAMES argument.
2782 ** methods with &OPTIONAL arguments no longer allow too many
2783 arguments to be passed in the call without error.
2784 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2785 option is consistent with the required arguments of the generic
2786 function lambda list.
2787 * bug fix: REQUIRE accepts a string designator. (Thanks to
2789 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2790 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2791 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2792 * bug fix: The compiler now checks for duplicated variables in macro
2794 * bug fix: SETQ on globals returns the correct value.
2795 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2796 not a COMPILER-ERROR (followed by some other strange error on
2797 choosing the CONTINUE restart).
2798 * bug fix: make.sh and friends are now more consistent in the way that
2799 they look for GNU "make".
2801 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2802 * experimental native threads support (on x86 Linux >=2.4 only).
2803 This is not compiled in by default: you need to add :SB-THREAD to
2804 the target features. See the "Beyond ANSI" chapter of the manual
2806 * fix for longstanding nonANSIism: The old distinction between
2807 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2808 The return value from CL:FIND-CLASS is now a CLOS class, and
2809 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2810 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2811 * An interface to the MetaObject Protocol, as described in Kiczales,
2812 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2813 MIT Press, 1991, is available from the SB-MOP package.
2814 * incompatible change: the SB-PCL package should now be considered
2815 a private implementation detail, and no longer a semi-private MOP
2817 * minor incompatible change: due to rearrangement for threads, the
2818 control stack and binding stack are now allocated at arbitrary
2819 addresses instead of being hardcoded per-port. Users affected by
2820 this probably have to be doing advanced things with shared
2821 libraries, and will know who they are.
2822 * minor incompatible change: Previously, all --eval forms used were
2823 processed with READ before any of them were processed with EVAL.
2824 Now each --eval form is processed with both READ and EVAL before
2825 the next --eval form is processed. (Thus package operations like
2826 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2827 work as the user might reasonably expect.)
2828 * minor incompatible change: *STANDARD-INPUT* is now only an
2829 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2831 * minor incompatible change: Y-OR-N-P is now character-oriented,
2832 not line oriented. Also, YES-OR-NO-P now works without errors.
2833 (thanks to Antonio Martinez)
2834 * sb-aclrepl module improvements: an integrated inspector, added
2835 repl features, and a bug fix to :trace command.
2836 * Known functions, which cannot be open coded by the backend, are
2837 considered to be able to check types of their arguments. (fixing
2838 a bug report by Nathan J. Froyd)
2839 * fixed a bug in computing method discriminating functions: It is
2840 now possible to define methods specialized on classes which have
2841 forward-referenced superclasses. (thanks to Gerd Moellmann)
2842 * fixed evaluation order in optional entries (reported by Gilbert
2844 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2845 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2846 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2847 required-by-AMOP NAME argument, as well as CLASS and
2848 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2849 * fixed bug 20: DEFMETHOD can define methods using names that are
2850 not the proper names of classes to designate class specializers.
2851 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2852 Unix file descriptors, instead of blowing up. (thanks to Antonio
2854 * Garbage collection refactoring: user-visible change is that a
2855 call to the GC function during WITHOUT-GCING will not do garbage
2856 collection until the end of the WITHOUT-GCING. If you were doing
2857 this you were probably losing anyway.
2858 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2859 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2860 TYPEP the latter but not the former.
2861 * The compiler issues a full WARNING for calls to undefined functions
2862 with names from the CL package.
2863 * MAP-INTO for a vector destination is open coded. (reported by
2864 Brian Downing on c.l.l)
2865 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2866 documentation string.
2867 * fixed some bugs revealed by Paul Dietz' test suite:
2868 ** COPY-ALIST now signals an error if its argument is a dotted
2870 ** Condition slots are now accessed more correctly in the presence
2871 of multiple initargs for a given slot.
2872 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2873 exclude restarts of the same name associated with a different
2875 ** DEFCLASS of forward-referenced classes with another
2876 forward-referenced class in the superclasses list no longer
2878 ** Condition slots are now initialized once each, not multiple
2879 times. (thanks to Gerd Moellmann)
2880 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2882 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2883 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2884 arguments contain duplicated elements.
2885 ** RESTART-CASE understands local macros.
2886 ** RESTART-CASE associates exactly its own restarts with a condition.
2887 ** ENDP in safe mode checks its argument to be of type LIST.
2888 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2889 in question is unbound.
2890 ** Optimized MAKE-INSTANCE functions no longer cause internal
2891 assertion failures in the presence of duplicate initargs.
2892 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2893 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2894 propagated even in the ctor optimized implementation of
2896 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2898 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2899 * a better implementation of SXHASH on (simple) bit vectors,
2900 measured both in execution speed and in distribution of results
2901 over the positive fixnums, has been installed. Likewise, a better
2902 implementation of EQUAL for simple bit vectors is now available.
2903 * fixed CEILING optimization for a divisor of form 2^k.
2904 * fixed bug 240 (emitting extra style warnings "using the lexical
2905 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2906 by Antonio Martinez)
2907 * fixed SXHASH, giving different results for NIL depending on type
2908 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2909 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2910 variable. (found by Rolf Wester)
2911 * SBCL does not ignore type declarations for special
2912 variables. (reported by rif on c.l.l 2003-03-05)
2913 * some bug fixes in contrib/sb-aclrepl/
2914 * fixed some bugs revealed by Paul Dietz' test suite:
2915 ** a bug in the CONS type specifier, whereby the CAR and CDR
2916 types got intertwined, has been fixed;
2917 ** the type system is now able to reason about the interaction
2918 between INTEGER and RATIO types more completely;
2919 ** APPEND, [N]REVERSE and NRECONC check that those their
2920 arguments, which must be proper lists, are really so;
2921 ** An array specialized to be unable to hold elements has been
2922 implemented, as required -- yes, really -- by ANSI;
2923 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2924 on malformed property lists;
2926 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2927 * incompatible packaging change: in line with Unix convention,
2928 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2929 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2930 * REQUIRE and PROVIDE are now optionally capable of doing something
2931 useful. See the documentation string for REQUIRE.
2932 * infrastructure for a managed SBCL contrib system: contributed
2933 modules in this release include:
2934 ** the ASDF system definition facility;
2935 ** an interface to the BSD Sockets API;
2936 ** an ACL-like convenience interface to the repl;
2937 (thanks to Kevin Rosenberg)
2938 ** an implementation of ROTATE-BYTE, with efficient implementation
2940 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2941 gives a closer approximation to the right answer than previously.
2942 (thanks to Raymond Toy)
2943 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2944 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2945 optional environment arguments, as required by ANSI.
2946 * fixed bugs in other functions taking environment objects, allowing
2947 calls with an explicit NIL environment argument to be compiled
2949 * fixed bug 228: primary return values from
2950 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2951 COMPILE or FUNCTION.
2952 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2953 :TYPE will no longer signal errors on innocuous objects.
2954 * fixed bug 231b: SETQ is better at respecting type declarations in
2955 the lexical environment.
2956 * fixed a bug in DEFCLASS: classes named by symbols with no or
2957 unprintable packages can now be defined.
2958 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2959 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2960 * fixed compiler failure related to checking types of functions.
2961 (reported by Robert E. Brown)
2962 * the compiler is now much more consistent in its error-checking
2963 treatment of bounding index arguments to sequence functions: in
2964 (SAFETY 3) code, errors will be signalled in almost all cases if
2965 invalid sequence bounding indices are passed to functions defined
2966 by ANSI to operate on sequences.
2967 * fixed a bug in the build procedure: documentation of SBCL-specific
2968 packages is now preserved and available in the final Lisp image.
2969 * lifted FDEFINITION lookup out of loops in the implementation of
2970 many list operations. (thanks to Robert E. Brown)
2971 * fixed a bug in the reader: the #n# reader macro now works for
2972 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2973 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2974 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2975 E. Brown and Tony Martinez respectively)
2976 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2977 SLOT-BOUNDP: in particular, it is now able to optimize them much
2978 better, and is now not vulnerable to having packages renamed.
2979 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2980 always called when a slot is not present in an instance. (thanks
2982 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2983 longer signal an error on unreading a character following EOF on
2984 the previous constituent stream. (thanks to Tony Martinez)
2985 * fixed some bugs revealed by Paul Dietz' test suite:
2986 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2987 not just nonnegative fixnums;
2988 ** the logical bit-array operators such as BIT-AND now accept an
2989 explicit NIL for their "opt-arg" argument (to indicate a
2990 freshly-consed result bit-array);
2991 ** ELT now signals an error on an invalid sequence index in safe
2993 ** the type system is now cleverer about negations of numeric
2994 types, and consequently understands the BIGNUM and RATIO types
2996 ** the type system is now cleverer about the interaction between
2997 INTEGER and RATIO types: while bugs still remain, many more
2998 cases are accurately computed;
2999 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3000 if it is in the last clause;
3001 ** CONSTANTLY now correctly returns a side-effect-free function in
3003 ** DECLARE is no longer treated as a special-operator; in
3004 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3005 * incremented fasl file version number due to the change in the
3006 DEFSTRUCT-SLOT-DESCRIPTION structure.
3008 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3009 * minor incompatible change: code processed by the "interpreter" or
3010 EVAL now has a compilation optimization policy of (DEBUG 2)
3011 (changed from (DEBUG 1)) to improve debuggability of interactive
3012 development, and to allow the use of the debug RETURN command in
3014 * an experimental implementation of the RETURN command for the
3015 debugger has been included. (thanks to Frederik Kuivinen)
3016 * fixed bug 62: constraints were not propagated into a loop.
3017 * fixed bug in embedded calls of SORT (reported and investigated by
3019 * fixed some bugs revealed by Paul Dietz' test suite:
3020 ** printing and reading of arrays with some dimensions having
3021 length 0 (thanks to Gerd Moellmann);
3022 ** BOA constructor with &AUX argument without a default value does
3023 not cause a type error;
3024 ** CONSTANTP now returns true for all self-evaluating objects.
3026 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3027 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3028 accessors that are related by inheritance, as specified in the
3029 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3030 Valtteri Vuorikoski)
3031 * The compiler is now able to inline functions that were defined in
3032 a complex lexical environment (e.g. inside a MACROLET).
3033 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3035 * Support for the upcoming FreeBSD-5.0 release has been included.
3036 (thanks to Dag-Erling Smorgrav)
3037 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3038 effect when it is not in a toplevel context.
3039 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3040 stand a better chance of being correct. (thanks to Gerd
3042 * fixed bug in COERCE, which now signals an error on coercing a
3043 rational to a bounded real type which excludes the expected
3045 * The compiler is now able to derive types more accurately from the
3046 COERCE and COMPILE functions.
3047 * fixed bug 223: functional binding is considered to be constant
3048 only for symbols in the CL package.
3049 * fixed bug 231: SETQ did not check the type of a variable being set
3050 (reported by Robert E. Brown)
3051 * A new optimization for MAKE-INSTANCE has been included, fixing
3052 various bugs (including relating to :ALLOCATION :CLASS slots and
3053 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3055 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3056 ** As required by ANSI, LOOP now disallows anonymous collection
3057 clauses such as COLLECT I in conjunction with aggregate boolean
3058 clauses such as THEREIS (= I 1);
3059 ** LOOP now signals an error when any variable is reused in the
3060 same loop (including the potentially useful construct analogous
3061 to WITH A = 1 WITH A = (1+ A);
3062 ** IT is only a special loop symbol within the first clause of a
3063 conditional loop clause;
3064 ** LOOP with a typed iteration variable over a hashtable now
3065 signals a type error iff it should.
3066 * fixed some other bugs revealed by Paul Dietz' test suite:
3067 ** FILE-STREAM now names the class previously known as FD-STREAM;
3068 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3069 argument) no longer signals an error;
3070 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3071 of the slot symbol, rather than using the current package
3072 ((:CONC-NAME "") continues to intern the slot's name in the
3074 * incremented fasl file version number, because of the incompatible
3075 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3076 of the new implementation of DEFINE-COMPILER-MACRO.
3078 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3079 * Support for building SBCL for MIPS platforms running in
3080 little-endian mode has now been checked in, and basic
3081 functionality on said platforms verified.
3082 * minor incompatible change: PCL now records the pathname of a file
3083 in which methods and the like are defined, rather than its
3085 * minor incompatible change: TRUENAME now considers the truename of
3086 a file naming a directory to be the pathname with :DIRECTORY
3087 component indicating that directory.
3088 * minor incompatible change: a NAMED clause in the extended form of
3089 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3090 reason for the previous behaviour is unclear.
3091 * more systematization and improvement of CLOS and MOP conformance
3092 in PCL (thanks to Gerd Moellman and Pierre Mai):
3093 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3095 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3096 primary methods with no specializers;
3097 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3099 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3100 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3101 FORWARD-REFERENCED-CLASSes; error reporting on
3102 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3104 ** SXHASH on CLOS instances now uses a slot internal to the
3105 instance to return different numbers on distinct instances,
3106 while preserving the same return value through invocations of
3108 ** DEFMETHOD signals errors when methods with longer incongruent
3109 lambda lists are added to generic functions;
3110 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3111 CLASS, as specified in AMOP;
3112 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3113 on the order returned by the primary method for classes of
3114 class STANDARD-CLASS;
3115 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3116 * fixed some bugs shown by Paul Dietz' test suite:
3117 ** DOLIST puts its body in TAGBODY;
3118 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3120 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3121 value producing form;
3122 ** if more variables are given to PROGV than values, extra
3123 variables are bound and made to have no value;
3124 ** NSUBSTITUTE on list arguments gets the right answer with
3126 ** ELT signals an error of type TYPE-ERROR when the index argument
3127 is not a valid sequence index;
3128 ** LOOP signals (at macroexpansion time) an error of type
3129 PROGRAM-ERROR when duplicate variable names are found;
3130 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3131 ** FUNCALL of special-operators now cause an error of type
3133 ** PSETQ now works as required in the presence of side-effecting
3134 symbol-macro places;
3135 ** NCONC accepts any object as its last argument;
3136 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3138 ** loop-for-as-package does not require a package to be explicitely
3140 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3142 * fixed bug 166: compiler preserves "there is a way to go"
3143 invariant when deleting code.
3144 * fixed bug 172: macro lambda lists with required arguments after
3145 &REST arguments now cause an error to be signalled. (thanks to
3147 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3148 bug. (thanks to Matthew Danish)
3149 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3151 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3152 arguments are lexically rebound. (thanks to Gerd Moellmann and
3154 * fixed bug 194: error messages are now more informative when there
3155 is no primary method applicable in a call to a generic
3156 function. (thanks to Gerd Moellmann)
3157 * fixed bug in command line argument checking (thanks to Julian
3159 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3160 (thanks to Matthew Danish)
3161 * incremented fasl file version number, because of the
3162 SXHASH-related changes in the layout of CLOS data structures
3164 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3165 * minor incompatible change: The runtime (the Unix executable named
3166 "sbcl") is now much pickier about the .core files it will load.
3167 Essentially it now requires .core files to descend from the same
3168 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3169 as the runtime does. (The intent is to prevent the crashes which
3170 can occur, and which can even be reported as mysterious failures,
3171 when people patch the sources or change the build parameters
3172 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3173 sbcl and .core files.)
3174 * fixed bug: VALUES-LIST is no longer optimized away.
3175 * fixed bug 142: The FFI conversion of C string values to Lisp
3176 string values no longer conses excessively. (thanks to Nathan
3177 Froyd porting Raymond Toy's fix to CMU CL)
3178 * began to systematize and improve MOP conformance in PCL (thanks to
3179 Nathan Froyd, Gerd Moellman and Pierre Mai):
3180 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3182 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3183 ** FINALIZE-INHERITANCE is now called on class finalization;
3184 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3185 argument precedence order.
3186 * fixed bug 202: The compiler no longer fails on functions whose
3187 derived types contradict their declared type.
3188 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3189 so it can be non-toplevel.
3190 * The fasl file version number has changed (because of the new
3191 implementation of DEFMACRO).
3192 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3193 safe code, that any length requirement by their type-specifier
3194 argument is valid. The exceptions to this are described in bug
3196 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3197 functions now check the types of their inputs as required by ANSI.
3198 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3199 introduced symbol is DECLAREd to be SPECIAL.
3200 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3201 * fixed bug: :COUNT argument to sequence functions may be negative
3202 * fixed bug: body of DO-SYMBOLS may contain declarations
3203 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3204 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3205 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3206 symbol macro only once
3207 * fixed printing of call frame when argument list is unavailable
3208 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3209 * compiler no longer signals WARNING on unknown keyword
3212 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3213 * A beta-quality port to the mips architecture running Linux,
3214 based on the old CMUCL backend, has been made. It has been tested
3215 on a big-endian kernel, and works sufficiently well to be able to
3216 rebuild itself; it has not been tested in little-endian mode.
3217 * fixed an inconsistency between gencgc.c and purify.c which made
3218 dumping/loading .core files unreliable
3219 * fixed bug 120a: The compiler now deals correctly with IFs where
3220 the consequent is the same as the alternative, instead of
3221 misderiving the return type. (thanks to Alexey Dejneka)
3222 * fixed bug 113: Logical pathnames are now dumpable (the logical
3223 host is resolved at load-time, throwing an error if it is not
3225 * fixed bug 174: FORMAT's error message is slightly clearer when a
3226 non-printing character is used in a format directive.
3227 * fixed several bugs in compiler checking of type declarations, i.e.
3228 violations of the Python "declarations are assertions" principle
3229 (thanks to Alexey Dejneka)
3230 * fixed several bugs in PCL's error checking (thanks to Gerd
3232 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3234 * fixed bug in compilation of functions as first class values
3235 (thanks to Antonio Martinez-Shotton)
3236 * The compiler's handling TYPE-ERRORs which it can prove will
3237 inevitably happen at runtime has been cleaned up and corrected
3238 in several ways. (thanks to Alexey Dejneka)
3239 * improved argument type checking for various basic arithmetic
3240 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3241 lavished on them in the past that they can be compiled in many
3242 ways in different special cases
3243 * fixed bug 181: compiler checks validity of user supplied type
3245 * cleaned up code flushing in optimization: Function calls which
3246 should signal errors for safety purposes (e.g. which ANSI says
3247 should signal errors when their arguments are of incorrect type)
3248 are no longer optimized away.
3249 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3250 * incremented fasl file version number, because changes in the
3251 implementation of sequence functions like COERCE caused
3252 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3253 in old inline expansions) to become undefined. (Actually these
3254 changes were later undone, so we might very well be binary
3255 compatible with 0.7.7 after all, but leaving the version number
3256 incremented seemed like the simplest and most conservative
3259 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3260 * An alpha-quality port to the parisc architecture running Linux,
3261 based on the old CMUCL backend, has been made. This, even more so
3262 than the other backends, should be considered still a work in
3263 progress; known problems include that the Linux kernel in 64-bit
3264 mode does not propagate the correct sigcontext structure to
3265 userspace, and consequently SBCL on a parisc64 kernel will not
3267 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3268 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3269 suppressed.) Also now that the compiler is looking at declarations
3270 in the environment, it checks optimization declarations as well,
3271 and suppresses inlining when (> DEBUG SPEED).
3272 * More fixes have been made to treatment of floating point exception
3273 treatment and other Unix signals. In particular, floating point
3274 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3275 * The detection and handling of control stack exhaustion (infinite
3276 or very deeply nested recursion) has changed. Stack exhaustion
3277 detection is now done by write-protecting pages at the OS level
3278 and applies at all optimization settings; when found, a
3279 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3280 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3281 be caught using IGNORE-ERRORS.
3282 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3283 correctly when transforming recursive LABELS functions to LETs.
3284 (thanks to Alexey Dejneka)
3285 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3286 that are names of constants or global variables.
3287 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3288 alien routines with docstrings.
3289 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3290 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3292 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3293 (thanks to Eric Marsden)
3294 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3295 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3296 * Bug fix: The SPARC backend can now compile functions involving
3297 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3298 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3299 to a functions expecting a PATHNAME-DESIGNATOR.
3300 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3301 lambda lists. (thanks to Alexey Dejneka)
3302 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3303 functions so that EXPORTing the name of the class doesn't cause
3304 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3305 to Antonio Martinez for reporting this)
3306 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3307 constant-folded without failing an assertion. (thanks to Einar
3308 Floystad Dorum for reporting this)
3309 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3310 types are now dealt with more correctly by the compiler.
3311 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3312 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3313 behaviour analogous to RENAME-FILE. This puts its behaviour more
3314 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3316 * The fasl file version number has changed again. (because of the
3317 bug fix involving the names of PCL MAKE-INSTANCE functions)
3319 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3320 * bug fix: Floating point exceptions are treated much more
3321 consistently on the x86/Linux and PPC/Linux platforms.
3322 * Array initialization with :INITIAL-ELEMENT is now much faster for
3323 cases when the compiler cannot open code the array creation, but
3324 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3325 array accesses have also seen a speed increase.
3326 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3327 specified. (This was at the root of some bad interactions between
3328 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3330 * bug fix: Internal error arguments for undefined functions are now
3331 computed correctly on the PPC/Linux platform.
3332 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3333 Raymond Toy's patch for CMU CL)
3334 * Support for the Solaris 9 operating environment has been included
3335 (thanks to Daniel Merritt)
3336 * A very ugly but hopefully complete draft of the missing FFI chapter
3337 of the manual has been created by reformatting the corresponding
3338 CMU CL manual chapter into (currently very ugly and incoherent)
3339 DocBook and bringing it up to date for SBCL behavior. Thus, the
3340 manual is now essentially complete, at least by my extreme
3341 once-and-only-once standards, whereby it's acceptable to refer to
3342 the doc strings of SB-EXT functions as the primary documentation.
3343 * The fasl file version number has changed again, due to cleanup of
3344 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3345 is no longer a static symbol.)
3347 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3348 * SBCL now builds with OpenMCL (version 0.12) as the
3349 cross-compilation host; also, more progress has been made toward
3350 bootstrapping under CLISP.
3351 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3353 * bug 158 fixed: The compiler can now deal with integer loop
3354 increments different from 1; fixing this turned out also to fix
3356 * bug 169 fixed: no more bogus warnings about using lexical bindings
3357 despite the presence of perfectly good SPECIAL declarations (thanks
3358 to David Lichteblau)
3359 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3360 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3361 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3363 * bug 180 fixed: Method combination specifications no longer ignore
3364 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3365 * bug fix: Structure type predicate functions now check their argument
3366 count as they should.
3367 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3368 correctly. (thanks to Pierre Mai)
3369 * minor incompatible change: The --noprogrammer option is deprecated
3370 in favor of the new --disable-debugger option, which is very similar.
3371 (The major difference is that it takes effect at a slightly different
3372 time at startup, causing handling of errors in --sysinit and
3373 --userinit files will be affected differently.) The
3374 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3375 been added to allow this functionality to be controlled from ordinary
3376 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3377 maintainers, who might want to run non-interactive scripts to
3378 build SBCL cores which will later be used interactively.)
3379 * minor incompatible change: The LOAD function no longer, when given
3380 a wild pathname to load, loads all files matching that pathname.
3381 Instead, an error of type FILE-ERROR is signalled.
3383 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3384 * bug 147 fixed: The compiler preserves its block link/count
3385 invariants more correctly now so that it doesn't crash. (thanks
3387 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3389 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3390 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3391 no-value-for-this-character as NIL (as specified by ANSI).
3392 * HOST-NAMESTRING on physical pathnames now returns a string that is
3393 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3394 (thanks to Christophe Rhodes)
3395 * The Alpha port handles icache flushing more correctly. (thanks to
3397 * More progress has been made toward bootstrapping under CLISP. (thanks
3398 to Christophe Rhodes)
3399 * The fasl file format has changed again, because dynamic loading
3400 on OpenBSD (which has non-ELF object files) motivated some cleanups
3401 in the way that foreign symbols are transformed and passed around.
3402 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3403 is no longer treated as whitespace by the reader, but instead as
3404 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3405 |A
\7fB|, instead of A as it used to.
3407 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3408 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3409 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3410 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3411 this as of 0.7.1.45, but was left out of the previous news section
3412 (thanks to Dan Barlow)
3413 * SBCL now runs on the Solaris operating system on SPARC architectures
3414 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3415 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3416 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3417 * SPARC backend cleanups, allowing builds of cores optimized for V8
3418 and V9 SPARCS, and also emission of code targeted to a particular
3419 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3421 * SBCL is closer to bootstrapping under CLISP, thanks to various
3422 fixes by Christophe Rhodes.
3423 * The fasl file format has changed again, to allow the compiler's
3424 INFO database to support symbol macros.
3425 * The user manual (in doc/) is formatted into HTML more nicely.
3426 (thanks to coreythomas)
3427 * The system is smarter about SUBTYPEP relationships, especially
3428 those involving NOT types (including types such as ATOM which are
3429 represented internally using NOT types). Thus SUBTYPEP is less
3430 likely to return (VALUES NIL NIL) in general, and in particular
3431 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3433 * The fasl file format has changed again, because the internal
3434 representation of types now includes a new slot to support the new
3435 SUBTYPEP-of-NOT-types logic.
3436 * (not a change in the main branch of SBCL, but a related prototype
3437 which can hopefully be merged into the main branch of SBCL in the
3438 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3439 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3440 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3441 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3442 dispatch macro character combinations. (thanks to Alexey Dejneka)
3443 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3444 that superficially look like logical namestrings correctly.
3445 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3447 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3448 default supplied-p) for &optional and &key arguments. (thanks to
3451 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3452 * incompatible change: The compiler is now less aggressive about
3453 tail call optimization, doing it only when (> SPACE DEBUG) or
3454 (> SPEED DEBUG). (This is an incompatible change because there are
3455 programs which relied on the old CMU-CL-style behavior to optimize
3456 away their unbounded recursion which will now die of stack overflow.)
3457 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3458 for non-GENCGC systems has been increased to 20M (since that
3459 seems much closer to the likely performance optimum for modern
3460 systems than the old 4M value was)
3461 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3462 and *DEBUG-PRINT-LEVEL*
3463 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3464 of CMU CL's support for SPARC, and various endianness and other
3465 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3466 * new syntactic sugar for the Unix command line: --load foo.bar is now
3467 an alternate notation for --eval '(load "foo.bar")'.
3469 ** The system now detects stack overflow and handles it gracefully,
3470 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3471 optimization settings. (This is a good thing in general, and
3472 its introduction in this version should be particularly timely
3473 for anyone whose code fails because of suppression of tail
3475 ** The system now hunts for the C variable "environ" in a more
3476 devious way, to avoid segfaults when the C library version differs
3477 between compile time and run time. (thanks to Christophe Rhodes)
3478 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3479 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3480 ** The compiler no longer issues bogus style warnings for undefined
3481 classes in the same source file as the DEFCLASSes which defined
3482 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3484 ** fixes in CONDITION class precedence list for undefined function
3485 errors (thanks to Alexei Dejneka)
3486 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3487 correctly. (thanks to Dan Barlow)
3488 ** portability fixes aiming at bootstrapping under CLISP (thanks
3489 to Dave McDonald and Christophe Rhodes)
3490 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3491 ** fixes in type translation and and type inference (thanks to
3493 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3494 ** various fixes in the new ports (thanks to Dan Barlow)
3495 * several changes related to debugging:
3496 ** suppression of tail recursion, as noted above
3497 ** stack overflow detection, as noted above
3498 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3499 is now the default. (For some time encapsulation has been more
3500 reliable than the breakpoint-based :ENCAPSULATE NIL
3501 implementation, at least on X86 systems; and I just noticed that
3502 encapsulation also seems closer to the spirit of the ANSI
3505 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3507 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3508 up properly again. (There was a packaging bug in 0.7.0 which
3509 left their definitions in SB-SYS::LOAD-FOREIGN and
3510 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3511 vital for most things which interface to C-level interfaces,
3512 like extensions working with sockets or databases or
3513 Perl-compatible regexes or whatever, and the need to fix
3514 this bug is the main reason that 0.7.1 was released so
3516 ** DEFGENERIC is now choosier about the methods it redefines, so that
3517 reLOADing a previously-LOADed file containing DEFGENERICs does
3518 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3519 cycle now works as it should. (thanks to Alexey Dejneka)
3520 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3521 fixed by Christophe Rhodes. (He actually submitted this patch
3522 months ago, and I delayed until after 0.7.0.)
3523 ** Bug 111 (internal compiler confusion about runtime checks on
3524 FUNCTION types) was fixed by Alexey Dejneka.
3525 * Some internal cleanups (getting rid of variables which aren't
3526 needed now that the byte interpreter is gone) caused the fasl
3527 file format number to change again.
3529 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3530 * major incompatible change: The default fasl file extension, i.e. the
3531 default extension for files produced by COMPILE-FILE, has changed
3532 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3534 ** There are many changes in the implementation of the compiler.
3535 SBCL is now essentially a compiler-only implementation of ANSI
3536 Common Lisp. EVAL still "interprets" a few special cases, but
3537 almost all the interesting cases are handled by creating
3538 a LAMBDA expression, calling COMPILE on it, then calling
3539 FUNCALL on the result.
3540 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3541 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3542 Since the code is newer, there might still be some new bugs
3543 (though not as many as before Martin Atzmueller's fixes:-). But
3544 the new code is substantially simpler and clearer, and hopefully
3545 any remaining bugs will be simpler, less fundamental, and more
3546 fixable then the bugs in the old code.
3547 ** The revised compiler is still a little unsteady on its feet.
3549 *** The debugging information it produces (particularly the names
3550 of FUNCTION objects) is sometimes much less useful than what
3551 the old compiler produced.
3552 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3553 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3554 within a MACROLET) has been temporarily weakened.
3555 ** There are new compiler optimizations for various functions:
3556 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3557 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3558 *** the math functions TRUNCATE, FLOOR, and CEILING
3559 *** the function-of-all-trades COERCE
3560 Mostly these should be transparent, but there's one
3561 potentially-annoying problem (bug 117): when the compiler
3562 inline-expands a function and does type analysis on the result,
3563 it can create control paths which have type mismatches, and
3564 when it can't prove that those control paths aren't taken,
3565 it will issue WARNINGs about the type mismatches. This is
3566 a particular problem in practice for the new sequence functions.
3567 It's not clear how this should be fixed, and for now, a
3568 workaround is given in the entry for 117 in the BUGS file.
3569 ** (Because of the interaction between the two previous items --
3570 occasional inlining problems and new inline expansions -- some
3571 of the new sequence function optimizations won't really kick in
3572 completely until debugging information, and then inlining, are
3573 straightened out in some future version.)
3574 * minor incompatible changes:
3575 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3576 DIRECTORY no longer implicitly promotes NIL slots of its
3577 pathname argument to :WILD. In particular, when you ask for the
3578 contents of a directory (which you used to be able to do without
3579 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3580 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3581 ** changes in behavior that ANSI explicitly defines to be
3582 implementation dependent:
3583 *** The new compiler-only implementation still conforms with ANSI,
3584 but acts a little different than before. Besides the obvious
3585 changes in performance tradeoffs (that the cost per form passed
3586 to EVAL has gone up, and the cost per form executed by EVAL
3587 has gone down), the behavior of the system changes a little
3588 because there are no longer any interpreted function objects.
3589 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3590 e.g. doing COMPILE on the output of interactive DEFUN is
3592 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3594 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3595 has changed from (:CL) to NIL.
3596 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3597 e.g. #\Newline and #\Tab, don't have names specified in the
3598 ANSI Common Lisp standard, is now based on their ASCII symbolic
3599 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3600 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3601 are no longer used for output.
3602 ** changes in internal implementation constants:
3603 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3604 4 million. (If your application spends a lot of time GCing and
3605 you have a lot of RAM, you might want to experiment with
3606 increasing it even more.)
3607 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3608 However, almost all old code should still continue to work without
3609 immediate update, as SB-C-CALL is now a (deprecated) nickname
3611 ** Old operator names in the style DEF-FOO are now deprecated in
3612 favor of new corresponding names DEFINE-FOO, for consistency with
3613 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3614 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3615 internal symbols, but a few supported extensions like
3616 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3617 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3618 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3619 etc. as you get deeper into recursive calls to the debugger
3620 command loop, instead of the old "5]", "5]]", "5]]]"
3621 sequence. (I was motivated to do this when squabbles between
3622 ILISP and SBCL left me very deeply nested in the debugger. In the
3623 short term, this change will probably provoke more ILISP/SBCL
3624 squabbles, but hopefully it will be an improvement in the long run.)
3625 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3626 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3627 function, since declaiming return types involving aliens is
3628 (1) annoyingly messy to do by hand and (2) vital to efficient
3629 compilation of code which calls such functions.
3630 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3631 longer reexported by the SB-EXT package. They're solely useful
3632 for alien code, so it seems more logical that you should get
3633 them from the SB-ALIEN package, not in SB-EXT.
3634 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3635 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3636 features. Instead, the code that they used to control is always
3637 built into the system.
3638 * many other bug fixes
3639 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3640 advantage of the new EVAL-WHEN stuff and to clean them up in
3641 general, and they are now more ANSI-compliant in a number of
3642 ways. Martin Atzmueller is responsible for a lot of this.
3643 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3645 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3646 *** correct ERROR type for various file operations
3647 *** some fixes for Lisp streams
3648 *** DEFMETHOD syntax checking
3649 *** changing old weird representation of debug information as
3650 strings (which, among their other deficiencies, don't transform
3651 correctly when you rename packages, and don't change their
3652 print representation when you change things like *PACKAGE*
3653 and *PRINT-LENGTH*) to symbols and lists of symbols
3654 He also made several improvements and fixed several bugs in DESCRIBE.
3655 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3657 *** misbehavior of WRITE-STRING/WRITE-LINE
3658 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3659 and several other LOOP problems as well
3660 *** DIRECTORY when similar filenames are present
3661 *** DEFGENERIC with :METHOD options
3662 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3663 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3664 *** argument ordering in FIND with :TEST option
3665 *** mishandled package designator argument in APROPOS-LIST
3666 *** various problems in the backquote readmacro
3667 *** a bug in APROPOS
3668 *** probably some others that I'm not describing very well here,
3669 since the CVS log documents them by reference to sbcl-devel
3670 messages, and the SourceForge archives aren't working well.:-(
3671 ** Dan Barlow improved the Alpha port (and is making progress on the
3672 PPC port, for those of you who think different).
3673 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3674 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3675 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3676 continues to work on the SPARC port (for those of you in a position
3677 to look down upon our little PC-compatible boxes from a great height).
3678 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3679 and uses that copy, rather than the current dynamic value, when
3680 it's trying to decide whether to truncate output. Thus e.g.
3681 (let ((*print-lines* 50))
3682 (pprint-logical-block (stream nil)
3684 (let ((*print-lines* 8))
3685 (print (aref possiblybigthings i) stream)))))
3686 should now truncate the logical block only at 50 lines, instead of
3687 often truncating it at 8 lines, as it did before.
3688 * The doc/cmucl/ directory, containing old CMU CL documentation
3689 from the time of the fork, is no longer part of the base system.
3690 SourceForge has shut down its anonymous FTP service, and with it
3691 my original plan for distributing the old CMU CL documentation
3692 there. For now, if you need these files you can download an old
3693 SBCL source release and extract them from it.
3694 * The fasl file version number changed again, for dozens of reasons,
3695 some of which are apparent above.
3697 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3698 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3699 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3700 patch, so that DISASSEMBLE output is much nicer.
3701 * The code in the SB-PROFILE package now seems reasonably stable.
3702 I still haven't decided what the final interface should look like
3703 (I'd like PROFILE to interact cleanly with TRACE, since both
3704 facilities use function encapsulation) but if you have a need
3705 for profiling now, you can probably use it successfully with
3706 the current CMU-CL-style interface.
3707 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3708 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3709 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3710 link now returns the dangling link itself, and for similar
3711 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3712 link itself. (In these cases the old code signalled an error and
3713 looped endlessly, respectively.) Thus, DIRECTORY now works even
3714 in the presence of dangling and cyclic symbolic links.
3715 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3716 is now a supported extension again, since the consensus on
3717 sbcl-devel was that it can be useful for ordinary development
3718 work, not just for debugging SBCL itself.
3719 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3720 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3721 that currently-defined functions might be redefined later with
3722 different return types.
3723 * Hash tables can be printed readably, as inspired by CMU CL code
3724 of Eric Marsden and SBCL code of Martin Atzmueller.
3725 * better error handling in CLOS method combination, thanks to
3726 Martin Atzmueller porting Pierre Mai's CMU CL patches
3727 * more overflow fixes for >16Mbyte I/O buffers
3728 * A bug in READ has been fixed, so that now a single Ctrl-D
3729 character suffices to cause end-of-file on character streams.
3730 In particular, now you only need one Ctrl-D at the command
3731 line (not two) to exit SBCL.
3732 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3734 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3735 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3736 does the right thing.
3737 * The implementation of some type tests, especially for CONDITION
3738 types, is now tidier and maybe faster, due to CMU CL code
3739 originally by Douglas Crosher, ported by Martin Atzmueller.
3740 * Some math functions have been fixed, and there are new
3741 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3742 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3743 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3744 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3745 likely to be built into sbcl-0.7.0.)
3746 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3747 CL, ported to SBCL by Martin Atzmueller)
3748 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3749 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3750 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3751 * There's a new slam.sh hack to shorten the edit/compile/debug
3752 cycle for low-level changes to SBCL itself, and a new
3753 :SB-AFTER-XC-CORE target feature to control the generation of
3754 the after-xc.core file needed by slam.sh.
3755 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3756 COMPILE-FILE is no longer supported, so that now every function
3757 gets an entry point, so that block compilation looks a little
3758 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3759 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3760 returns the number of bytes consed since the system started,
3761 rather than the number consed since the first time the function
3762 was called. (The new definition parallels ANSI functions like
3763 CL:GET-INTERNAL-RUN-TIME.)
3764 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3765 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3766 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3767 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3768 * Fasl file version numbers are now independent of the target CPU,
3769 since historically most system changes which required version
3770 number changes have affected all CPUs equally. Similarly,
3771 the byte fasl file version is now equal to the ordinary
3774 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3775 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3776 is no longer recognized. I apologize for this, because it was
3777 listed in SB-EXT as a supported extension, but I found that
3778 its existing behavior was poorly specified, as well as incorrectly
3779 specified, and it looked like too much of a mess to straighten it
3780 out. I have enough on my hands trying to get ANSI stuff to work..
3781 * many patches ported from CMU CL by Martin Atzmueller, with
3782 half a dozen bug fixes in pretty-printing and the debugger, and
3783 half a dozen others elsewhere
3784 * fixed bug 13: Floating point infinities are now supported again.
3785 They might still be a little bit flaky, but thanks to bug reports
3786 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3787 as flaky as they were.
3788 * The --noprogrammer command line option is now supported. (Its
3789 behavior is slightly different in detail from what the old man
3790 page claimed it would do, but it's still appropriate under the
3791 same circumstances that the man page talks about.)
3792 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3793 are now supported, and enabled by default. Thus, the compiler can
3794 handle many floating point and complex operations much less
3795 inefficiently. (Thus e.g. you can implement a complex FFT
3797 * The compiler now detects type mismatches between DECLAIM FTYPE
3798 and DEFUN better, and implements CHECK-TYPE more correctly, and
3799 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3800 thanks to patches from Martin Atzmueller.
3801 * various fixes to make the cross-compiler more portable to
3802 ANSI-conforming-but-different cross-compilation hosts (notably
3803 Lispworks for Windows, following bug reports from Arthur Lemmens)
3804 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3805 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3806 fixed thanks to Pierre Mai's CMU CL patches.
3807 * fixes to keep the system from overflowing internal counters when
3808 it tries to use i/o buffers larger than 16M bytes
3809 * fixed bug 45a: Various internal functions required to support
3810 complex special functions have been merged from CMU CL sources.
3811 (When I was first setting up SBCL, I misunderstood a compile-time
3812 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3813 * improved support for type intersection and union, fixing bug 12
3814 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3815 more obscure bugs as well
3816 * some steps toward byte-compiling non-performance-critical
3817 parts of the system, courtesy of patches from Martin Atzmueller
3818 * Christophe Rhodes has made some debian packages of sbcl at
3819 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3820 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3821 stable, but are nonetheless usable. When he's ready, I'd be happy
3822 to add them to the SourceForge "File Releases" section. (And if
3823 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3824 * new fasl file format version number (because of changes in
3825 internal representation of (OR ..) types to accommodate the new
3826 support for (AND ..) types, among other things)
3828 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3829 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3831 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3833 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3834 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3835 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3836 * DESCRIBE now works on structure objects again.
3837 * Most function call argument type mismatches are now handled as
3838 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3839 know whether the function will be redefined before the call is
3840 executed. (The compiler could flag local calls with full WARNINGs,
3841 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3842 it doesn't keep track of enough information to know whether calls
3843 are local in this sense.)
3844 * Compiler output is now more verbose, with messages truncated
3845 later than before. (There should be some supported way for users
3846 to override the default verbosity, but I haven't decided how to
3847 provide it yet, so this behavior is still controlled by the internal
3848 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3849 src/compiler/ir1util.lisp.)
3850 * Fasl file format version numbers have increased again, because
3851 support for the Gray streams extension changes the layout of the
3852 system's STREAM objects.
3853 * The Gray subclassable streams extension now works, thanks to a
3854 patch from Martin Atzmueller.
3855 * The full LOAD-FOREIGN extension (not just the primitive
3856 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3857 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3858 but like most other programs, it defaults to copying the Unix
3859 environment from the original process instead of starting the
3860 new process in an empty environment.
3861 * Extensions which manipulate the Unix environment now support
3862 an :ENVIRONMENT keyword option which doesn't smash case or
3863 do other bad things. The CMU-CL-style :ENV option is retained
3864 for porting convenience.
3865 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3866 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3868 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3870 * A patch from Martin Atzmueller seems to have solved the SIGINT
3871 problem, and as far as we know, signal-handling now works cleanly.
3872 (If you find any new bugs, please report them!)
3873 * The system no longer defaults Lisp source file names to types
3874 ".l", ".cl", or ".lsp", but only to ".lisp".
3875 * The compiler no longer uses special default file extensions for
3876 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3877 seems to expect a single default extension for all compiled code,
3878 and there's no compelling reason to try to stretch the standard
3879 to allow two different extensions.) Instead, byte-compiled files
3880 default to the same extension as native-compiled files.
3881 * Fasl file format version numbers have increased again, because
3882 a rearrangement of internal implementation packages made some
3883 dumped symbols in old fasl files unreadable in new cores.
3884 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3885 many fewer weird special cases.
3886 * Bug #17 (differing COMPILE-FILE behavior between logical and
3887 physical pathnames) has been fixed, and some related misbehavior too,
3888 thanks to a patch from Martin Atzmueller.
3889 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3890 by Tim Moore, ported to SBCL by Martin Atzmueller.
3891 * Martin Atzmueller fixed several filesystem-related problems,
3892 including bug #36, in part by porting CMU CL patches, which were
3893 written in part by Paul Werkowski.
3894 * More compiler warnings in src/runtime/ are gone, thanks to
3895 more patches from Martin Atzmueller.
3896 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3899 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3901 * DESCRIBE now works on CONDITION objects.
3902 * The debugger now handles errors which arise when trying to print
3903 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3905 * The build system now uses an additional file, customize-target-features.lisp,
3906 to allow local modifications to the target *FEATURES* list. (The point of
3907 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3908 debugging features enabled, without having to worry about propagating it
3909 into everyone's system when I do a "cvs update".) When no
3910 customize-target-features.lisp file exists, the target *FEATURES* list
3911 should be constructed the same way as before.
3912 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3913 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3914 warn about behavior which is undefined under the ANSI spec. Note: This
3915 is specified by ANSI, but it's not very popular with programmers.
3916 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3917 macro in the SBCL sources for an example of a workaround which you
3918 might use to make portable ANSI-standard code which does what you want.
3919 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3920 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3921 when it's used as a non-toplevel form. (This is required in order
3922 to implement the DEFCONSTANT-EQX macro.)
3923 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3924 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3925 funny things, due to bugs in the implementation of EVAL-WHEN
3926 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3927 non-toplevel DEFCONSTANTs are uncommon.)
3928 * The core file version number and fasl file version number have been
3929 incremented, because the old noncompliant DEFCONSTANT behavior involved
3930 calling functions which no longer exist, and because I also took the
3931 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3932 * fixed bug 1 (error handling before read-eval-print loop starts), and
3933 redid debugger restarts and related debugger commands somewhat while
3935 ** The QUIT debugger command is gone, since it did something
3936 rather different than the SB-EXT:QUIT command, and since it never
3937 worked properly outside the main toplevel read/eval/print loop.
3938 Invoking the new TOPLEVEL restart provides the same functionality.
3939 ** The GO debugger command is also gone, since you can just invoke
3940 the CONTINUE restart directly instead.
3941 ** The TOP debugger command is also gone, since it's redundant with the
3942 FRAME 0 command, and since it interfered with abbreviations for the
3944 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3946 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3947 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3948 some porting work by Martin Atzmueller)
3949 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3950 with a boatload of patches which he ported to SBCL. Now that those
3952 ** The system tries to make sure that its low-priority messages
3953 are prefixed by semicolons, to help people who like to use
3954 syntax highlighting in their ILISP buffer. (This patch
3955 was originally due to Raymond Toy.)
3956 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3957 patches originally written by Raymond Toy.
3958 ** The compiler understands coercion between single-value and
3959 multiple-VALUES type expressions better, getting rid of some very
3960 weird behavior, thanks to patches originally by Robert MacLachlan
3961 and Douglas Crosher.
3962 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3963 lambda lists, thanks to a patch originally by Pierre Mai.
3964 ** The system no longer bogusly warns about "abbreviated type
3966 ** The compiler gets less confused by inlining and RETURN-FROM,
3967 thanks to some patches originally by Tim Moore.
3968 ** The system no longer hangs when dumping circular lists to fasl
3969 files, thanks to a patch originally from Douglas Crosher.
3970 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3971 undefined function error.
3972 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3973 in structure slot initforms) for now, documented workaround instead:-|
3974 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3975 * fixed bug 5: added stubs for various Gray stream functions called
3976 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3977 installed, at least appropriate type errors are generated
3978 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3979 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3980 consistently in DEFMETHOD forms.
3981 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3982 it doesn't seem to affect SBCL after all
3983 * The C runtime system now builds with better optimization and many
3984 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3986 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3988 * The system is now under CVS at SourceForge (instead of the
3989 CVS repository on my home machine).
3990 * The new signal handling code has been tweaked to treat register
3991 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3992 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3993 downstream, e.g. in debug-int.lisp, has implicit dependencies
3994 on the unsignedness of integer representation of machine words,
3995 and that caused the system to bomb out with infinite regress
3996 when trying to recover from type errors involving signed values,
3997 e.g. (BUTLAST '(1 2 3) -1).)
3998 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3999 (This was one of the bugs Peter Van Eynde reported back in July.)
4000 * The system now uses code inspired by Colin Walters' O(N)
4001 implementation of MAP (from the cmucl-imp@cons.org mailing
4002 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4003 inline the MAP operation, and there is more than one
4004 sequence argument to the MAP call (so that it can't just
4005 do ETYPECASE once and for all based on the type of the
4006 single sequence argument). (The old non-inline implementation
4007 of the general M-argument sequence-of-length-N case required
4008 O(M*N*N) time when any of the sequence arguments were LISTs.)
4009 * The QUIT :UNIX-CODE keyword argument has been renamed to
4010 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4011 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4013 * Raymond Wiker's patches to port dynamic loading from Linux to
4014 FreeBSD have been added.
4015 * The BUGS file is now more nearly up to date, thanks in large part
4016 to Martin Atzmueller's review of it.
4017 * The debugger now flushes standard output streams before it begins
4018 its output ("debugger invoked" and so forth).
4019 * The core version number and fasl file version number have both
4020 been incremented, because of incompatible changes in the layout
4022 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4023 used to not be called for a saved Lisp image.)
4024 * Martin Atzmueller's version of a patch to fix a compiler crash,
4025 as posted on sbcl-devel 13 September 2000, has been installed.
4026 * Instead of installing Martin Atzmueller's patch for the
4027 compiler transform for SUBSEQ, I deleted the compiler transform,
4028 and transforms for some similar consing operations.
4029 * A bug in signal handling which kept TRACE from working on OpenBSD
4031 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4032 away by constant folding
4033 * The system now defines its address space constants in one place
4034 (in the Lisp sources), and propagates them automatically elsewhere
4035 (through GENESIS and the sbcl.h file). Therefore, patching the
4036 address map is less unnecessarily tedious and error-prone. The
4037 Lisp names of address space constants have also been systematized.
4038 * CVS tags like dollar-Header-dollar have been removed from
4039 the sources, because they have never saved me trouble and
4040 they've been source of trouble working with patches and other
4041 diff-related operations.
4042 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4043 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4045 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4047 * The system has been ported to OpenBSD.
4048 * The system now compiles with a simple "sh make.sh" on the systems
4049 that it's supported on. I.e., now you no longer need to tweak
4050 text in the target-features.lisp-expr and symlinks in src/runtime/
4051 by hand, the make.sh takes care of it for you.
4052 * The system is no longer so grossly inefficient when compiling code
4053 involving vectors implemented as general (not simple) vectors (VECTOR T),
4054 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4055 which dares to use the various sequence functions on non-simple
4056 vectors, takes less of a performance hit.
4057 * There is now a primitive type predicate VECTOR-T-P
4058 to test for the (VECTOR T) type, so that e.g.
4059 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4060 can now be compiled with some semblance of efficiency. (The old code
4061 turned the type declaration into a full call to %TYPEP at runtime!)
4062 * AREF on (VECTOR T) is still not fast, since it's still compiled
4063 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4064 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4065 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4066 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4067 without making full calls to SUBTYPEP at runtime.
4068 (Some analogous efficiency issues for non-simple vectors specialized to
4069 element types other than T, or for non-simple multidimensional arrays,
4070 have not been addressed. They could almost certainly be handled the
4071 same way if anyone is motivated to do so.)
4072 * The changes in array handling break binary compatibility, so
4073 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4074 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4075 instead of (VALUES T T).
4076 * By following the instructions that Dan Barlow posted to sbcl-devel
4077 on 2 July 2000, I was able to enable primitive dynamic object
4078 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4079 functionality is not implemented (since it calls ld to resolve
4080 library references automatically, requiring RUN-PROGRAM for its
4081 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4082 try to resolve library references) is now supported.
4083 * The system now flushes the standard output streams when it terminates,
4084 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4085 them at several other probably-convenient times, e.g. in each pass of
4086 the toplevel read-eval-print loop, and after evaluating a form given
4087 as an "--eval" command-line option. (These changes were motivated by a
4088 discussion of stream flushing issues on cmucl-imp in August 2000.)
4089 * The source transform for TYPEP of array types no longer assumes
4090 that an array whose element type is a not-yet-defined type
4091 is implemented as an array of T, but instead punts, so that the
4092 type will be interpreted at runtime.
4093 * There is now some support for cross-compiling in make.sh: each of
4094 the phases of make.sh has its own script. (This should be transparent
4095 to people doing ordinary, non-cross-compile builds.)
4096 * Since my laptop doesn't have hundreds of megabytes of memory like
4097 my desktop machine, I became more motivated to do some items on
4098 my to-do list in order to reduce the size of the system a little:
4099 ** Arrange for various needed-only-at-cold-init things to be
4100 uninterned after cold init. To support this, those things have
4101 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4102 symbols with such names are now uninterned after cold init).
4103 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4104 fasl files for cold load.
4105 ** Remove the old compiler structure pooling code (which used to
4106 be conditional on the target feature :SB-ALLOC) completely.
4107 ** Redo the representation of some data in cold init to be more compact.
4108 (I also looked into supporting byte compiled code at bootstrap time,
4109 which would probably reduce the size of the system a lot, but that
4110 looked too complicated, so I punted for now.)
4111 * The maximum signal nesting depth in the src/runtime/ support code has
4112 been reduced from 4096 to 256. (I don't know any reason for the very
4113 large old value. If the new smaller value turns out to break something,
4114 I'll probably just bump it back up.)
4115 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4117 * Many, many bugs reported by Peter Van Eynde have been added to
4118 the BUGS list; some have even been fixed.
4119 * While enabling dynamic object file loading, I tried to make the
4120 code easier to understand, renaming various functions and variables
4121 with less ambiguous names, and changing some function calling
4122 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4123 * While trying to figure out how to do the OpenBSD port, I tried to
4124 clean up some of the code in src/runtime/. In particular, I dropped
4125 support for non-POSIX signal handling, added various comments,
4126 tweaked the code to reduce the number of compilation warnings, and
4127 renamed some files to increase consistency.
4128 * To support the new automatic configuration functionality in make.sh,
4129 the source file target-features.lisp-expr has been replaced with the
4130 source file base-target-features.lisp-expr and the machine-generated
4131 file local-target-features.lisp-expr.
4132 * fixed a stupid quoting error in make.sh so that using CMU CL
4133 "lisp -batch" as cross-compilation host works again
4135 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4137 * DESCRIBE no longer tries to call itself recursively to describe
4138 bound/fbound values, so that it no longer fails on symbols which are
4139 bound to themselves (like keywords, T, and NIL).
4140 * DESCRIBE now works on generic functions.
4141 * The printer now prints less-screwed-up representations of closures
4142 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4143 * A private symbol is used instead of the :EMPTY keyword previously
4144 used to mark empty slots in hash tables. Thus
4145 (DEFVAR *HT* (MAKE-HASH-TABLE))
4146 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4147 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4148 now does what ANSI says that it should. (You can still get
4149 similar noncompliant behavior if bang on the hash table
4150 implementation with all the symbols you get back from
4151 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4152 This breaks binary compatibility, since tests for equality to
4153 :EMPTY are wired into things like the macroexpansion of
4154 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4156 * There's now a minimal placeholder implementation for CL:STEP,
4157 as required by ANSI.
4158 * An obscure bug in the interaction of the normal compiler, the byte
4159 compiler, inlining, and structure predicates has been patched
4160 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4161 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4163 * Missing ordinary arguments in a macro call are now detected even
4164 when the macro lambda list contains &KEY or &REST.
4165 * The debugger no longer complains about encountering the top of the
4166 stack when you type "FRAME 0" to explicitly instruct it to go to
4167 the top of the stack. And it now prints the frame you request even
4168 if it's the current frame (instead of saying "You are here.").
4169 * As specified by ANSI, the system now always prints keywords
4170 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4172 * The default initial SIZE of HASH-TABLEs is now smaller.
4173 * Type information from CLOS class dispatch is now propagated
4174 into DEFMETHOD bodies, so that e.g.
4175 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4177 is now basically equivalent to
4178 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4179 (DECLARE (TYPE SINGLE-FLOAT X))
4181 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4182 operation, without having to do run-time type dispatch.
4183 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4184 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4185 * Since I don't seem to be making big file reorganizations very often
4186 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4187 is overflowing my ability to conveniently back them up), I've finally
4188 checked the system into CVS. (The CVS repository is on my home system,
4189 not at SourceForge -- putting it on SourceForge might come later.)
4190 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4191 high-level GC-NOTIFY-FOO functions send their output. (There's
4192 still very little control of where low-level verbose GC functions
4193 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4194 less than it used to -- the GC-NOTIFY-FOO functions are now under
4195 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4196 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4197 in only one place in the source code, and propagates it automatically
4198 everywhere that it's needed. Thus e.g. when I bump the version from
4199 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4200 * The C source files now include boilerplate legalese and documentation
4201 at the head of each file (just as the Lisp source files already did).
4202 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4203 to his page will be replaced with a link to his new CLiki service.
4205 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4207 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4208 * The build process now looks for GNU make under the default name "gmake",
4209 instead of "make" as it used to. If GNU make is not available as "gmake"
4210 on your system, you can change this default behavior by setting the
4211 GNUMAKE environment variable.
4212 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4213 can build without error under CMU CL.
4215 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4217 * There is now a partial SBCL user manual (with some new text and some
4218 text cribbed from the CMU CL manual).
4219 * The beginnings of a profiler have been added (starting with the
4220 CMU CL profiler and simplifying and cleaning up). Eventually the
4221 main interface should be through the TRACE macro, but for now,
4222 it's still accessed through vaguely CMU-CL-style functions and macros
4223 exported from the package SB-PROFILE.
4224 * Some problems left over from porting CMU CL to the new
4225 cross-compilation bootstrap process have been cleaned up:
4226 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4227 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4228 part when getting rid of the compiler *BACKEND* stuff.)
4229 ** The value of *NULL-TYPE* was screwed up, because it was
4230 being initialized before the type system knew the final
4231 definition of the 'NULL type. This screwed up several key
4232 optimizations in the compiler, causing inefficiency in all sorts
4233 of places. (I found it because I wanted to understand why
4234 GET-INTERNAL-RUN-TIME was consing.)
4235 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4236 PROCLAIM FTYPE information. Unfortunately this broke binary
4237 compatibility again, since now the forms output by DEFGENERIC
4238 to refer to functions which didn't exist in 0.6.3.
4239 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4240 can use the new (as of 0.6.3) transform for SOME into MAP into
4242 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4243 values to (INTEGER 0 1000000), so that the time code will no longer
4244 occasionally get blown up by Linux returning 1000000 microseconds
4245 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4246 its output conform to the ANSI spec. (Alas, this makes its output
4247 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4248 * A full call to MAP NIL with a single sequence argument no longer conses.
4249 * fixes to problems pointed out by Martin Atzmueller:
4250 * The manual page no longer talks about multiprocessing as though
4251 it were currently supported.
4252 * The ILISP support patches have been removed from the distribution,
4253 because as of version 5.10.1, ILISP now supports SBCL without us
4254 having to maintain patches.
4255 * added a modified version of Raymond Toy's recent CMU CL patch for
4256 EQUALP comparison of HASH-TABLE
4258 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4260 * The system still can't cross-compile itself with
4261 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4262 entails), but at least it can compile more of itself that way
4263 than it used to be able to, and various buglets which were uncovered
4264 by trying to cross-compile itself that way have now been fixed.
4265 * This release breaks binary compatibility again. This time
4266 at least I've incremented the FASL file format version to 2, so that the
4267 problem can be detected reliably instead of just causing weird errors.
4268 * various new style warnings:
4269 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4270 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4271 ** using the lexical binding of a variable named in the *FOO* style
4272 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4273 as specified by ANSI.
4274 * *RANDOM-STATE* is no longer automatically initialized from
4275 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4276 default behavior of the system is to repeat its behavior every time
4277 it's run. If you'd like to change this behavior, you can always
4278 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4279 old convention there was no comparably easy way to get the system to
4280 repeat its behavior every time it was run.
4281 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4282 FTYPE declarations has been removed, in favor of their ANSI
4283 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4284 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4285 the types of their sequence arguments can be determined at compile time.
4286 This is done through a new open code expansion for MAP which eliminates
4287 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4288 argument types can be determined at compile time.
4289 * The optimizer now transforms COERCE into an identity operation when it
4290 can prove that the coerced object is already of the correct type. (This
4291 can be a win for machine generated code, including the output of other
4292 optimization transforms, such as the MAP transform above.)
4293 * Credit information has been moved from source file headers into CREDITS.
4294 * Source file headers have been made more standard.
4295 * The CASE macro now compiles without complaining even when it has
4298 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4300 * (Note that the way that the PCL macroexpansions were rewritten
4301 to accommodate the change in DEFGENERIC below breaks binary
4302 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4303 not run under sbcl-0.6.2. Once we get out of alpha releases,
4304 i.e. hit release 1.0.0, we'll probably try to maintain binary
4305 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4306 and sbcl-1.4.4. Until then, however, it might be fairly common
4307 for maintenance releases to break binary compatibility.)
4308 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4310 * The reporting of unhandled signals has been changed to print some
4311 explanatory text as well as the report form. (Previously only
4312 the report form was printed.)
4313 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4314 it defines, so that the compiler no longer issues undefined function
4315 warnings for compiled-but-not-yet-loaded generic functions.
4316 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4317 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4318 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4319 section of the ANSI spec are supported.
4320 * The "" nickname for the "KEYWORD" package has been removed.
4321 The reader still handles symbol tokens which begin with a package marker
4322 as keywords, but it doesn't expose its mechanism for doing so in the
4323 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4324 * The system now issues STYLE-WARNINGs for contradictory TYPE
4325 proclamations. (Warnings for contradictory FTYPE proclamations would
4326 be nice too, but those can't be done usefully unless the type system
4327 is made smarter about FUNCTION types.)
4328 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4329 systematized, so that "*target-*.lisp is supposed to exist only on the
4330 target and imply that there's a related file which exists on the
4331 host, and *host-*.lisp is supposed to exist only on the host and imply
4332 that there's a related file which exists on the target. This involves a
4333 lot of renaming. Hopefully the acute confusion caused by the renaming
4334 will be justified by the reduction in chronic confusion..
4335 ** runtime-type.lisp -> early-target-type.lisp
4336 ** target-type.lisp -> late-target-type.lisp
4337 ** early-host-format.lisp -> early-format.lisp
4338 ** late-host-format.lisp -> late-format.lisp
4339 ** host-error.lisp -> misc-error.lisp
4340 ** early-error.lisp -> early-target-error.lisp
4341 ** late-error.lisp -> late-target-error.lisp
4342 ** host-defboot.lisp -> early-defboot.lisp
4343 ** code/misc.lisp -> code/target-misc.lisp
4344 ** code/host-misc.lisp -> code/misc.lisp
4345 ** code/numbers.lisp -> code/target-numbers.lisp
4346 ** code/early-numbers.lisp -> numbers.lisp
4347 ** early-host-type.lisp -> early-type.lisp
4348 ** late-host-type.lisp -> late-type.lisp
4349 ** host-typep.lisp -> typep.lisp
4350 ** load.lisp -> target-load.lisp
4351 ** host-load.lisp -> load.lisp
4352 ** host-disassem.lisp -> disassem.lisp
4353 ** host-insts.lisp -> insts.lisp
4354 ** byte-comp.lisp -> target-byte-comp.lisp
4355 ** host-byte-comp.lisp -> byte-comp.lisp
4356 ** host-signal.lisp -> signal.lisp
4357 ** host-defstruct.lisp -> defstruct.lisp
4358 ** late-target-type.lisp -> deftypes-for-target.lisp
4359 Furthermore, several other previously target-only files foo.lisp (e.g.
4360 hash-table.lisp and random.lisp) have been split into a target-and-host
4361 foo.lisp file and a target-only target-foo.lisp file, with their key type
4362 definitions in the target-and-host part, so that the cross-compiler will
4363 know more about target types.
4364 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4365 gone away. In their place are various *BACKEND-FOO* variables
4366 corresponding to the slots of the old structure.
4367 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4368 into the target SBCL has been fixed.
4369 * The chill.lisp system for loading cold code into a running SBCL
4371 * Support for the CMU CL "scavenger hook" extension has been removed.
4372 (It was undocumented and unused in the CMU CL sources that SBCL was
4373 derived from, and stale in sbcl-0.6.1.)
4374 * Various errors in the cross-compiler type system were detected
4375 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4376 (enabling various consistency checks). Many of them were fixed,
4377 but some hard problems remain, so the compiler is back to
4378 running without *TYPE-SYSTEM-INITIALIZED* for now.
4379 * As part of the cross-compiler type system cleanup, I implemented
4380 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4381 * I have started adding UNCROSS calls throughout the type system
4382 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4383 on cross-compiler input files will be able to go away, and various
4385 * CONSTANTP now returns true for quoted forms (as explicitly required
4388 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4390 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4391 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4392 (MAKE-LIST -1). In the longer term, it ought to become true
4393 that declarations are assertions even at SAFETY 1. For now, it's not
4394 quite true even at SAFETY 3, but it's at least more nearly true..
4395 (Note that this change seems to increases the size of the system by
4396 O(5%) and to decrease the speed of the compiler by 20% or more.)
4397 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4398 for the problem of printing dozens of lines of distracting information
4399 about low-level system machinery as part of the top stack frame
4400 on entry to the debugger when an undefined function was called.
4401 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4402 is not reset to COMMON-LISP-USER.
4403 * Compilation of stuff related to dyncount.lisp has been made conditional
4404 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4405 smaller. The various dyncount-related symbols have been moved into
4406 a new "SB-DYNCOUNT" package.
4407 * tty-inspect.lisp has been renamed to inspect.lisp.
4408 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4409 feature has gone away. (When we eventually port to other flavors of
4410 libc and/or Unix, we'll try to make the differences between flavors
4411 invisible at the user level.)
4412 * Various other *FEATURES* tags, and/or their associated conditionals,
4413 have been removed if obsolescent, or given better documentation, or
4414 sometimes given more-mnemonic names.
4416 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4418 * tidied up "make.sh" script
4419 * tidied up system directory structure
4420 * better "clean.sh" behavior
4421 * added doc/FOR-CMUCL-DEVELOPERS
4422 * many many small tweaks to output format, e.g. removing possibly-confusing
4423 trailing #\. character in DESCRIBE-INSTANCE
4424 * (EQUALP #\A 'A) no longer signals an error.
4425 * new hashing code, including EQUALP hashing
4426 * tidied up Lisp initialization and toplevel
4427 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4428 * command line argument processing
4429 * added POSIX-GETENV function to deal with Unix-ish environment variables
4430 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4431 terminating SBCL on EOF
4432 * non-verbose GC by default
4433 * There is no more "sbcl" shell script; the sbcl file is now the C
4434 runtime executable (just like CMU CL).
4435 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4437 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4438 symbol and package information as Lisp native symbol and package objects
4439 instead of strings naming symbols and strings naming packages. This way,
4440 whenever packages are renamed (as in warm init), debug information is
4441 transformed along with everything else.
4442 * tweaked the optimization policy declarations which control the building
4443 of SBCL itself. Now, among other things, the system no longer saves
4444 source location debugging information. (This helps two problems at once
4445 by reducing SBCL size and by keeping SBCL from trying to look for its
4446 sources -- which may not exist -- when reporting errors.)
4447 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4448 debugging and testing purposes
4449 * cleaned up printing, making the printer call PRINT-OBJECT for
4450 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4451 methods, giving nearly-ANSI behavior
4452 * converted almost all special variables to use *FOO* naming convention
4453 * deleted PARSE-TIME functionality, since it can be done portably
4454 * moved some files out of cold init into warm init
4455 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4457 * regularized formatting of source files
4458 * added an install.sh script
4459 * fixed ridiculous memory usage of cross-compiler by making
4460 compiler/alloc.lisp not try to do pooling unless it can hook
4461 itself into the GC of the cross-compilation host. Now the system
4462 builds nicely on my old laptop.
4463 * added :SB-ALLOC in target-features.lisp-expr
4464 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4465 was not implemented)
4466 * re-did condition handling and note reporting in the compiler. Notes
4467 are no longer handled by signalling conditions. Style warnings
4468 and warnings are handled more correctly and reported in such a way
4469 that it's easy to find one or the other in your output (so that you
4470 can e.g. figure out which of many problems caused COMPILE-FILE to
4472 * changed the severity of several compiler warnings from full WARNING
4473 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4474 compiler note reporting so that it doesn't use the condition system
4475 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4476 COMPILE-FILE command)
4477 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4478 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4479 no longer supported, since their implementation was deeply intertwingled
4480 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4481 * removed bogus "support" for compiler macros named (SETF FOO), and
4482 removed the compiler macro for SETF INFO (but only after making a fool
4483 of myself on the cmucl-imp mailing list by posting a bogus patch for
4484 DEFINE-COMPILER-MACRO..)
4485 * Compiled files containing forms which have side effects on the Lisp
4486 reader (such as DEFPACKAGE forms) are now handled more correctly.
4487 (Compiler queuing of top level lambdas has been suppressed by setting
4488 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4489 * deleted various currently-unused source files, e.g. gengc.lisp. They
4490 may be added back at some point e.g. when porting to other architectures,
4491 but until they are it's distracting to distribute them and to try to
4493 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4494 there were so many of them they're just distractions, and UNCROSS is
4495 known to be able to handle the current sources
4496 * moved PROFILE functionality into TRACE, so that it will be clear
4497 how the wrapping and unwrapping of functions when you profile them
4498 interacts with the wrapping and unwrapping of functions when you
4499 trace them. (Actually, the functionality isn't there yet, but at least
4500 the interface specification is there. Hopefully, the functionality will
4501 arrive with some future maintenance release.)
4502 * removed host-oops.lisp
4503 * changed signature of QUIT function to allow UNIX-CODE argument
4504 * fixed READ-SEQUENCE bug
4505 * tweaked verbose GC output so that it looks more like the progress
4506 output that ANSI specifies for functions like LOAD
4507 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4508 * added <http://sbcl.sourceforge.com> to the banner information printed by