1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
3 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
4 unparsing of directory pathnames as files. Analogously,
5 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
6 filename to parse into a directory pathname.
7 * optimization: SUBSEQ and COPY-SEQ are 30-80% faster for strings
8 and vectors whose element-type or simplicity is not fully known at
10 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
12 * bug fix: some sequence functions elided bounds checking when
14 * bug fix: too liberal weakening of union-type checks when SPEED >
16 * bug fix: more bogus fixnum declarations in ROOM implementation
19 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
20 * new feature: MAKE-HASH-TABLE now experimentally accepts a
21 :SYNCHRONIZED argument, which makes the hash-table safe for
22 concurrent accesses (but not iteration.) See also:
23 SB-EXT:WITH-LOCKED-HASH-TABLE, and
24 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
25 * optimization: CONCATENATE on strings is an order of magnitue faster
26 in code compiled with (> SPEED SPACE).
27 * optimization: SUBSEQ is ~50% faster on lists.
28 * bug fix: bug 417 fixed -- source location reporting is now more robust.
29 * bug fix: SUBSEQ on a list will now correctly signal an error if if
30 END is smaller then START.
31 * bug fix: SB-PROFILE will no longer report extra consing for nested
32 calls to profiled functions.
33 * bug fix: ROOM implementation had bogus fixnum declarations which
34 could cause type-errors when calling ROOM on large images.
35 * bug fix: if file compilation is aborted, the partial fasl is now
36 deleted, and COMPILE-FILE returns NIL as the primary value.
37 * bug fix: number of thread safety issues relating to SBCL's internal
38 hash-table usage have been fixed.
39 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
40 be returned from its body when the values were being returned
41 using unknown-values return convection and the W-P-O was wrapped
42 inside an UNWIND-PROTECT.
43 * bug fix: sb-posix should now compile again under Windows, enabling
46 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
47 * incompatible change: hash-table accessor functions are no longer
48 automatically protected by locks. Concurrent accesses on the same hash-table
49 from multiple threads can give inconsistent results or even corrupt the
50 hash-table completely. Multi-threaded applications should do their own
51 locking at the correct granularity. In the current implementation it is
52 still safe to have multiple readers access the same table, but it's not
53 guaranteed that this property will be maintained in future releases.
54 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
55 and will signal an error at runtime.
56 * enhancement: SB-THREAD package now exports a semaphore interface.
57 * enhancement: CONS can now stack-allocate on x86 and
58 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
60 * enhancement: nested lists can now be stack allocated on
61 platforms providing stack allocation support.
62 * enhancement: dynamic-extent support has been extended to support
63 cases where there are multiple possible sources for the stack
65 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
66 if the mutex is uncontested on Linux.
67 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
68 as specified by the CLHS. (thanks to Tobias Rittweiler)
69 * bug fix: NaN comparison now works on x86-64.
70 * bug fix: CLOSE :ABORT T on a stream with pending output now
72 * bug fix: instances of non-standard metaclasses using standard
73 instance structure protocol sometimes missed the slot type checks
75 * bug fix: known functions can be inlined in high-debug code.
77 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
78 * minor incompatible change: the MSI installer on Windows no longer
79 associates .lisp and .fasl files with the installed SBCL.
80 * minor incompatible change: :UNIX is no longer present in *FEATURES*
81 on Windows. (thanks to Luis Oliviera)
82 * new feature: SB-CLTL2 contrib module now implements
83 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
84 * optimization: scavenging weak pointers is now more efficient,
85 requiring O(1) instead of O(N) per weak pointer to identify
86 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
87 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
88 method bodies using constant slot-names when the first argument is
89 a specializer parameter for the method.
90 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
91 names, when the class of the instance is a direct instance
92 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
93 as slow as the constant slot-name case.)
94 * optimization: member type construction is now O(N) instead
96 * optimization: UNION and NUNION are now O(N+M) for large
97 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
99 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
100 now more readable in environments like Slime which display it.
101 (thanks to Tobias C. Rittweiler)
102 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
103 was able to infer the correct argument type for the object on which
104 the CAS operation was being performed.
105 * bug fix: copy propagation interfered with parallel assignment
106 semantics in local calls. (reported by Paul Khuong)
107 * bug fix: the signed modular fixnum shift compiled to wrong code on
108 x86 and x86-64. (spotted by a slight modification to some of
111 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
112 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
113 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
114 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
116 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
117 and generic function now print the object identity for anonymous
118 instances. (thanks to Attila Lendvai)
119 * enhancement: as an extension to MOP, SBCL now supports SETF of
120 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
121 (thanks to Attila Lendvai)
122 * enhancement: sb-cover has an alternate annotation mode which puts
123 the form annotations at the car of the form, instead of annotating
124 it completely. The benefit of this mode is that it shows better
125 which forms were instrumented by the compiler.
126 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
127 by printer control variables. (reported by Dan Corkill)
128 * bug fix: system leaked memory when delayed output was performed by
129 the OS in smaller chunks then expected. (thanks to David Smith)
130 * bug fix: system leaked memory when file streams were not closed
132 * bug fix: large objects written to slow streams that were modified
133 after the write could end up with the modified state written to
134 the underlying file descriptor.
135 * bug fix: multiple threads operating in parallel on the same stream
136 could cause buffer-overflows.
137 * bug fix: source location information is stored correctly
138 (broken since 1.0.6). This bug would generally show up as the
139 Slime debugger higlighting the wrong form when the sldb-show-source
141 * bug fix: Branch forms are again annotated as branches in the sb-cover
143 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
144 on x86-64. (reported by Gregory Vanuxem)
145 * bug fix: EQUALP could return wrong results for structures with raw
146 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
147 word). (reported by Vjacheslav Fyodorov)
148 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
151 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
152 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
153 atomic compare-and-swap operations on threaded platforms.
154 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
155 allows assining a global minimum value to optimization qualities
156 (overriding proclamations and declarations).
157 * enhancement: closed over variables can be stack-allocated on x86
159 * performance bug fix: GETHASH and (SETF GETHASH) are once again
161 * optimization: slot definition lookup is now O(1). This speeds up
162 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
163 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
164 * optimization: ASSOC and MEMBER can now be open-coded for all
165 combinations of keyword arguments when second argument is constant
166 and SPEED >= SPACE. In other cases a specialized version is
168 * bug fix: using obsoleted structure instances with TYPEP and
169 generic functions now signals a sensible error.
170 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
171 (reported by Kristoffer Kvello)
172 * bug fix: backtrace construction is now more careful when making
173 lisp-objects from pointers on the stack, to avoid creating bogus
174 objects that can be seen by the GC.
175 * bug fix: defaulting of values in contexts expecting more than 7
176 variables now works on x86-64. (reported by Christopher Laux)
177 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
179 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
180 as the property-list of a symbol.
181 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
182 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
183 situations requested, are once again file-compileable. (reported
186 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
187 * MOP improvement: support for user-defined subclasses of
188 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
189 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
190 the expansion of DEFMETHOD, is responsible for generating a form
191 which creates a list of specializers when evaluated. Additional
192 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
193 debugging and introspective support.
194 * minor incompatible change: the (unsupported) spinlock interface
195 has changed: free spinlock now has the value NIL, and a held spinlock
196 has the owning thread as its value.
197 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
198 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
200 * enhancement: name of a socket-stream is now "a socket" instead of
202 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
203 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
204 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
205 for generic functions. (Thanks to Tobias C. Rittweiler)
206 * optimization: bignum printing speed has been improved by 20-40%
207 (depending on the bignum size.)
208 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
210 * bug fix: the cache used by the CLOS to store precomputed effective
211 methods, slot offsets, and constant return values is now thread and
213 * bug fix: generic function dispatch function updating is now thread
214 and interrupt safe (in the sense that the known issues have been
216 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
217 * bug fix: interning EQL-specializers is now thread and interrupt safe.
218 * bug fix: asdf systems with dependencies to the SB-POSIX or
219 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
220 * bug fix: interrupt safety of applicable method computation has been
223 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
224 * new contrib: sb-cover, an experimental code coverage tool, is included
226 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
227 significantly faster.
228 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
229 produce more efficient code on x86-64 if the compiler can determine
230 that (+ INDEX <constant>) does not require a bounds check and FOO
231 has an element type at least 8 bits wide.
232 * enhancement: a new, experimental synchronous timeout facility is
233 provided. Refer to SB-SYS:WITH-DEADLINE for details.
234 * enhancement: when a symbol name conflict error arises, the
235 conflicting symbols are always printed with a package prefix.
236 (thanks to Kevin Reid)
237 * enhancement: stepping is now once again supported on the SPARC and
238 MIPS platforms. (It is also now more likely to work on CheneyGC
240 * enhancement: sb-sprof can now also track and report accurate call
242 * bug fixes: the treatment of non-standard subclasses of
243 SB-MOP:SPECIALIZER is more correct.
244 * incompatible change: PURIFY no longer copies the data from the
245 dynamic space into the static and read-only spaces on platforms
246 that use the generational garbage collector
247 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
249 * bug fix: GC race condition occasionally resulting in crashes with
250 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
252 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
254 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
255 a STYLE-WARNING to be signalled (regression from 1.0.4.)
256 * bug fix: an asynchronous interrupt could previously leave the
257 system running with GC inhibited.
258 * bug fix: a DECLARE form evaluated at top-level now causes an error
259 rather than silently (or verbosely) returning NIL.
260 * bug fix: trying to dissassemble functions compiled at high DEBUG could
261 result in a error being signalled due to source form lookup errors.
262 (reported by Peter Graves)
264 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
265 * incompatible change: removed writer methods for host-ent-name,
266 host-ent-addresses -- changing the values did not update the DNS
268 * minor incompatible change: changed experimental JOIN-THREAD interface
269 * documentation: the manual now lists reader and writer methods
270 in class slot documentation sections. (thanks to Richard M Kreuter)
271 * documentation: unwinding from asyncronous events has been
272 documented as unsafe.
273 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
274 in multithreaded application code.
275 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
276 platforms. (thanks to James Anderson for the optimization hint)
277 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
279 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
280 variants no longer cons.
281 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
282 their NOT- variants no longer cons.
283 * optimization: EQUAL hash tables no longer use SXHASH for objects
284 of all data types, but instead use an EQL hash for types for which
285 EQUAL is the same as EQL.
286 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
287 are significantly faster.
288 * optimization: new STRING-OUTPUT-STREAM implementation is much
289 faster for large outputs and conses 30% less on average.
290 * enhancement: XREF information is now collected for references made
291 to global variables using SYMBOL-VALUE and a constant argument.
292 * enhancement: SIGINT now causes a specific condition
293 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
294 * bug fix: FILE-POSITION used to signal an error in some cases where
295 ANSI requires it to return NIL.
296 * bug fix: ADJUST-ARRAY is now interrupt-safe.
297 * bug fix: adding and removing fd-handlers is now interrupt-safe.
298 * bug fix: inlined calls to C now ensure 16byte stack alignment on
300 * bug fix: bad type declaration in the CLOS implementation has
301 been fixed. (reported by James Anderson)
302 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
303 (thanks to Lutz Euler)
304 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
305 required by the CLHS. (reported by Samium Gromoff)
306 * bug fix: dead unbound variable references now signal an error.
307 (reported by Marco Monteiro)
308 * bug fix: / with an unused value was being deleted in safe code.
309 (thanks to Marco Monteiro and Kevin Reid)
310 * bug fix: number of characters that can be written onto a single
311 line in a file is unlimited.
312 * bug fix: some GC deadlocks caused by asynchronous interrupts have
313 been fixed by inhibiting interrupts for when GC is disbled.
314 * bug fix: some interrupt safety issues with GETHASH, (SETF
315 GETHASH), CLRHASH and REMHASH have been fixed.
316 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
317 specifier no longer causes infinite recursion.
318 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
319 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
320 * bug fix: an &environment argument with the correct variable information
321 is passed to macros that are expanded during byte compilation.
322 (reported by Samium Gromoff)
323 * bug fix: a logic error was causing the hash function for bignums to
324 have excessive amounts of collisions. (reported by Faré Rideau)
325 * bug fix: modifying the contents of an array could change the return
326 value of SXHASH on that array, which is only allowed for strings
327 and bit vectors. (bug introduced in 0.9.16)
328 * improvement: the x86-64/darwin port now passes all tests (except
329 for the debugger tests) but should still be considered
330 experimental until this is fixed.
331 * improvement: a style-warning is signaled for CASE (etc) clauses with
332 duplicate keys. (thanks to Kevin Reid)
333 * improvement: macos/x86 and macos/x86-64 now use mach exception
334 handlers for dealing with illegal instructions (for trapping and
335 error handling) and memory protection violations (for GC).
337 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
338 * new platform: experimental support for x86-64/darwin (MacOS).
339 * incompatible change: the thread-safe (on most platforms) getaddrinfo
340 and getnameinfo sockets functions are used instead of gethostbyaddr
341 and gethostbyname, on platforms where the newer functions are available.
342 As a result, the ALIASES field of HOST-ENT will always be NIL on these
344 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
345 don't print the startup banner, but behave as if --noinform was passed
346 as a command line argument. (thanks to Kevin Reid)
347 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
348 * optimization: code using alien values with undeclared types is much faster.
349 * optimization: the compiler is now able to open code SEARCH in more cases.
350 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
351 * bug fix: using standardized COMMON-LISP special variables as loop
352 variables no longer signals bogus package lock violations. (reported
354 * bug fix: declaring local loop variables to be of a range-limited type
355 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
356 (reported by Andras Simon)
357 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
358 bugs remain on x86-64.)
359 * bug fix: the #= and ## reader macros now interact reasonably with
360 funcallable instances.
361 * bug fix: type-checks for function arguments were compiled using the
362 compiler policy settings of the wrong lexical environment (in 1.0.2
364 * bug fix: SHADOW accepts characters as string designators, as required
365 by the spec (thanks to Eric Marsden)
366 * bug fix: fixed GC safety issues when foreign functions are called with
367 non-base strings as arguments
368 * bug fix: more consistent error output for fatal-compiler-warnings, like
370 * bug fix: functions with proper names were showing up as NIL in
371 backtraces (reported by Edi Weitz, regression in 1.0.2)
373 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
374 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
375 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
376 produce more efficient code on the x86 if the compiler can determine
377 that (+ INDEX <constant>) does not require a bounds check and FOO
378 has an element type at least 8 bits wide.
379 * bug fix: references to undefined variables are handled the same way
380 in toplevel forms as in the normal compiler
381 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
383 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
384 (thanks to Jon Buller)
385 * improvement: faster compilation times for complex functions
386 * improvement: added readlink support to SB-POSIX (thanks to Richard
389 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
390 * improvement: experimental support for mach exception handling on
391 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
393 * improvement: support for GBK external format.
394 (thanks to Chun Tian (binghe))
395 * improvement: the debugger now displays variables that have been closed
396 over, in code compiled with (DEBUG 2) or higher
397 * improvement: support for executable cores on NetBSD (thanks to
399 * new feature: added a RESTART-FRAME debugger command
400 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
401 be used to provide an implementation for FILE-POSITION on Gray streams
402 (thanks to Eric Marsden)
403 * new feature: add syslog, openlog and closelog support to SB-POSIX
404 (thanks to Richard Kreuter)
405 * optimization: the function call overhead in code compiled with
406 a high DEBUG optimization setting is significantly
407 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
408 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
409 (thanks to Tony Martinez)
410 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
411 evaluated once (reported by Kevin Reid)
412 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
413 works again (thanks to Kilian Sprotte)
414 * bug fix: an error is signaled for tagbodies with duplicate tags
415 (thanks to Stephen Wilson)
416 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
417 * bug fix: Win32 port can now handle foreign code unwinding Lisp
418 stack frames from alien callbacks.
419 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
420 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
421 65k pages (thanks to David Woodhouse)
422 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
424 changes in sbcl-1.0.1 relative to sbcl-1.0:
425 * new platform: FreeBSD/x86-64, including support for threading.
426 * new feature: the compiler stores cross-referencing information
427 abount function calls (who-calls), macroexpansion (who-macroexpands)
428 and special variables (who-binds, who-sets, who-references) for code
429 compiled with (< SPACE 3). This information is available through the
430 sb-introspect contrib.
431 * new feature: users may subclass SEQUENCE, and have instances of
432 these classes interoperate with standard Common Lisp functions if
433 a number of methods are defined. (This feature is experimental
434 and the interface subject to change based on feedback from SBCL
435 users and the general community)
436 * improvement: sb-sprof traces call stacks to an arbitrary depth on
437 x86/x86-64, rather than the previous fixed depth of 8
438 * improvement: another pthread back-end of mutex "pthread-futex".
439 * bug fix: non-ascii command-line arguments are processed correctly
440 (thanks to Yaroslav Kavenchuk)
441 * bug fix: non-required arguments were not passed correctly when a method
442 defined using DEFMETHOD was called from a mop-generated method using
443 CALL-NEXT-METHOD (reported by Pascal Costanza)
444 * bug fix: recursion is now permitted in accessors through
445 SLOT-UNBOUND. (reported by Pascal Costanza)
446 * bug fix: an error was signaled at startup if the HOME environment
447 variable was defined, but had an empty value (reported by Peter Van Eynde)
448 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
449 signal a PROGRAM-ERROR, not a BUG.
450 * bug fix: SB-POSIX:READDIR works when built with large file support.
451 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
453 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
454 for better automatic indentation support. (thanks to Matt Pillsbury)
455 * optimization: loading generic functions no longer takes O(n^2) time,
456 proportional to the amount of methods in the generic function
457 (reported by Todd Sabin and Jeremy Brown)
458 * optimization: the FIND and POSITION family of sequence functions
459 are significantly faster on arrays whose element types have been
461 * improvements to the Windows port:
462 ** Intermittent heap corruption problems have been fixed. (thanks
463 to Alastair Bridgewater)
464 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
466 ** Lisp is able to unwind foreign exception frames from alien
467 callbacks. (thanks to Alastair Bridgewater)
469 changes in sbcl-1.0 relative to sbcl-0.9.18:
470 * improvement: experimental support for threading on FreeBSD/x86.
471 (thanks to NIIMI Satoshi)
472 * improvement: runtime option --dynamic-space-size can be used
473 to set the size of the dynamic space reserved on startup.
474 * improvement: floating point modes in effect are now saved in
475 core, and restored on startup.
476 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
477 startup, not time since first call to GET-INTERNAL-REAL-TIME.
478 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
479 threads are not running after *SAVE-HOOKS* have run.
480 * improvement: writes to CLOS instance slots are type-checked in code
481 compiled with (SAFETY 3)
482 * improvement: floating-point exception handling on FreeBSD (thanks to
484 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
485 (thanks to Zach Beane)
486 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
488 * improvement: added support for the Shift-JIS external format.
489 (contributed by NIIMI Satoshi)
490 * improvement: callbacks are supported on Linux/PPC. (thanks to
492 * bug fix: compiler bug triggered by a (non-standard) VALUES
493 declaration in a LET* was fixed. (reported by Kaersten Poeck)
494 * bug fix: file compiler no longer confuses validated and already
495 dumped structurres. (reported by Kaersten Poeck)
496 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
497 fill-pointer signals a type-error as required. (thanks to
499 * bug fix: disassemly of funcallable instances works.
500 * bug fix: single stepping on PPC.
501 * bug fix: fix thread-safety problems in the type system (generally
502 manifesting as nonsensical errors like "STRING is a bad type specifier
503 for sequences" or "The value 1 is not of type FIXNUM")
504 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
505 (reported by Josip Gracin)
506 * bug fix: an error is signaled for attempts to displace arrays with
507 incompatible element types (thanks to Mario Mommer)
508 * bug fix: more correct handling of wide characters in debug info
509 (bug reported by Attila Lendvai and fixed by Juho Snellman)
510 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
511 and don't cause extra consing
512 * optimization: MAP and MAP-INTO are significantly faster on vectors
513 whose elements types have been declared.
514 * Improvements to SB-SPROF:
515 ** Support for allocation profiling
516 ** Reduced profiling overhead, especially for long profiling runs
517 * Improvements to the Windows port:
518 ** floating point exceptions are now reported correctly.
519 ** stack exhaustion detection works partially.
520 ** more accurate GET-INTERNAL-REAL-TIME.
521 ** better RUN-PROGRAM behaviour in case of errors in the child
523 ** PROBE-FILE now simplifies pathnames correctly.
524 ** DIRECTORY now works correctly with :WILD-INFERIORS.
525 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
526 1st 1970 (although time timezone data will be unreliable).
527 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
529 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
530 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
531 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
533 * bug fix: two potential GC deadlocks affecting threaded builds.
534 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
535 returns T (reported by Anton Kazennikov)
536 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
537 with non-variable places
538 * bug fix: remove a race condition in the setting of
539 funcallable-instance functions, this should make threaded CLOS
540 code more stable against memory faults.
541 * bug fix: corruption of specials when unbinding is interrupted by an
542 asynchronous unwind (reported by Hannu Koivisto)
543 * improvement: the debugger will now also display local variables that
544 are only used once, for code compiled with a DEBUG optimization quality
547 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
548 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
549 * incompatible change: External-format support for FFI calls. The
550 SB-ALIEN:C-STRING no longer implies an ASCII
551 external-format. Instead the string is subject to external-format
552 conversion. Additionally return values of type SB-ALIEN:C-STRING
553 are by default of type (SIMPLE-ARRAY CHARACTER), not
554 SIMPLE-BASE-STRING. For an alien type that behaves like the old
555 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
556 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
557 * incompatible change: SB-EXT package no longer contains the
558 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
559 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
561 * incompatible change: the single-stepper is no longer available
562 on Alpha, Mips and Sparc platforms.
563 * minor incompatible change: the direct superclasses of
564 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
565 not (STANDARD-OBJECT FUNCTION). This makes the
566 class-precedence-lists of GENERIC-FUNCTION and
567 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
569 * minor incompatible change: the default stream external format on
570 non-unicode SBCL is no longer detected from the locale, but is always
572 * new feature: Add a version of evaluator that uses an interpreter instead
573 of the compiler. EVAL still uses the compiler by default, to switch it
574 to use the interpreter, set the value of the variable
575 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
576 * minor incompatible change: the single-stepper REPL has been merged
577 with the normal debugger (see the "Stepping" heading of the debugger help
578 for more details). The debugger command STEP will no longer switch
579 to the single-stepper REPL.
580 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
581 for a type now works.
582 * bug fix: loading of default sysinit file works. (thanks to Leonid
584 * bug fix: better detection of circularities in the file-compiler.
585 (reported by Marco Monteiro)
586 * bug fix: the CL pathname functions now work with files that have
587 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
588 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
589 systems with Unix98 pty semantics.
590 * bug fix: ASDF-INSTALL will now work with bsd tar.
591 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
593 * bug fix: timers expiring in dead threads no longer cause a
594 type-error (reported by Paul "Nonny Mouse").
595 * bug fix: thanks to more lightweight single-stepper instrumentation,
596 code compiled with (DEBUG 3) will compile and execute significantly faster,
597 and will have more accurate type-inferencing than before
598 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
599 whose bindings are modified
600 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
601 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
602 ** replace dummy implementations of CL:MACHINE-INSTANCE and
603 CL:SOFTWARE-VERSION with proper version
605 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
606 * feature: implemented the READER-METHOD-CLASS and
607 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
608 as specified by AMOP.
609 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
611 * optimization: faster LOGCOUNT implementation on x86 and x86-64
612 (thanks to Lutz Euler)
613 * optimization: hashing of general arrays and vectors has been
614 improved. (reported by Any Fingerhut)
615 * enhancement: SB-INTROSPECT is now able to find definitions of
616 profiled functions. (thanks to Troels Henriksen)
617 * enhancement: compiler-macro expansion applies now to FUNCALL forms
618 as well, allowing compiler-macros for SETF-functions to expand.
619 * enhancement: step-instrumentation no longer wraps known
620 single-value functions with multiple-value context, allowing
621 better type inference.
622 * fixed bug #337: use of MAKE-METHOD in method combination now works
623 even in the presence of user-defined method classes. (reported by
624 Bruno Haible and Pascal Costanza)
625 * fixed bug #339(c): if there are applicable methods not part of any
626 long-form method-combination group, call INVALID-METHOD-ERROR.
627 (reported by Bruno Haible)
628 * fixed bug #361: the :FUNCTION initarg in the protocol for
629 initialization of methods can now be used to override
630 internally-produced optimized functions. (reported by Bruno
632 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
633 system-provided lambda expression no longer cause warnings about
634 unbound #:|pv-table| symbols.
635 * bug fix: improved the handling of type declarations and the
636 detection of violations for keyword arguments with non-constant
638 * bug fix: potentially erronous calls to PATHNAME and
639 MERGE-PATHNAMES were being flushed in some cases.
640 (reported by Richard Kreuter)
641 * bug fix: compiled calls to TYPEP were mishandling obsolete
642 instances. (reported by James Bielman and Attila Lendvai)
643 * bug fix: format strings with ~> without matching ~< no longer
644 trigger an AVER, but signal an understandable error instead.
645 (reported by Antonio Martinez)
646 * bug fix: specifying an output-file without a pathname-type for
647 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
648 of input-file instead of "fasl". (reported by Robert Dodier)
649 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
650 (reported by James Y Knight).
651 * bug fix: compiler-macros-function did not consider the environment
652 argument for shadowing by local functions.
653 * bug fix: compiler-macros expansion was inhibited by local INLINE
655 * bug fix: inline expansions of known functions were subject to
656 step-instrumentation in high debug policies, leading to problems
658 * bug fix: compiler failed to differentiate between different CONS
660 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
661 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
662 * bug fix: #368: incorrect use of expressed vs. upgraded array
664 * bug fix: #306a: more precise unions of array types.
665 * thread-safety improvements:
666 ** CONDITION-WAIT could return early on Linux, if the thread was
667 interrupted and subsequently continued with SIGCONT.
668 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
670 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
671 * added support for the ucs-2 external format. (contributed by Ivan
673 * minor incompatible change: pretty printing of objects of type
674 (cons symbol) is, in the default pprint-dispatch-table, now
675 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
677 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
678 called later in a class's lifetime, possibly as late as when the
679 first instance of the class is created. Previously,
680 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
681 class became finalizeable.
682 * fixed bug: FILE-POSITION sometimes returned inconsistent results
683 for multibyte external-format streams. (thanks to "vbzoli")
684 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
685 with :ALLOCATION :CLASS inherited from superclasses of the
687 * fixed bug: anonymous classes can now be created using the :NAME
688 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
689 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
690 * fixed bug: core-files saved with :EXECUTABLE T can again be
691 executed when SBCL_HOME isn't set. (reported by James Knight)
692 * fixed bug: toplevel LOCALLY forms with declarations could
693 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
694 * fixed bug: printing from several different threads using different
695 values of *print-case* could cause invalid output, due to
696 some internal special variables of the printer not being bound
697 thread-locally (reported by Max Mikhanosha)
698 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
699 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
700 (reported by Pascal Costanza)
701 * fixed bug: FINALIZE-INHERITANCE is called from
702 REINITIALIZE-INSTANCE on classes when the class has previously
703 been finalized, as required by AMOP.
704 * minor code generation optimizations:
705 ** better register allocation in CLOS dispatching functions
706 ** overflow detection when coercing signed bytes to fixnums on x86-64
707 ** is now implemented with one IMUL instruction instead of three shifts
708 ** more efficient bit-vector access on x86 and x86-64
709 ** more efficient access to raw structure slots on x86 and x86-64
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
714 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
715 * feature: thread support on Solaris/x86, and experimental thread support
717 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
719 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
720 default method for SLOT-UNBOUND.
721 * minor incompatible change: prevent the user from specializing the
722 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
723 somewhat counter to the intent of the protocol, I (CSR) think, and
724 additionally it just doesn't work in SBCL as currently
725 implemented, thanks to optimizations (that are always valid for
726 the other three SLOT-VALUEish functions, but not for the setter).
727 * minor incompatible change: the :SB-LDB feature is now enabled by
728 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
729 the low-level debugger.
730 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
731 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
732 down to an SB-SYS:FD-STREAM, allowing direct communication in
733 more cases. Temporary files and pipes are still used as a fallback
735 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
736 on the non-reentrant C functions localtime(3) and gmtime(3).
737 * bug fix: global optimization policy was not visible in LOCALLY and
739 * bug fix: class objects can be used as specializers in methods.
740 (reported by Pascal Costanza)
741 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
742 with a type error. (reported by blitz_ on #lisp)
743 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
744 failed with a type error. (reported by Pascal Bourguignon)
745 * bug fix: merging pathnames against defaults with :DIRECTORY
746 starting with '(:RELATIVE :BACK) should preserve the :BACK.
747 (reported by James Y Knight)
748 * bug fix: saving large (>2GB) cores on x86-64 now works
749 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
750 constant index and a (simple-array (signed-byte 32)) array
751 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
752 error (patch by Robert J. Macomber)
753 * feature: TIME now displays the amount of run-time spent in GC
754 * bug fix: The debugger now does a better job of respecting
755 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
756 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
757 workaround for bug 403.)
758 * bug fix: floating point exception handling now works on FreeBSD
759 (thanks to NIIMI Satoshi)
760 * fixed some bugs revealed by Paul Dietz' test suite:
761 ** MISC.641: LET-conversion were not supposed to work in late
763 * improvements to the Win32/x86 port:
764 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
765 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
767 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
768 * new feature: source path information is generated for macro-expansion
769 errors for use in IDE's like Slime (thanks to Helmut Eller)
770 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
771 * bug fix: compiler does not loop forever on an invalid type in
773 * improvement: compilation of most CLOS applications is significantly
775 * optimization: added a limited bytecode compiler for simple toplevel
776 forms, speeding up compilation and FASL loading
777 * bug fix: the statistical profiler now properly distinguishes anonymous
780 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
781 * minor incompatible change: in sbcl-0.9.11 (but not earlier
782 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
783 into subclasses of the finalized class. Now user calls to
784 FINALIZE-INHERITANCE finalize just the one class, and calls by the
785 system return before any subclasses are finalized.
786 * minor incompatible change: The reader no longer ignores errors
787 regarding non-existent packages in #+ and #- feature tests.
788 * new feature: command line options --no-sysinit, --no-userinit to
789 inhibit loading the corresponding init files
790 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
791 for sending data through UDP sockets (thanks to François-René Rideau)
792 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
793 error is signalled instead (thanks to François-René Rideau)
794 * minor incompatible change: Error signalling behaviour of lexical
795 operations violating package locks has changed slightly. Refer to
796 documentation on package locks for details.
797 * enhancement: EVAL can process IF-expressions without resorting to the
799 * bug fix: Some operations involving SIN, COS, or TAN weren't being
800 constant-folded properly due to missing out-of-line functions.
801 (reported by Mika Pihlajamäki)
802 * bug fix: LISTEN sometimes returned T even in cases where no data was
803 immediately available from the stream
804 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
805 were reversed. (reported by Levente Mészáros)
806 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
807 when given a vector displaced to an adjustable array. (reported
809 * bug fix: derivation of float boundaries from numbers outside the
810 appropriate float range (reported by John Wiseman)
811 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
813 * bug fix: precomputing cache entries for generic functions with
814 some subclasses of specializers as yet invalid does not attempt to
815 fill a cache line with a negative offset. (reported by Levente
817 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
818 allow condition class objects as arguments to DOCUMENTATION and
819 (SETF DOCUMENTATION); only find and set documentation for
820 structure names for the STRUCTURE doc-type. (suggested by Gary
822 * improvements to the Win32/x86 port:
823 ** (user-homedir-pathname) and default initialization file
824 locations now know about the user's "Documents and Settings"
825 directory (thanks to Yaroslav Kavenchuk)
826 ** run-program is implemented (thanks to Mike Thomas)
827 ** sockets support (thanks to Timothy Ritchey)
828 ** better backtrace support (thanks to Alastair Bridgewater)
829 ** sb-grovel supported
830 ** asdf-install and sb-posix work somewhat
831 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
832 * improvements to the Solaris/x86 port:
833 ** works on Solaris 11/Solaris Express
834 ** floating-point exception handling support
835 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
836 * fixed some bugs revealed by Paul Dietz' test suite:
837 ** REMOVE-METHOD returns its generic function argument even when
838 the method is not one of the generic functions' methods.
839 ** objects declared as MEMBER types can be admissible arguments to
841 ** printing characters should simply be printed by the FORMAT ~:C
843 ** compiler failure when compiling functions with hairy constant
844 defaults for optional parameters.
845 ** compiler produces wrong code when MAYBE-INLINE-expanding a
846 function, which is already optimized.
848 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
849 * new platform: experimental support for SBCL x86/Darwin, including
850 MacOS X 10.4.5 on Intel.
851 * new feature: Unicode character names are now known to the system
852 (through CHAR-NAME and NAME-CHAR).
853 * new feature: the filesystem location of SBCL's core file is
854 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
855 * minor incompatible change: the contrib modules SB-POSIX and
856 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
857 this change is to make it easier to distribute
858 location-independent binaries.
859 * bug fix: as implied by AMOP, standardized classes no longer have
860 slots named by external symbols of public packages. (reported by
862 * optimization: calling functions via constant symbols -- as in
863 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
864 via the function object as in (FUNCALL #'FOO).
865 * optimization: on x86 and x86-64, the calling convention for the
866 case of dealing with an unknown number of values has been altered
867 to be friendlier to the prediction heuristics implemented,
868 particularly with reference to CALL and RET pairing. (thanks to
869 Alastair Bridgewater)
870 * optimization: on x86, the code for access to thread-local symbol
871 values and binding/unbinding of thread-local symbols is smaller.
872 (thanks to Alastair Bridgewater)
873 * enhancement: CONSTANTP is now able to determine constantness of
874 more complex forms, including calls to constant-foldable standardized
875 functions and some special forms beyond QUOTE.
876 * fixed bug: occasional GC crashes on Solaris/x86
877 * optimization: x86-64 supports stack allocation of results of simple
878 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
879 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
880 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
881 (thanks to James Knight)
882 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
883 by the standard (thanks to Stephen Compall)
885 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
886 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
887 be used for bundling the runtime and the core file into one
888 executable binary. This feature is not currently supported on all SBCL
889 platforms. (thanks to James Bielman and NIIMI Satoshi)
890 * new feature: a generational or ephemeral garbage collector is now
891 the default on the PowerPC platform (both Linux and Darwin). The
892 old Cheney (stop and copy) collector is a build-time option.
893 (thanks to Cyrus Harmon, after Raymond Toy)
894 * minor incompatible change: the method by which SBCL finds its
895 contributed modules has changed; it no longer relies on symbolic
896 links from an $SBCL_HOME/systems directory, but searches directly
897 in the subdirectories of $SBCL_HOME.
898 * enhancement: the dynamic heap size on the Linux/PPC platform is
899 markedly larger, even using the older Cheney garbage collector.
900 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
901 certain complicated string types. (reported by Gary King)
902 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
903 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
904 * fixed bug: DOCUMENTATION on structures no longer signals an
905 error if no structure type of the right name exists. (reported by
907 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
908 applicable non-standard primary or :AROUND method on
909 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
910 no longer cause unbound variable errors. (reported by Kevin Reid)
911 * fixed bug: metacircle resolution in cases where methods have slots
912 added before the slots from STANDARD-METHOD. (reported by Jean
914 * fixed bug: the Power PC assembly code for calling into lisp
915 sometimes computed its return address wrongly (depending on the
916 15th bit of the address, fixed at link time). This used to
917 manifest itself in a segmentation violation while building PCL.
918 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
919 many others over the years)
920 * fixed bug: ROOM no longer reports silly numbers for stack usage if
921 the stack pointer (treated as a signed integer) is negative.
922 (thanks to Peter van Eynde)
924 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
925 * new platform: experimental support for the Windows operating
926 system has been added. (thanks to Alastair Bridgewater)
927 * fixed several bugs in and robustified the PPC FFI (including
928 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
929 * bug fix: allow non-simple string symbol names (reported by Paul
931 * bug fix: interrupt handling on NetBSD (thanks to Richard M
933 * bug fix: saving a core corrupted callbacks on x86/x86-64
934 * bug fix: closed a loophole in metacircularity detection and
935 grounding in the PCL implementation of CLOS.
936 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
937 (reported by tomppa on #lisp)
938 * bug fix: clear the direction flag on Lisp -> C transitions, as
939 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
940 (reported by Andrej Grozin and Hendrik Maryns)
941 * optimization: major improvements to GC efficiency on GENCGC platforms
942 * optimization: faster implementation of EQUAL
943 * optimization: emit more efficient opcodes for some common
944 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
945 * optimization: several other minor code-generation improvements on x86-64
946 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
948 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
949 * minor incompatible change: (SETF CLASS-NAME) and (SETF
950 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
951 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
952 by ANSI as a generic function, and for consistency (SETF
953 GENERIC-FUNCTION-NAME) is treated likewise.
954 * fixed bug #233.b: make constraint propagation notice when a variable
955 value is changed after it is referenced but before it is used
956 * fixed bug #296: no more arbitrary behaviour from filenames with
957 odd characters as --load arguments. (reported by Adam Warner)
958 * fixed bug #390: :CHARACTER-SET pathname components now work as
959 expected. (reported by Tim Daly Jr)
960 * fixed bug #391: complicated :TYPE intersections in slot
961 definitions no longer cause an error in PCL internals.
962 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
963 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
964 returning the number of octets which would be written to the
965 file-stream. (thanks to Robert J. Macomber)
966 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
967 arguments to the internal stream functions. (thanks to David
969 * bug fix: stack exhaustion is now handled on the x86/SunOS
970 platform. (thanks to Daisuke Homma)
971 * bug fix: more accurate ROOM results on GENCGC platforms
972 * optimization: improved type inference for arithmetic-for
973 index variables in LOOP
974 * optimization: faster floating-point SQRT on x86-64
975 * fixed some bugs revealed by Paul Dietz' test suite:
976 ** DOCUMENTATION returns NIL instead of "" for method combinations
977 that don't have a docstring
979 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
980 * minor incompatible change: (SETF CLASS-NAME) and (SETF
981 GENERIC-FUNCTION-NAME) are no longer generic functions, and
982 therefore are not customizeable by user code (as seems to be at
983 least permitted and maybe required by AMOP). As a consolation,
984 however, the SBCL implementation of these functions now calls
985 REINITIALIZE-INSTANCE as specified by AMOP.
986 * bug fix: slot-definition documentation is propagated as per ANSI
987 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
988 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
989 argument, as do eslotd initialization methods. (from Pascal
990 Costanza's "Closer" project)
991 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
992 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
994 * bug fix: it is now possible to have more than one subclass of
995 STANDARD-GENERIC-FUNCTION without causing stack overflow.
996 (reported by Bruno Haible, Pascal Costanza and others)
997 * bug fix: the dependent update protocol now works for generic
998 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1000 * bug fix: condition-class instances corresponding to
1001 DEFINE-CONDITION forms are now created eagerly. (reported by
1002 Kalle Olavi Niemitalo on comp.lang.lisp)
1003 * bug fix: floating point printing is more accurate in some
1004 circumstances. (thanks to Simon Alexander)
1005 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1006 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1007 * bug fix: callbacks on OS X now preserve stack-alignment.
1008 (thanks to Cyrus Harmon)
1009 * enhancement: the x86-64 disassembler is much better at
1010 disassembling SSE instructions. (thanks to Lutz Euler)
1011 * enhancement: CHECK-TYPE on a variable now allows type-inference
1012 (thanks to Vincent Arkesteijn)
1013 * optimization: improved performance of EUC-JP external format.
1014 (thanks to NIIMI Satoshi)
1015 * optimization: performance improvements to IO on file streams of
1016 :ELEMENT-TYPE CHARACTER
1017 * optimization: much faster memory allocation on x86-64
1018 * optimization: faster garbage collections (latency and throughput)
1020 * optimization: faster bignum subtraction and division on x86 and x86-64
1021 (thanks to Lutz Euler)
1022 * optimization: more accurate type-inference for loops that use a
1023 floating point index variable or a negative step.
1025 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1026 * bug fix: add a workaround to SBCL looping infinitely at startup on
1027 Linux kernels with apparently buggy implementations of personality().
1028 (thanks to Svein Ove Aas)
1029 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1030 (thanks to David Lichteblau)
1031 * bug fix: local bindings shadow global symbol macros for
1032 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1033 * optimization: non-open coded uses of numeric comparison operators
1034 (e.g. >) no longer cons when called with more than one parameter
1035 on platforms supporting dynamic-extent allocation.
1036 * enhancement: saving cores with foreign code loaded is now supported
1037 on MIPS/Linux in addition to the previously supported platforms.
1038 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1039 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1040 * bug fix: division by zero in sb-sprof when no samples were collected
1041 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1042 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1043 (thanks to Alexander Kjeldaas)
1044 * new feature: ignoring the return values of destructive functions
1045 like NREVERSE signals a compile-time style-warning.
1046 (thanks to Kevin Reid)
1047 * new feature: the alignment of alien structure fields can be
1048 explicitly specified. (thanks to Cyrus Harmon)
1049 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1050 and dump core on SIGQUIT
1052 ** incompatible change: threads do not inherit values of specials
1053 from their parents (see manual)
1054 ** bug fix: threads stacks belonging to dead threads are freed by the
1055 next exiting thread, no need to gc to collect thread stacks anymore
1056 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1057 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1058 ** bug fix: dynamic variable and thread start related gc lossage
1059 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1061 * fixed some bugs revealed by Paul Dietz' test suite:
1062 ** SUBTYPEP is slightly more accurate on heinously complicated
1063 CONS types where some of the members have uncertain (in the
1064 NIL, NIL sense) type relationships to each other.
1066 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1067 no longer purified unless :PURIFY T is explicitly specified.
1068 ** Non-purified cores are significantly smaller than before
1070 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1071 * new feature: timers based on Zach Beane's excellent timer package
1072 * added support for the following external formats: koi8-u,
1073 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1074 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1075 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1076 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1077 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1078 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1079 * incompatible change: a threaded SBCL will no longer revert to
1080 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1081 * bug fix: interrupts are disabled until startup is complete; no
1082 more sigsegvs when receiving a signal to soon
1083 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1085 * bug fix: add a workaround for the memory randomization features in
1086 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1087 workaround will only be in effect on systems with the proc filesystem
1089 * bug fix: printing objects of type HASH-TABLE signals a
1090 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1092 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1093 have been running for over 50 days. (reported by Gilbert Baumann)
1094 * bug fix: the logic for getting names of functions gets less
1095 confused when confronded with alternate-metaclass
1096 funcallable-instances. (reported by Cyrus Harmon)
1097 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1098 consistent, even on internal alternate-metaclass objects.
1099 * bug fix: flush closure information collected by physical
1100 environment analysis prepass before the main pass. (bug reported
1102 * bug fix: compiler pack phase does not modify a hashtable, which is
1103 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1104 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1105 may be expanded. (reported by Alexander <malishev> on c.l.l)
1106 * bug fix: no more style-warnings when compiling code using LAST
1107 under high-SPEED low-SPACE settings. (reported by David Wallin
1109 * fixed some incompatibilities between SBCL's MOP and the MOP
1111 ** the METAOBJECT class is now implemented;
1112 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1113 STANDARD-OBJECT, as required;
1114 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1115 now compatible (as required by VALIDATE-SUPERCLASS); there
1116 remains a consistency requirement over the presence or absence
1117 of the FUNCTION class in the superclass list of finalized
1118 classes; see the manual for more details;
1119 ** the :AROUND method for COMPUTE-SLOTS on
1120 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1121 requested slot ordering.
1123 ** bug fix: parent thread now can be gc'ed even with a live
1125 ** bug fix: binding a special with PROGV to no value is not
1126 the same as that symbol not having been bound (thanks to
1128 ** bug fix: binding specials is thread safe (thanks to
1130 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1131 and SB-SPROF do not die with 'no handler for signal XX in
1132 interrupt_handle_now(..)' anymore
1133 ** bug fix: WITH-TIMEOUT works with multiple running threads
1134 * fixed some bugs revealed by Paul Dietz' test suite:
1135 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1136 the :method-class keyword argument.
1138 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1139 * new port: the Solaris operating system on x86 processors is now
1140 mostly supported, though some rough edges in the environment
1141 remain. (thanks to Daisuke Homma)
1142 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1143 space for its heap. (thanks to Thiemo Seufer)
1144 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1145 closures. (thanks to Thiemo Seufer)
1146 * minor incompatible change: eof selects abort in the debugger.
1147 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1148 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1150 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1151 deprecated, and will go away in a future revision of SBCL.
1152 * minor incompatible change: GC-ON and GC-OFF are no longer
1153 implemented with a counter, it does not matter how many times gc
1154 is switched on or off
1155 * bug fix: discriminating functions for generic function classes
1156 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1157 make invalid assumptions about method precedence order. (reported
1159 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1160 platforms. (thanks to Thiemo Seufer)
1161 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1162 * bug fix: correct pathname printing with printer escaping is on.
1163 (thanks to Kevin Reid)
1164 * bug fix: complex VOP definitions in "user-space" no longer trigger
1165 package locks. (reported by Zach Beane)
1166 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1167 for funcallable-instance functions, and is no different from
1169 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1170 characters in the current readtable. (reported by Nicholas Neuss)
1171 * bug fix: SUBTYPEP on various CONS types returns more a more
1172 accurate acknowledgment of its certainty.
1173 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1174 lists in some cases. This partially fixes bug 384.
1175 * flush all standard streams before prompting in the REPL and the
1177 * bug fix: signal handling and triggering gc do not conflict
1178 directly anymore, in particular a high frequency sb-sprof does
1179 not prevent gc from running
1180 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1181 approximation for timezone and DST information between the
1182 universal time epoch and the smallest negative 32-bit time_t.
1183 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1184 year 1899 so long as the encoded time is non-negative.
1185 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1186 the correct answer for non-constant rotate amounts when performing
1187 an inline 32-bit rotation.
1189 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1190 there is only one thread in the session
1191 ** bug fix: memory leak for streams created in one thread and
1192 written to in another
1193 ** bug fix: lockup when compiled with gcc4
1194 ** bug fix: race that allows the gc to be triggered when gc is
1196 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1197 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1199 ** bug fix: finished threads are gc'ed properly
1200 * fixed some bugs revealed by Paul Dietz' test suite:
1201 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1202 has arguments with a different set of applicable methods from
1203 the orignal arguments.
1204 ** The type error thrown by MAP now has a correct expected-type
1206 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1207 name a compiled function.
1208 ** Three MISC tests where a large bignum was improperly coerced to
1209 a float that couldn't represent that bignum during type
1210 derivation were fixed.
1211 ** SUBTYPEP can now handle types involving the negation of a
1212 list-form FUNCTION type.
1213 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1214 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1215 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1217 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1218 * New feature: Experimental support for bivalent streams: streams
1219 opened with :element-type :default now allow character and binary
1220 (unsigned-byte 8) I/O
1221 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1222 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1224 * bug fix: no more highly sporadic "couldn't check whether ~S is
1225 readable" when reading a stream and an interrupt hits in the middle
1226 of a select system call
1227 * compiler better recognizes complex arrays (reported by Eduardo
1229 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1230 for variables on SBCL built with linkage-tables. (reported by Luis
1232 * various error reporting improvements.
1233 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1234 (Thanks to Thiemo Seufer)
1235 * bug fix: nested reader invokations work correctly
1236 * bug fix: it is possible to have simultaneous references to foreign
1237 code and foreign data with the same name.
1239 ** added x86-64 support
1240 ** incompatible change: the threading api now works with thread
1241 objects instead of thread ids
1242 ** bug fix: threads are protected from signals and interruption when
1243 starting up or going down
1244 ** bug fix: a race where an exiting thread could lose its stack to gc
1245 ** bug fix: don't halt on infinite error in threads if possible
1246 ** fixed numerous gc deadlocks introduced in the pthread merge
1247 ** bug fix: fixed thread safety issues in read and print
1248 ** bug fix: debugger doesn't hang on session lock if interrupted at
1249 an inappropriate moment
1250 ** bug fix: run-program is now thread safe(r)
1251 ** bug fix: inner with-recursive-lock no longer releases the mutex
1252 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1253 * fixed some bugs revealed by Paul Dietz' test suite:
1254 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1255 EXPECTED-TYPE slots filled.
1256 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1258 ** MAKE-CONDITION accepts classes as type-designators.
1259 ** COMPILE may never return NIL.
1260 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1261 range before calling Unix time functions
1263 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1264 * numerous signal handling fixes to increase stability
1265 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1266 * minor incompatible change: we now correctly canonize default
1267 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1268 as per the MOP, rather than the historical (INITARG INITFUNCTION
1269 INITFORM). (reported by Bruno Haible)
1270 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1271 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1272 for more information.
1273 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1274 pathname is a directory pathname.
1275 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1276 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1278 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1279 :METACLASS argument in addition to a class name. (reported by
1280 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1281 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1282 (reported by Bruno Haible; patch from Gerd Moellmann)
1283 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1285 * bug fix: more cleanups to the floating point exception handling on
1286 x86-64 (thanks to James Knight)
1287 * bug fix: the compiler does not try to constant fold calls of
1288 COERCE and BIT-* functions when they return freshly constructed
1289 objects. (reported by Nikodemus Siivola and Paul Dietz)
1290 * optimization: improved the MIPS versions of generic (in the
1291 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1292 * optimization: direct conversion of (unsigned-byte 32) to floats on
1293 the PowerPC platform.
1294 * optimization: structure instances with raw slots now use less
1295 memory, and probably show better memory locality. (thanks to
1297 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1298 are treated as requests for stack allocation on the x86-64,
1299 Alpha-32, PPC and SPARC platforms.
1300 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1301 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1302 (thanks to Tony Martinez)
1304 ** gcing a dead thread can no longer lead to lockups
1305 ** threads block signals until they are set up properly
1306 ** errno is no longer shared by threads
1307 ** interrupt-thread restores the eflags register on x86
1308 ** fixed some lockups due to gc/thread interaction
1309 * fixed some bugs revealed by Paul Dietz' test suite:
1310 ** invalid dotted lists no longer raise a read error when
1311 *READ-SUPPRESS* is T
1312 ** signal an error if a symbol that names a declaration is used
1313 as the name of a type, or vice versa
1314 ** allow using the (declare (typespec var*)) abbreviation for
1315 (declare (type typespec var*)) with all type specifiers
1316 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1317 ** always return NIL from PPRINT-POP when OBJECT is NIL
1318 ** don't signal errors when pretty-printing malformed LABELS,
1319 FLET or MACROLET forms
1320 ** declarations in a DOLIST body are also in scope for the
1322 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1324 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1327 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1328 * fixed cross-compiler leakages that prevented building a 32-bit
1329 target with a 64-bit host compiler.
1330 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1331 opened with :IF-EXISTS :OVERWRITE.
1332 * fixed bug 281: error for an invalid qualifier in a short-form method
1333 combination method is not signalled until the faulty method is called.
1334 * bug fix: iteration variable type inferrer failed to deal with open
1335 intervals. (reported by Alan Shields)
1336 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1337 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1338 or more runtime options were provided to the sbcl binary.
1339 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1340 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1341 * optimization: there's now a fast-path for fixnum arguments in the
1342 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1343 * optimization: the code generated on x86-64 is more compact thanks
1344 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1345 * bug fix: floating-point exception handling now partly works on
1346 x86-64. (Thanks to James Knight)
1347 * improvement to the MIPS backend from Thiemo Seufer: C-style
1348 64-bit long long arguments and return values to alien functions
1350 * fixed some bugs revealed by Paul Dietz' test suite:
1351 ** the type-error signalled from WARN has a filled-in DATUM slot.
1352 ** the type-error required when a stream is not associated with
1353 a file has the stream as its datum.
1354 ** type-errors on single-floats on x86-64 no longer have
1355 :INVALID-OBJECT as the datum
1356 ** the type-errors signalled for invalid function names now have
1357 a correct expected type
1358 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1359 for structures defined with a :TYPE. Documentation strings for
1360 typed structures are no longer immediately discarded
1361 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1362 broadcast streams and synonym streams. FILE-LENGTH now also works
1363 on broadcast streams.
1365 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1366 * incompatible change: the --noprogrammer option, deprecated since
1367 version 0.7.5, has been removed. Please use the equivalent
1368 --disable-debugger option instead.
1369 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1371 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1372 inoperational for a while now) has been completely removed.
1373 * null lexical environments are now printed as #<NULL-LEXENV>,
1374 significantly reducing the amount of clutter in typical backtraces.
1375 * documentation on weak pointers, finalization, and after GC hooks
1376 has been added to the manual.
1377 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1378 as other specialized array types, is much faster. SUBSEQ and
1379 COPY-SEQ on such arrays have also been sped up.
1380 * optimization: EQL is now more efficient when at least other argument
1381 is known to be of type (OR FIXNUM (NOT NUMBER)).
1382 * fixed bug: the runtime is now less vulnerable to changes in the
1383 size of the SBCL object on OS X, and virtual memory is reserved for
1384 all spaces that need to be at a fixed address.
1385 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1386 * fixed bug: finalizers and after GC hooks that cause consing are now
1388 * fixed bug: compiler error messages and summaries are now printed to
1389 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1390 * fixed inference of the upper bound of an iteration variable.
1391 (reported by Rajat Datta).
1392 * fixed bug 211e: calling local functions with duplicated constant
1393 keyword argument no longer causes a bogus style warning about an
1395 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1396 ftype declaration to be disregarded. (reported by Dave Roberts)
1397 * fixed bug 373: caused by erronous compilation of references to alien
1398 variables in the runtime on ppc/darwin.
1399 * fixed bug 376: CONJUGATE type deriver.
1400 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1401 (reported by Baughn on #lisp)
1402 * a cross-compiler bug on non-x86 platforms has been identified and
1403 fixed. (thanks to Bruno Haible)
1404 * improvements to the MIPS runtime code for increased stability.
1405 (thanks to Thiemo Seufer)
1406 * increased the maximimum compact environment size to allow
1407 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1408 * partial workaround for bug 135: don't name the function we're
1409 calling for hairy cases of EVAL, so as not to accumulate one environment
1410 entry per call to EVAL. (reported by Kevin Reid)
1411 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1412 * optimization: type testing for non-vector arrays should be faster.
1413 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1415 * bug fix: setting 31st element of a bit vector to zero did not work
1417 * bug fix: redefining a class definition which failed due to a
1418 previous accessor / function clash now works (but see BUGS entry
1419 #380 for more problems in this area). (thanks to Zach Beane)
1420 * the long form of DEFINE-METHOD-COMBINATION disables method group
1421 checking when given a single method group with pattern *. (thanks
1423 * on x86 compiler supports stack allocation of results of simple
1424 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1425 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1426 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1427 * fixed some bugs related to Unicode integration:
1428 ** the restarts for recovering from input and output encoding
1429 errors only appear when there is in fact such an error to
1431 * fixed some bugs revealed by Paul Dietz' test suite:
1432 ** MISC.549 and similar: late transformation of unsafe type
1433 assertions into derived types caused unexpected code
1435 ** SCALE-FLOAT type deriver is less wrong.
1436 ** type derivers for EXP, LOG and similar functions compute result
1437 types for complex arguments better.
1438 ** (MISC.563) CONJUGATE type deriver works for very restricted
1440 ** out-of-line type testers for character strings are available.
1441 ** EQUAL compiler transform understands specialness of objects
1443 ** accessing double-floats stored on the stack now works on x86-64.
1444 ** debugger internals could sometimes create invalid lispobjs,
1445 resulting in GC crashes.
1446 ** MISC.548: type check weakening can convert required type into
1448 ** initialization forms for bindings are not in scope of free special
1451 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1452 * incompatible change: thread support for non-NPTL systems has
1453 been removed - locking is buggy and unreliable. A threaded
1454 SBCL build will now warn at startup and refuse to create
1455 new threads, unless futex support is detected in the kernel
1456 * incompatible change: the top level REPL now has only an ABORT
1457 restart associated with it, not TOPLEVEL and ABORT as it used to.
1458 TOP and TOPLEVEL are now available as debugger commands for
1459 returning to the top level.
1460 * incompatible change: forms evaluated in the REPL now use the
1461 global optimization policy.
1462 * incompatible change: user- and system-initialization files are
1463 no longer processed with LOAD, but by READ and EVAL; hence the
1464 global optimization policy, startup package, readtable, etc,
1466 * The .fasl file format number has been incremented because of
1467 various incompatible changes.
1468 * internal entry point details and argument counts no longer appear
1469 in backtraces unless explicitly requested by setting
1470 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1471 * built-in and standard functions no longer have names like "top
1472 level local call to FOO".
1473 * fixed bug 32: functions defined in non-null lexical environments
1474 now have more legible printed representation
1475 * fixed bug 33: functions defined in non-null lexical environemnts
1476 are now more amenable to inspection by INSPECT.
1477 * workaround for bug 354: XEPs no longer appear in backtraces unless
1478 explicitly requested.
1479 * fixed bug: receiving the signal which results from attempting to
1480 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1481 SIGBUS on other *BSDs) on architectures where the C stack is also
1482 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1483 to debugger confusion or wild instability. (reported by Cheuksan
1485 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1486 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1487 (reported by Lutz Euler)
1488 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1489 compiled code. (reported by Paul Dietz)
1490 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1491 specialized parameter to a method to something that is not TYPEP
1492 the specializer is now possible.
1493 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1494 face of package deletion.
1495 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1496 pathname. (reported by David Tolpin on c.l.l)
1497 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1498 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1499 STRING-STREAM and FILE-STREAM, respectively.
1500 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1501 than previously; in addition, its test suite is now run on build.
1502 (thanks to Luke Gorrie)
1503 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1504 * optimization: added a immediate representation for single-floats
1506 * fixed some bugs revealed by Paul Dietz' test suite:
1507 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1508 ** attempting to create a package with a colliding nickname causes
1509 correctable errors to be signalled.
1510 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1511 ** several x86-64 backend bugs related to sign-extension of immediate
1514 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1515 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1516 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1518 * fixed loading of multiply forward-referenced layouts.
1519 (thanks to Cheuksan Wang)
1520 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1521 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1522 (see COMPILE-FILE documentation for details of :PRINT option)
1523 * fixed bugs 19 and 317: fixed-format floating point printing is
1524 more accurate. This also fixes a bug reported by Adam Warner
1525 related to the ~@F format directive.
1526 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1528 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1529 dispatch table if the from-char is a dispatch macro character.
1530 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1531 by the wordsize no longer ignore the last word. (reported by Lutz
1533 * fixed bug in type checking of dynamic-extent variables. (reported
1535 * optimization: sequence traversal functions use their freedom to
1536 coerce function designators to functions.
1537 * optimization: code with many calls to CLOS methods specialized on
1538 CLOS classes has had redundant type checks removed.
1539 * optimization: added declarations to speed up operations that access
1540 the internal character database (for example STRING-UPCASE)
1541 * optimization: comparison operations between floats and sufficiently small
1542 fixnums no longer create extra rationals
1543 * fixed some bugs related to Unicode integration:
1544 ** portions of multibyte characters at the end of buffers for
1545 character-based file input are correctly transferred to the
1546 start of the buffer at the next read.
1547 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1548 passing it through to OPEN.
1549 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1550 argument given to internal calls to OPEN.
1551 ** the built-in comment readers (introduced by character sequences
1552 ";" and "#|") are more forgiving to encoding errors; they will
1553 STYLE-WARN and then attempt to resync the stream at a character
1554 boundary. (thanks to Teemu Kalvas)
1555 * fixed some bugs in the x86-64 port:
1556 ** Negative short int return values from c-calls are sign-extended
1558 ** The stack is aligned to 16-bytes for c-calls, as required by
1559 the ABI. (thanks to Cheuksan Wang)
1560 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1561 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1562 for enums. (thanks to Vincent Arkesteijn)
1563 ** Multiple small optimizations and bugfixes for floating point
1565 * fixed some bugs revealed by Paul Dietz' test suite:
1566 ** Space, Tab, Linefeed, Return and Page have the invalid
1567 secondary constituent character trait.
1568 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1570 ** WITH-INPUT-FROM-STRING should only update the index place on
1572 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1573 works more reliably.
1574 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1575 with one-dimensional arrays with fill pointers.
1576 ** TYPE-OF failed on a complex with an integer realpart and a
1578 ** compiler failure during type inference for the code of form
1579 (IF (EQL X (THE ...)) ...) (MISC.535).
1581 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1582 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1583 platform. The port supports 61-bit fixnums, large memory spaces
1584 and reloading shared object files.
1585 * enhancement: saving cores with foreign code loaded is now
1586 supported on ppc/Darwin in addition to the previously supported
1588 * enhancement: the statistical profiler now walks deeper into the
1589 call stack for more meaningful call-graphs and accrued time
1590 reports (x86/x86-64 only). It also now reports time spent in
1592 * enhancement: it is now possible to trace most individual methods
1593 of a generic function in addition to tracing the generic function
1595 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1597 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1598 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1599 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1600 produces an error. (thanks to Vincent Arkesteijn)
1601 * bug fix: NAMESTRING on pathnames with :WILD components in their
1602 directories works correctly. (thanks to Artem V. Andreev)
1603 * fixed bug 125: compiler preserves identity of closures. (reported
1605 * bug fixed: functions with &REST arguments sometimes failed with
1606 "Undefined function" when compiled with (DEBUG 3). (reported by
1608 * bug fix: overflow during compiling of setting element of a bit
1609 vector with constant index and value. (reported by Timmy Douglas)
1610 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1611 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1612 * on x86 compiler partially supports stack allocation of dynamic-extent
1614 * GO and RETURN-FROM do not check the extent of their exit points
1615 when compiled with SAFETY 0.
1616 * fixed some bugs related to Unicode integration:
1617 ** encoding and decoding errors are now much more robustly
1618 handled; it should now be possible to recover even from invalid
1619 input or output to the terminal. (thanks to Teemu Kalvas)
1620 ** provided a first cut at implementing STRING-TO-OCTETS and
1621 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1622 ** altered the SB-MD5 contributed module slightly, changing the
1623 interface just enough for it to be supportable for builds where
1624 lisp characters are not eight bits.
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** the FORMATTER-generated functions for ~V[ conditionals require
1627 the correct number of arguments.
1628 ** READ-FROM-STRING returns the mandated second value when applied
1629 to displaced strings.
1630 ** the #\Rubout and #\Backspace characters are treated as invalid
1631 constituent characters by the tokenizer.
1633 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1634 * new feature: reloading changed shared object files with
1635 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1636 * new feature: references to foreign variables and functions
1637 can now be compiled and loaded before the corresponding shared
1638 object file is loaded, as long as the foreign definitions are
1639 available at runtime.
1640 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1641 supported. (thanks to Dan Debertin)
1642 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1643 just keywords. (thanks to Vincent Arkesteijn)
1644 * fixed bug #331: structure-class instances corresponding to
1645 DEFSTRUCT forms are now created eagerly.
1646 * fixed bug #345: backtraces from calls to undefined functions work
1647 on x86 as well. Related bug #61 is now also partially fixed on x86
1648 (backtraces from throws to unknown catch tags.)
1649 * bug fix: lambda-list parsing is now stricter vrt. order and number
1650 of lambda-list keywords.
1651 * bug fix: as specified by AMOP, an error is signalled if a
1652 class-option appears multiple times in a DEFCLASS form. (reported
1654 * bug fix: RANDOM can be compiled when the compiler derives the type
1655 of its numeric argument as a disjoint set of small integers.
1656 (reported by Paul Dietz)
1657 * bug fix: starting a core saved with shared objects loaded when
1658 those objects are not available no longer causes threaded SBCL to
1659 hang. (reported by Sean Ross)
1660 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1662 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1663 argument, which must always be NIL. (reported by Kalle Niemitalo)
1664 * bug fix: printing 1.0d+23 no longer results in an error.
1665 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1666 * bug fix: structure slot setters preserve evaluation order. (thanks
1668 * fixed some bugs related to Unicode integration:
1669 ** RUN-PROGRAM can allow its child to take input from a Lisp
1670 stream. (reported by Stefan Scholl)
1671 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1672 locales. (reported by Ken Causey)
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** INCF, DECF and REMF evaluate their place form as specified in
1676 ** FORMATTER expands ~{ ~} iteration directives with V or #
1677 parameters correctly.
1678 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1679 consequent uses no arguments correctly.
1680 ** the system has a partial understanding of the (COMPLEX RATIO)
1682 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1685 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1686 * new feature: a build-time option (controlled by the :SB-UNICODE
1687 keyword feature, enabled by default) for building the system with
1688 support for the entire 21-bit character space defined by the
1690 * new feature: the system now has rudimentary external-format
1691 support; the primary user-visible change at this time is that
1692 characters with the high bit set (such as lower-case-e-acute) will
1693 print correctly to a terminal in a UTF-8 environment.
1694 * minor incompatible change: BASE-CHAR no longer names a class;
1695 however, CHARACTER continues to do so, as required by ANSI.
1696 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1697 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1698 has been moved to the SB-EXT package (temporarily re-exported from
1700 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1701 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1703 * fixed bug #351: better error-handlind and reporting for malformed
1705 * fixed bug #350: bignum-printing is now more memory-efficient,
1706 allowing printing of very large bignums, eg. (expt 2 10000000).
1707 (reported by Bruno Haible)
1708 * fixed bug #302: better primitive-type selection for intersection
1710 * fixed bug #308: non-graphic characters now all have names, as
1711 required. (reported by Bruno Haible)
1712 * bug fix: redefining a class with different superclasses now correctly
1713 removes it from the direct-subclasses of its previous superclasses.
1714 (reported by David Morse)
1715 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1716 new value now works. (reported by Bruno Haible)
1717 * bug fix: correct canonicalization of multiple non-standard slot
1718 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1719 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1720 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1721 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1722 now exists, an signals an error.
1723 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1724 during macroexpansion for non-iteration variables as well. (reported
1725 by Bruno Haible for CMUCL)
1726 * bug fix: Cyclic structures and unprintable objects in compiler
1727 messages no longer cause errors. (reported by Bruno Haible)
1728 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1729 and READ-FROM-STRING all now return a primary value of NIL if
1730 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1731 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1732 (reported by Bruno Haible for CMUCL)
1733 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1734 do not designate a condition. (reported by Bruno Haible for
1736 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1737 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1738 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1739 permitting resolution in favour of any of the conflicting symbols.
1740 (reported by Bruno Haible for CMUCL)
1741 * bug fix: EQUAL compiler optimizations is less aggressive on
1742 strings which can potentially compare true despite having distinct
1743 specialized array element types.
1744 * bug fix: unit enumerations can be defined without dividing by
1745 zero. (thanks to Vincent Arkesteijn)
1746 * FORMAT compile-time argument count checking has been enhanced.
1747 (report from Bruno Haible for CMUCL)
1748 * a partial workaround for the bug 262: the compiler does not try to
1749 inline-expand a local function doing RETURN-FROM from a deleted
1750 BLOCK. (thanks to Peter Denno for the bug report and to David
1751 Wragg for the simple test case)
1752 * fixed some bugs revealed by Paul Dietz' test suite:
1753 ** RENAME-PACKAGE allows all package designators as new package
1755 ** constraint propagation and lambda variable substitution are
1756 more cautious in dealing with partially deleted code.
1757 ** compiler sometimes lost reoptimization passes.
1758 ** CERROR, given a condition as condition designator, passes its
1759 remaining arguments to the continue format control without
1761 ** Case-altering FORMAT directives work correctly on non-ASCII
1763 ** The REAL type specifier handles bounds outside the floating
1764 point ranges without signalling FLOATING-POINT-OVERFLOW.
1765 ** Functions with IR1-transformations can create intercomponent
1766 references to global functions.
1767 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1769 ** FORMAT ~R treats a nil value for its first parameter correctly.
1771 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1772 * enhancement: saving cores with foreign code loaded is now
1773 supported on x86/NetBSD and sparc/Linux in addition to the previously
1774 supported platforms.
1775 * bug fix: on some platforms repeated installations caused multiple
1776 copies of HTML documentation to be installed -- should not happen
1777 any more. (reported by Stefan Scholl)
1778 * bug fix: parsing self-recursive alien record types multiple times
1779 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1780 original patch by Helmut Eller for CMUCL)
1781 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1782 (thanks to Richard Kreuter)
1783 * bug fix: defining classes whose accessors are methods on existing
1784 generic functions in other (locked) packages no longer signals
1785 bogus package lock violations. (reported by François-René Rideau)
1786 * bug fix: special variables as DEFMETHOD parameters no longer have
1787 associated bogus type declarations. (reported by David Wragg and
1789 * bug fix: read-write consistency on streams of element-type
1790 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1791 * bug fix: redefiniton of the only method of a generic function with
1792 no DEFGENERIC no longer emits a full WARNING. In addition,
1793 redefinition of generic functions with no DEFGENERIC to an
1794 incompatible lambda list now signals an error. (thanks to Zach
1796 * bug fix: DEFGENERIC now works even when there's a function of the
1797 same name in an enclosing lexical environment. (thanks to Zach
1799 * fixed compiler failure, caused by instrumenting code during
1800 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1801 * optimization: added loop analysis and improved register allocation
1802 * fixed some bugs revealed by Paul Dietz' test suite:
1803 ** POSITION on displaced vectors with non-zero displacement
1804 returns the right answer.
1805 ** (SIMPLE-STRING) is a valid type specifier for sequence
1807 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1809 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1810 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1812 ** The justification version of the FORMAT ~< directive treats
1813 non-zero minpad parameter correctly.
1815 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1816 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1817 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1818 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1819 the supported interface.
1820 * new feature: Single-stepping of code compiled with DEBUG 2 or
1821 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1822 * new feature: saving cores with foreign code loaded is now
1823 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1824 Timothy Moore's work for CMUCL)
1825 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1826 parameters to * instead of NIL if no initform is supplied.
1827 (reported by Johan Bockgaard)
1828 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1829 correctly when FUNCALL appears as the car of the form. Note:
1830 despite this FUNCALL forms are not currently subject to
1831 compiler-macro expansion. (port of Raymond Toy's fix for the
1832 same from CMUCL, reported by Johan Bockgaard)
1833 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1834 (thanks for Teemu Kalvas)
1835 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1836 case instead of terminating the iteration (thanks for Julian
1837 Squires, Sean Champ and Raymond Toy)
1838 * bug fix: incorrect expansion of defgeneric that caused a style
1839 warning. (thanks for Zach Beane)
1840 * bug fix: slot accessor effective method computation works properly
1841 for classes with multiple non-standard applicable methods on
1842 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1843 * on x86 compiler supports stack allocation of results of LIST and
1844 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1845 CMUCL implementation by Gerd Moellmann)
1846 * fixed some bugs revealed by Paul Dietz' test suite:
1847 ** FORMAT strings with both the ~<~:;~> form of the justification
1848 directive and pretty-printing directives cause an error.
1850 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1851 * incompatible change: the internal functions
1852 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1853 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1854 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1855 instead of the old functions.
1856 * new feature: on platforms where "dladdr" is available foreign
1857 function names now appear in backtraces. (based on Helmut Eller's
1859 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1860 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1862 * documentation: ASDF manual is now distributed with SBCL.
1863 * bug fix: SBCL can now load its contributed modules with REQUIRE
1864 even if the system-provided entries have been removed from
1865 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1867 * bug fix: backtraces involving undefined functions or assembly
1868 routines are more informative. (thanks to Brian Downing)
1869 * bug fix: mutually referent alien structures now work correctly.
1870 (reported by Rick Taube)
1871 * bug fix: structures defined by WITH-ALIEN can be referred to
1872 within other definitions in the same WITH-ALIEN.
1873 * bug fix: division operators (MOD, TRUNCATE and the like) with
1874 constant zero divisors and integer dividends no longer generate
1876 * bug fix: provide default methods for INPUT-STREAM-P and
1877 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1878 * bug fix: improve the detection and resolution of MOP metacycles
1879 (where effective method computation depends on the generic function
1880 having its effective method computed). (reported by Bruno Haible)
1881 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1882 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1883 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1884 * bug fix: pathnames with relative directory components can be
1885 represented relative to default pathnames.
1886 * optimization: in taking the GCD of bignums, reduce the two bignums
1887 to approximately the same size (using Euclid's algorithm) before
1888 applying the more sophisticated binary GCD. (thanks to Juho
1890 * optimization: COUNT on bitvectors now operates word-at-a-time.
1891 * optimization: ASH with a positive, but not necessarily constant,
1892 (leftwards) shift, when in a modular context, is compiled to a
1894 * fixed some bugs revealed by Paul Dietz' test suite:
1895 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1896 if the corresponding argument is NIL.
1897 ** FORMAT directives accepting a minpad parameter treat negative
1899 ** literal commas as character parameters to FORMAT directives are
1901 ** literal spaces directly after ~<Newline> directives within a
1902 format-logical-block (~:< ~@:>) do not induce :FILL-style
1903 conditional newlines.
1904 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1905 as specified: it no longer includes conditional newlines).
1906 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1908 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1909 :PER-LINE-PREFIX argument does not evaluate to a string.
1911 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1912 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1913 the manual for details; for now, package locks can be disabled by
1914 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1915 if no major problems are found then it is likely that they will be
1916 compiled in unconditionally.
1917 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1918 unsupported operators on all platforms. To load a shared library
1919 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1920 object file, link it into a shared library outside of SBCL and
1921 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1923 * bug workaround: ROOM T and the GC were not working together
1924 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1925 reporting facility aren't preserved. That reporting has been
1926 disabled, so now until and unless someone figures out how to make it
1927 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1928 * minor incompatible change: as threatened around sbcl-0.8.0, the
1929 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1930 The symbols are also exported from SB-PCL for backwards
1931 compatibility, but more so than before SB-PCL should be treated as
1932 an implementation-internal package.
1933 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1935 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1936 It is known as of this release not to work on the Alpha, however.
1937 * fixed bug #167: errors signalled due to illegal syntax in method
1938 bodies are now more legible.
1939 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1940 designators and can hence be used with TYPEP.
1941 * fixed bug #333: CHECK-TYPE now ensures that the type error
1942 signalled, if any, has the right object to be accessed by
1943 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1944 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1945 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1947 * fixed bug #334: programmatic addition of slots using specialized
1948 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1949 and :ALLOCATION :CLASS slots.
1950 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1951 float exponents. (rereported by Peter Seibel)
1952 * fixed a bug: #\Space (and other whitespace characters) are no
1953 longer considered to be macro characters in standard syntax by
1954 GET-MACRO-CHARACTER.
1955 * fixed bug: initialization of condition class metaobjects no longer
1956 causes an instance of the condition to be created. (reported by
1958 * fixed bug: it is now possible to have slots such that
1959 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1960 description is neither :INSTANCE nor :CLASS.
1961 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1962 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1963 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1964 non-standard SLOT-DEFINITION-ALLOCATION values.
1965 * fixed bugs: various race conditions handling exiting threads.
1966 CL-PPCRE's thread torture test now passes.
1967 * fixed arguably-a-bug: GC time proportional to number of threads in
1968 system even when most of them are idle
1969 * optimization: improved performance of BIT and SBIT on bit-vectors.
1970 * .fasl file incompatibility: The fasl file version number has
1971 been incremented because of changes associated with package locks.
1973 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1974 * minor incompatible change: the system no longer provides
1975 optimization hints (conditions which are TYPEP
1976 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1977 mode; these hints are still emitted when higher SPEED optimization
1979 * new contrib module: a sampling profiler (profiling by statistical
1980 sampling, rather than by instrumenting functions) is available as
1981 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1982 * the behaviour of the standard function ED is now customizeable by
1983 third parties through a hook variable: see ED's documentation
1984 string for information on the protocol.
1985 * the compiler no longer emits efficiency notes for (FUNCALL X)
1986 when the type of X is uncertain under default optimization
1988 * fixed bug 276: mutating a binding of a specialized parameter to a
1989 method to something that is not TYPEP the specializer is now
1991 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1992 exhibit the required behaviour on the x86 platform. (thanks to
1993 Peter van Eynde, Eric Marsden and Bruno Haible)
1994 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1995 even for difficult arguments. (reported by Peter Graves)
1996 * fixed bug 141a: the backquote printer now descends quoted
1998 * fixed another bug in backquote printing: no more destructive
1999 modification of the form's list structure. (reported by Brian
2001 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2002 move between its address being taken and the call to
2003 interrupt_thread, fixing a crashing race condition.
2004 * the SB-POSIX contrib implementation has been adjusted so that it
2005 no longer exhibits ridiculously poor performance when constructing
2006 instances corresponding to C structs.
2008 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2009 * minor incompatible change: the sb-grovel contrib now treats C
2010 structures as alien (in the sense of SB-ALIEN) objects rather than
2011 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2012 has implications for memory management of client code
2013 (sb-grovel-returned objects must now be manually managed) and for
2014 type safety (alien objects now have full types).
2015 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2016 used to control emission of compiler diagnostics, rather than the
2017 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2018 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2019 quality should be considered deprecated.
2020 * install.sh now installs the user manual as well
2021 * (not quite a new documentable feature, but worth considering in
2022 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2023 beginnings of a semantically meaningful condition hierarchy is
2024 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2026 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2027 designator as the defaults argument.
2028 * fixed bug: Displaced arrays whose displaced-to array has become
2029 too small now cause ARRAY-DIMENSION to signal an error, providing
2030 for safer bounds-checking. (reported by Bruno Haible)
2031 * fixed bug: DEFCLASS slot definitions with identical :READER and
2032 :WRITER names now signal a reasonable error. (reported by Thomas
2034 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2036 * fixed bug: Invalid slot specification errors now print correctly.
2037 (thanks to Zach Beane)
2038 * fixed bug 320: Shared to local slot value transfers in class
2039 redefinitions now happen corrently from superclasses as
2040 well. (reported by Bruno Haible)
2041 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2043 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2044 improper lists correctly. (reported by Bruno Haible)
2045 * fixed bug 313: source-transform for <fun-name> was erroneously
2046 applied to a call of a value of a variable with name <fun-name>.
2047 (reported by Antonio Menezes Leitao)
2048 * fixed bug 307: The obsolete instance protocol ensures that
2049 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2050 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2051 remove dead unknown-values globs from the middle of the stack.
2052 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2053 easier to resume long computations after using *BREAK-ON-SIGNALS*
2054 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2055 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2056 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2057 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2058 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2059 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2061 * fixed a (fairly theoretical) bug in string printing: if
2062 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2063 does not have array-element-type equal to the most general string
2065 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2066 function name. (thanks to Nikodemus Siivola)
2067 * fixed bug: compile-time format string checker failed on
2068 non-closed ~{. (reported by Thomas F Burdick)
2069 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2070 don't warn on legal loop code involving more than one aggregate
2071 boolean. (thanks to Nikodemus Siivola)
2072 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2073 in keyword position to MAKE-PACKAGE was not being reported as
2075 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2076 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2078 * fixed bugs 280 and 312: the checking for multiple definitions in a
2079 file is less likely to become confused by uses of inline
2081 * fixed bug: the #S reader macro performs the keyword coercion
2082 specified for slot names. (reported by Kalle Niemitalo)
2083 * fixed bug: lambda lists may contain symbols whose names start with
2084 & but are not lambda-list-keywords; their occurrence triggers a
2086 * fixed bug 321: define-method-combination argument lambda lists do
2087 not cause invalid code to be generated when &optional or &aux
2088 variables have default values. (reported by Bruno Haible)
2089 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2090 have CLOS classes; this appears also to have fixed reported
2091 instability in stack exhaustion detection.
2092 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2093 MUFFLE-WARNING when no associated restart is present is now
2095 * optimization: rearranged the expansion of various defining macros
2096 so that each expands into only one top-level form in a
2097 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2099 * optimization: used a previously little-used slot in symbols to
2100 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2102 * fixed some bugs revealed by Paul Dietz' test suite:
2103 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2105 ** Fixed an optimization bug related to inheritance of initforms
2106 from local to shared slots.
2107 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2108 ** CAST optimizer forgot to flush argument derived type.
2109 ** print/read consistency on floats is now orders of magnitude
2110 more likely. (thanks also to Bruno Haible for a similar report
2112 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2113 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2114 for objects of type REAL. Make it so.
2115 ** SXHASH is specified (infelicitously) to respect similarity,
2116 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2117 it so. (thanks to Markus Ziegler)
2118 ** on the Alpha, the compiler succeeds in compiling functions
2119 returning a known number of arguments greater than 63.
2120 ** fixed handling of invalid NIL arguments in keyword position
2122 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2124 ** provided a readably-printable representation for RANDOM-STATE
2126 ** ensured that pathnames, where they have a namestring, always
2127 print using #P"..." syntax.
2129 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2130 * Support for the forthcoming 2.0 version of the NetBSD kernel
2131 running on x86 hardware has been added. (thanks to Perry
2132 E. Metzger most immediately, and others for their past work)
2133 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2134 current release version; SBCL's OpenBSD support had been broken
2135 since about the time of OpenBSD's switch to ELF binary format.)
2136 * [placeholder for DX summary]
2137 ** user code with &REST lists declared dynamic-extent, under high
2138 speed or space and low safety and debug optimization policy.
2139 * The manual has been converted to Texinfo format and the debugger
2140 chapter from the cmucl manual has been added.
2141 * A facility has been added to extract documentation strings from
2142 sbcl and store them as Texinfo-formatted snippets for inclusion in
2143 the manual (via Texinfo's @include directive)
2144 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2145 the test case to Dave Roberts)
2146 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2147 pointers. (reported by Sean Ross)
2148 * bug fix: PROFILE output is printed nicely even for large numerical
2149 values. (thanks to Zach Beane)
2150 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2151 greater than 32 handle EOF correctly.
2152 * bug fix: on X86 an immediate argument of the IMUL instruction is
2153 correctly printed in disassembly. (thanks to Lutz Euler)
2154 * bug fix: class slots in redefined classes preserve their old
2155 values. (thanks to Bruno Haible and Nikodemus Siivola)
2156 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2157 succeeds. (reported by Marco Baringer)
2158 * fixed some bugs revealed by Paul Dietz' test suite:
2159 ** READ-SEQUENCE now works on ECHO-STREAMs.
2160 ** RATIONALIZE works more according to its specification. (thanks
2163 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2164 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2165 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2166 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2167 should matter to you only if you rebind the printer control
2168 variables and then find you want different bindings in the
2169 debugger than in the ordinary execution of your program.)
2170 * The runtime build system has been tweaked to support building
2171 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2172 assembler and linker. (thanks to Nikodemus Siivola)
2173 * Unbound, undefined, undeclared variables now trigger full
2174 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2175 more often programmer error than deliberate exploitation of undefined
2177 * optimization: the hash algorithm for strings has changed to one
2178 that is less vulnerable to spurious collisions. (thanks to Juho
2180 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2181 needless bounds checking. (thanks to Juho Snellman)
2182 * optimization: implemented multiplication as a modular
2183 (UNSIGNED-BYTE 32) operation on the PPC backend.
2184 * fixed some bugs revealed by Paul Dietz' test suite:
2185 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2187 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2188 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2189 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2191 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2193 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2195 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2196 their output stream on EOF from read.
2197 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2198 have been read to end-of-file.
2199 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2201 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2202 description of determination of which consecutive characters
2204 ** Printing the "Space" character with escaping on now yields "#\\ ",
2205 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2206 ** Reading floating-point numbers with *READ-BASE* set to a number
2207 less than 10 works correctly.
2208 ** Reading floating-point numbers with *READ-BASE* set to a number
2209 more than 10 works correctly.
2210 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2211 the readtable currently in effect.
2213 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2214 * minor incompatible change: parsing of namestrings on a physical
2215 (Unix) host has changed; numbers after the final #\. in a
2216 namestring are no longer interpreted as a version field. This is
2217 intented to be largely invisible to the user, except that the
2218 meaning of the namestring "*.*.*" has changed: it now refers to a
2219 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2220 should usually be replaced by
2221 (make-pathname :name :wild :type :wild :version :wild)
2222 with the added benefit that this is more likely to be portable.
2223 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2224 option to OPEN now signals an error if the file being opened
2225 exists; this may have an impact on existing code.
2226 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2227 signals received, on the PowerPC platforms (both Linux and
2228 Darwin). (thanks to Pierre Mai for pointing out the location of
2230 * several fixes on OS X: The system now builds and runs cleanly on
2231 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2232 (thanks to Brian Mastenbrook)
2233 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2234 second-resolution: integer multiples of 1/3600 between -24 and 24.
2235 (thanks to Vincent Arkesteijn)
2236 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2237 type when called with 1 argument; PEEK-CHAR checked type of
2238 PEEK-TYPE only after having read first character from a
2239 stream. (reported by Peter Graves)
2240 * bug fix: the garbage collector now has much better locality
2241 behaviour, and in particular no longer treats all memory as being
2242 exhausted when many small objects point to each other in a deeply
2244 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2245 recognized as being TYPEP their class.
2246 * bug fix: the PUSHNEW documentation string has been corrected.
2247 (thanks to Vincent Arkesteijn)
2248 * bug fix: defaulting of the value for the last of an atypically
2249 large number of multiple values being bound was not being
2250 performed correctly on the Alpha or PPC platforms
2251 * optimization: implemented multiplication as a modular
2252 (UNSIGNED-BYTE 32) operation on the x86 backend.
2253 * optimization: SEARCH on simple-base-strings can now be open-coded.
2254 (see also contrib/compiler-extras.lisp for inspiration for
2255 teaching the compiler about the Boyer-Moore algorithm).
2256 * value, returned by MAX (and MIN) called with several EQUALP, but
2257 not EQL, arguments now does not depend on compiler settings.
2258 * fixed some bugs revealed by Paul Dietz' test suite:
2259 ** in stack analysis liveness information is propagated from
2260 non-local entry points.
2261 ** pathwise CAST removing failed when the CAST node did not start
2263 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2264 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2266 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2267 host is already defined.
2268 ** RENAME-FILE works on streams instead of signalling an internal
2270 ** PEEK-CHAR uses the current readtable when determining whether
2271 or not a character is whitespace.
2272 ** MERGE-PATHNAMES handles the case when the pathname does not
2273 specify a name while the default-pathname specifies a version.
2274 ** Pathnames now stand a better chance of respecting print/read
2276 ** Attempting to use standardized file system operators with a
2277 pathname with invalid :DIRECTORY components signals a
2279 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2281 ** DIRECTORY on logical pathnames is more correct.
2282 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2283 signal a TYPE-ERROR if their argument is not a stream.
2284 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2285 designator argument does not designate a stream.
2286 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2287 examining the synonym.
2288 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2290 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2291 element-type larger than ([UN]SIGNED-BYTE 32).
2293 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2294 * When built with the :SB-FUTEX feature, threaded builds now take
2295 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2296 for faster/more reliable mutex and condition variable support.
2297 * Incompatible change (but one you probably shouldn't have been using
2298 anyway): the interface and code for arbitrating between multiple
2299 threads in the same user session has been redesigned.
2300 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2301 error when called without an explicit environment argument.
2302 (thanks to Nikodemus Siivola)
2303 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2304 confused about their position. (thanks to Adam Warner and Gerd
2306 * bug fix: Pretty printing backquoted forms with unquotations in the
2307 argument list position of various code constructs such as LAMBDA
2308 now works correctly. (reported by Paul Dietz)
2309 * bug fix: Pretty printing unquotations no longer loses all
2310 stream position information.
2311 * optimization: performance of string output streams is now less
2312 poor for multiple small sequence writes.
2313 * optimization: performance of CSUBTYPEP in the presence of complex
2314 expressions involving CONS and NOT many times has been improved.
2315 (reported by Paul Dietz)
2316 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2318 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2319 streams for the REPL, for the debugger, and so on. (thanks to
2321 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2322 * fixed some bugs revealed by Paul Dietz' test suite:
2323 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2324 is ignored if it is too small, rather than propagating through
2325 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2326 ** extremely complex negations of CONS types were not being
2327 sufficiently canonicalized, leading to inconsistencies in
2329 ** VALUES tranformer lost derived type.
2331 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2332 * fixed a bootstrapping bug: the build process no longer assumes
2333 that the various BOOLE-related constants have the same value in
2334 host and target lisps. (noted by Paul Dietz' test suite on an
2335 SBCL binary built from CLISP)
2336 * The system can now be dynamically linked on the MIPS platform,
2337 which enables dynamic loading of foreign code from Lisp. (thanks
2338 to Ralf Baechle for discussions on the MIPS ABI)
2339 * The system now records debugging information for its own source
2340 files in a filesystem-position-independent manner, relative to
2341 the "SYS" logical host.
2342 * fixed a compiler bug: MV-LET convertion did not check references
2343 to the "max args" entry point. (reported by Brian Downing)
2344 * tweaked disassembly notes to be less confident about proclaiming
2345 some instruction as an LRA. (thanks to Brian Downing)
2346 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2347 now each have their own history, command character, and other
2348 characteristics. (thanks to David Lichteblau)
2349 * fixed some bugs revealed by Paul Dietz' test suite:
2350 ** compiler failure in compiling LOGAND expressions including a
2352 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2353 shift greater than 32.
2354 ** FLUSH-DEST did not mark blocks for type check regeneration.
2355 ** HANDLER-CASE failed to accept declarations in handler clauses
2356 in some circumstances.
2358 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2359 * New code in contrib/sb-introspect (still probably not entirely
2360 stable yet) provides some support for smart Lisp development
2361 environments like SLIME.
2362 * The conditions signalled for errors occurring when loading .fasl
2363 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2364 in a way which should help ASDF recover gracefully.
2365 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2366 stuff has been cleaned up. If you code contrib/ stuff, this might
2367 affect you, and you can look at contrib/README, contrib/STANDARDS,
2368 and/or the 0.8.4.27 diff to check.
2369 * In full calls the compiler now does not generate checks for declared
2370 argument types for all arguments.
2371 * various threading fixes
2372 ** and some experimental patches which didn't make it into
2373 the main tree for this release, but which are shipped in
2374 contrib/experimental-thread.patch as a possible fix for some
2375 failures (deadlock, spinning...) in GC-intensive multithreaded
2377 * fixed PPC build problem (source code incompatibility of different
2378 library versions): added offsetof() hackery which attempts to divine
2379 where glibc maintainers put uc_mcontext today
2380 * fixed bug 282: compiler does not trust type assertions while passing
2381 arguments to a full call.
2382 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2383 &OPTIONAL VALUES type parameter.
2384 * fix bug 214: algorithm for noting rejected templates is now more
2385 similar to that of template seletion. (also reported by rydis on
2387 * fixed bug 141b: printing backquoted information readably and prettily
2388 inserts a space where necessary.
2389 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2390 (CHARACTER 10) are now reported as errors, rather than propagated
2391 as unknown types. (reported by piso on #lisp)
2392 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2393 with values NIL and :ERROR. (thanks to Milan Zamazal)
2394 * fixed bug 191c: CLOS now does proper keyword argument checking as
2395 described in CLHS 7.6.5 and 7.6.5.1.
2396 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2397 longer raise an error; further, using a list as a for-as-arithmetic
2398 counter now raises a meaningful error.
2399 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2400 now understood by sequence creation functions such as MAKE-SEQUENCE
2402 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2403 a STRING-INPUT-STREAM.
2404 * compiler enhancement: SIGNUM is now better able to derive the type
2406 * type declarations inside WITH-SLOTS are checked. (reported by
2408 * fixed some bugs revealed by Paul Dietz' test suite:
2409 ** incorrect optimization of TRUNCATE for a positive first
2410 argument and negative second.
2411 ** compiler failure in let-convertion during flushing dead code.
2412 ** compiler failure while deriving type of TRUNCATE on an
2413 interval, containing 0.
2414 ** ASH of a negative bignum by a negative bignum count now returns
2416 ** intersection of CONS types now canonicalizes properly, fixing
2417 inconsistencies in SUBTYPEP.
2419 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2420 * incompatible change: The --disable-debugger command line
2421 option now clobbers the debugger at a more fundamental
2422 level, by redefining #'INVOKE-DEBUGGER instead of by
2423 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2424 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2425 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2426 enter the debugger REPL and then suffer recursive errors
2427 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2428 while under the new system, BREAK in --disable-debugger mode
2429 terminates the system just as an unhandled error would.
2430 * fixed compiler performance when processing loops with a step >1;
2431 * bug fix: DOCUMENTATION now retrieves generic function
2432 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2433 support has been systematized, and now supports the methods
2434 specified by ANSI, along with a default method and a method for
2435 slot documentation. (reported by Nathan Froyd)
2436 * bug fix: effective methods associated with a generic function are
2437 no longer cached over a change of that generic function's method
2438 combination. (reported by Andreas Fuchs)
2439 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2440 names a non-existent directory. (thanks to Andreas Fuchs)
2441 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2442 the PPC platform, lead to stack corruption; this has been fixed.
2443 (reported by Rainer Joswig)
2444 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2445 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2446 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2447 arithmetic in general, is now much more reliable on the PPC
2449 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2450 platform now returns the right answer.
2451 * optimization: restored some effective method precomputation in
2452 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2453 precomputation is now tunable.
2454 * optimization: compiler-internal data structure use has been
2455 reviewed, and changes have been made that should improve the
2456 performance of the compiler by about 20%.
2457 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2458 simple-base-strings and simple-bit-vectors is improved.
2459 * optimization: the optimization of 32-bit logical and arithmetic
2460 functions introduced in version 0.8.3 on the x86 has been
2461 implemented on the mips, ppc and sparc platforms; an
2462 implementation of the same facility, but for 64-bit arithmetic,
2463 has been added for the alpha.
2464 * microoptimization: the compiler is better able to make use of the
2465 x86 LEA instruction for multiplication by constants.
2466 * bug fix: in some situations compiler did not report usage of
2467 generic arithmetic in (SPEED 3) policy.
2468 * bug 145b fix: compiler used wrong type specifier while converting
2469 MEMBER-types to numeric.
2470 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2472 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2473 index argument now works properly on the Alpha platform.
2474 * bug fix: floating point exception treatment on the Alpha platform
2476 * bug fix: FILE-POSITION works much better on string input and
2477 output streams. (thanks to Nikodemus Siivola)
2478 * bug fix: many threading/garbage collection symptoms sorted.
2479 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2480 might be pseudo-atomic.
2481 * internal change: Stopping for GC is now done with signals not
2482 ptrace. GC is now done in whichever thread wanted it, instead of
2484 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2486 * bug fix: result form in DO is not contained in the implicit
2488 * incompatible change: ICR structure is changed; the value part of
2489 CONTINUATION is now called LVAR; corresponding functions are
2490 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2491 * added type deriver for ISQRT (thanks to Robert E. Brown).
2492 * bug fix: better support for loading from the command line when an
2493 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2495 * fixed some bugs revealed by Paul Dietz' test suite:
2496 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2497 ** ROUND and FROUND now give the right answer when given very
2498 small float arguments.
2499 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2501 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2502 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2503 ** LCM should return a non-negative integer.
2504 ** PARSE-INTEGER returned the index of a terminator instead of the
2505 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2506 ** PARSE-INTEGER returned an incorrect index being applied to a
2508 ** LCM with two arguments of 0 returns 0 rather than signalling
2510 ** unsigned addition of a 32-bit constant with the high bit set no
2511 longer causes an internal compiler error on the x86.
2512 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2513 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2514 with negative last argument.
2515 ** byte specifiers with zero size and position no longer cause
2516 an error during type derivation.
2517 ** bignum multiplication on the Alpha platform now returns the
2519 * porting: The system now builds on SuSE AMD64, although it still
2520 generates a 32-bit binary.
2521 * .fasl file incompatibility: The fasl file version number has
2522 been incremented (because of the changes to internal compiler
2523 data structures referred to above).
2525 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2526 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2527 more accurately, on the Darwin kernel running on PowerPC hardware.
2528 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2529 * Compiler code deletion notes now signal a condition of type
2530 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2531 an associated MUFFLE-WARNING restart.
2532 * The compiler now performs limited argument count validation of
2533 constant format strings in FORMAT, and where appropriate in ERROR,
2534 CERROR and WARN. (thanks to Gerd Moellmann)
2535 * New ASDF-INSTALL contrib can be used for automatic download and
2536 installation of third-party Lisp code from CCLAN or other sites
2538 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2539 which forces another thread to execute a function supplied by the
2541 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2542 now accept and act upon their :ELEMENT-TYPE keyword argument.
2543 (reported by Martin Atzmueller, Edi Weitz)
2544 * bug fix: FILE-POSITION now accepts position designators up to
2545 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2546 is the greater. (thanks to Patrik Nordebo)
2547 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2548 Kalle Olavi Niemitalo)
2549 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2550 printable. (reported by Eric Marsden)
2551 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2552 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2553 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2554 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2555 (thanks to Stig Sandoe)
2556 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2557 UNREAD-CHAR now correctly includes the unread character in the
2558 target sequence. (thanks to Gerd Moellmann)
2559 * bug fix in threaded builds: the system can now be suspended and
2560 resumed by shell job control with minimal disruption.
2561 * bug fixes in times and timezones >2038AD
2562 * better handling of "where is GNU make?" problem in build scripts
2563 (thanks to Nikodemus Siivola)
2564 * new optimization: inside a named function any reference to a
2565 function with the same name is considered to be a self-reference;
2566 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2567 optimization quality.
2568 * new optimization on x86: logical functions and + now have
2569 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2570 used when the result is truncated to 32 bits.
2571 * VALUES declaration is partially enabled.
2572 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2573 * fixed some bugs revealed by Paul Dietz' test suite:
2574 ** The system now obeys the constraint imposed by
2575 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2576 types form a lattice under type intersection.
2577 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2578 ** ASSOC now ignores NIL elements in an alist.
2579 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2580 and (1+ MOST-POSITIVE-FIXNUM) answers.
2581 ** The addition of a method with invalid qualifiers to a generic
2582 function does not cause an error to be signalled immediately;
2583 a warning is signalled, and the error is generated only on
2584 calling the generic function.
2585 * changed .fasl file version number, in part to add type codes for
2586 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2587 obscure ANSI requirements
2589 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2590 * fixed bug 148: failure to inline-expand a local function left
2591 garbage, confusing the compiler.
2592 * fixed bugs 3cd: structure slot readers perform type check if the
2593 slot can have an invalid value (i.e. it is either not initialized
2594 or can be written with a less specific slot writer).
2595 * bug fix: the compiler now traps array references to elements off
2596 the end of an array; previously, the bounds checking in some
2597 circumstances could go off-by-one.
2598 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2600 * type declarations for array element types now obey the description
2601 on the CLHS page "Declaration TYPE", as per discussions on
2602 sbcl-help around 2003-05-08. This means that a declaration
2603 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2604 declaration, all references to BAR will be asserted or assumed
2605 (with THE, so dependent on compiler policy) to involve objects of
2606 type FOO. Note that no such declaration is implied in
2607 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2608 * declared types of functions from the "Conditions"
2609 chapter. (reported by Paul Dietz)
2610 * bug fix: CERROR accepts a function as its first argument.
2611 * bug fix: NTH an NTHCDR accept a bignum as index
2612 arguments. (reported by Adam Warner)
2613 * optimization: character compare routines now optimize comparing
2614 against a constant character. (reported by Gilbert Baumann)
2615 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2616 argument now works properly on the MIPS platform.
2617 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2618 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2619 declared to have a functional type, does not cause an error
2621 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2622 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2623 * fixed bug: if last continuation of a deleted block has a
2624 destination, this destination should be deleted too. (reported by
2626 * fixed a bug in the bootstrap process: the host compiler's values
2627 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2628 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2629 test case from Patrik Nordebo)
2630 * improved the ability of the disassembler on the PPC platform to
2631 provide helpful disassembly notes.
2632 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2633 the class in more cases than previously.
2634 * bug fix: FILE-POSITION now understands :START and :END for
2635 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2636 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2637 8) data. (thanks to David Lichteblau)
2638 * bug fix: it is possible to add a method to a generic function
2639 without lambda list.
2640 * bug fix: reader failed to signal END-OF-FILE inside an
2641 object representation. (reported by Nikodemus Siivola)
2642 * fixed some bugs revealed by Paul Dietz' test suite:
2643 ** LAST and [N]BUTLAST should accept a bignum.
2644 ** condition slot accessors are methods.
2645 ** (VECTOR NIL) is a subtype of STRING.
2647 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2648 * minor incompatible change: some nonsensical specialized lambda
2649 lists (used in DEFMETHOD) which were previously ignored now signal
2651 * minor incompatible change: the system is now aware of the types of
2652 variables in the COMMON-LISP package, and will signal errors for
2653 most violations of these type constraints (where previously they
2654 were silently accepted).
2655 * minor incompatible change: COMPILE-FILE now uses the freedom
2656 afforded (ANSI 3.2.2.3) to use derived function types for
2657 functions defined in the same file. This also permits the system
2658 to warn on static type mismatches and function
2659 redefinition. (Currently it does not work with high DEBUG level.)
2660 * minor incompatible change: VALUES declaration is disabled.
2661 * When issuing notes, the compiler now signals a condition of type
2662 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2663 restart for use in user handlers. It is expected that the
2664 COMPILER-NOTE condition will eventually become a condition
2665 supertype to a hierarchy of note types, which will then be
2666 handleable in a similar fashion. However, at the moment, no such
2667 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2668 but it's an implementation detail, not a classification for the
2670 * Changes in type checking closed the following bugs:
2671 ** type checking of unused values (192b, 194d, 203);
2672 ** template selection based on unsafe type assertions (192c, 236);
2673 ** type checking in branches (194bc).
2674 * A short form of VALUES type specifier has ANSI meaning (it has
2675 increased the number of situations when SBCL cannot perform type
2677 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2678 keywords or constants is permissible.
2679 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2680 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2681 classes. (thanks to Antonio Martinez)
2682 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2683 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2684 argument to a pretty-print stream. Instead, it leaves any such
2685 operations to DESCRIBE-OBJECT methods.
2686 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2688 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2689 lambda lists. (thanks to David Lichteblau)
2690 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2691 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2692 (thanks to Gerd Moellmann)
2693 * fixed bug 246: increased compilation speed of long
2694 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2696 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2698 * DEFCONSTANT now throws a condition of type
2699 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2700 constant to a non-EQL value; CONTINUE and ABORT restarts
2701 respectively change and preserve the value.
2702 * fixed bug 63: The code walker, part of the implementation of CLOS,
2703 is now better at handling symbol macros.
2704 * bug fix: There is no longer an internal implementation type named
2705 CL:LENGTH. (reported by Raymond Toy)
2706 * bug fix: In macro-like defining macros/special operators the
2707 implicit block does not enclose the lambda list.
2708 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2709 enclosed in parentheses) are not suitable as type specifiers, and
2710 their use properly signals an error now.
2711 * bug fix: An argument count mismatch for a type specifier in code
2712 being compiled no longer causes an unhandled error at compile
2713 time, but signals a compile-time warning.
2714 * fixed simple vector readable printing
2715 * bug fix: DESCRIBE takes more care over whether the class
2716 precedence list slot of a class is bound before accessing it.
2717 (reported by Markus Krummenacker)
2718 * bug fix: FORMATTER can successfully compile pretty-printer format
2719 strings which use variants of the ~* directive inside.
2720 * bug fix: SEARCH now applies its TEST predicate to the elements of
2721 the arguments in the correct order. (thanks to Wolfhard Buss)
2722 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2723 in type inference. (thanks to Robert E. Brown)
2724 * bug fix: Reading in symbols with an explicit package name of ""
2725 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2726 * fixed some bugs revealed by Paul Dietz' test suite:
2727 ** NIL is now allowed as a structure slot name.
2728 ** Arbitrary numbers, not just REALs, are allowed in certain
2729 circumstances in LOOP for-as-arithmetic clauses.
2730 ** Multiple class redefinitions before slot access no longer
2731 causes a type error.
2732 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2733 association between the name and a class.
2734 ** Generic functions with non-standard method-combination and over
2735 five methods all of which return constants no longer return NIL
2736 after the first few invocations. (thanks to Gerd Moellmann)
2737 ** CALL-NEXT-METHOD with no arguments now passes the original
2738 values of the arguments, even in the presence of assignment.
2739 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2741 ** DISASSEMBLE works with closures and funcallable instances.
2742 ** ADD-METHOD now returns the generic function, not the new method.
2743 ** FIND-METHOD signals an error if the lengths of the specializers
2744 is incompatible with the generic function, even if the ERRORP
2746 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2747 which its argument is a member.
2748 ** DEFCLASS only redefines the class named by its class-name
2749 argument if that name is the proper name of the class;
2750 otherwise, it creates a new class.
2751 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2752 of the UNBOUND-SLOT condition to the name of the slot.
2753 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2754 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2755 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2756 SLOT-MAKUNBOUND in the specified fashion.
2758 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2759 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2760 cross-compilation host. As a consequence, we can now bootstrap our
2761 way up to SBCL starting with a bare gcc toolchain and human-readable
2762 source code (first the source to CLISP, then the source to SBCL).
2763 * A contributed module containing a partial implementation of the
2764 simple-streams interface has been included. (thanks to Rudi
2766 * A contributed module implementing the RFC1321 Message Digest
2767 Algorithm, known as MD5, has been included.
2768 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2769 no longer has any effect, as the code controlled by this feature
2770 has been deleted. (As far as we know, no-one has ever built using
2771 this feature, and its semantics were confused in any case).
2772 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2773 work on conditions (as required by the ANSI specification),
2774 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2775 expected behaviour on conditions. Users should note, however,
2776 that such behaviour is not required by the ANSI specification,
2777 and so use of this behaviour may render their code unportable.
2778 * fixed some bugs revealed by Paul Dietz' test suite:
2779 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2781 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2782 ** SXHASH on condition objects no longer returns NIL.
2783 ** :ALLOCATION :CLASS slots are better treated; their values are
2784 updated on class redefinition, and initforms inherited from
2785 superclasses are applied.
2786 ** REMOVE-METHOD returns its generic function argument even when
2787 no method was removed.
2788 ** SHARED-INITIALIZE now initializes the values of the requested
2789 slots, including those with :ALLOCATION :CLASS.
2790 ** ALLOCATE-INSTANCE now works on structure classes defined via
2791 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2793 ** SLOT-EXISTS-P now works on conditions, as well as structures
2795 ** MAKE-LOAD-FORM now has the required methods on
2796 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2797 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2798 keyword, and now implements the SLOT-NAMES argument.
2799 ** methods with &OPTIONAL arguments no longer allow too many
2800 arguments to be passed in the call without error.
2801 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2802 option is consistent with the required arguments of the generic
2803 function lambda list.
2804 * bug fix: REQUIRE accepts a string designator. (Thanks to
2806 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2807 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2808 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2809 * bug fix: The compiler now checks for duplicated variables in macro
2811 * bug fix: SETQ on globals returns the correct value.
2812 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2813 not a COMPILER-ERROR (followed by some other strange error on
2814 choosing the CONTINUE restart).
2815 * bug fix: make.sh and friends are now more consistent in the way that
2816 they look for GNU "make".
2818 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2819 * experimental native threads support (on x86 Linux >=2.4 only).
2820 This is not compiled in by default: you need to add :SB-THREAD to
2821 the target features. See the "Beyond ANSI" chapter of the manual
2823 * fix for longstanding nonANSIism: The old distinction between
2824 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2825 The return value from CL:FIND-CLASS is now a CLOS class, and
2826 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2827 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2828 * An interface to the MetaObject Protocol, as described in Kiczales,
2829 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2830 MIT Press, 1991, is available from the SB-MOP package.
2831 * incompatible change: the SB-PCL package should now be considered
2832 a private implementation detail, and no longer a semi-private MOP
2834 * minor incompatible change: due to rearrangement for threads, the
2835 control stack and binding stack are now allocated at arbitrary
2836 addresses instead of being hardcoded per-port. Users affected by
2837 this probably have to be doing advanced things with shared
2838 libraries, and will know who they are.
2839 * minor incompatible change: Previously, all --eval forms used were
2840 processed with READ before any of them were processed with EVAL.
2841 Now each --eval form is processed with both READ and EVAL before
2842 the next --eval form is processed. (Thus package operations like
2843 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2844 work as the user might reasonably expect.)
2845 * minor incompatible change: *STANDARD-INPUT* is now only an
2846 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2848 * minor incompatible change: Y-OR-N-P is now character-oriented,
2849 not line oriented. Also, YES-OR-NO-P now works without errors.
2850 (thanks to Antonio Martinez)
2851 * sb-aclrepl module improvements: an integrated inspector, added
2852 repl features, and a bug fix to :trace command.
2853 * Known functions, which cannot be open coded by the backend, are
2854 considered to be able to check types of their arguments. (fixing
2855 a bug report by Nathan J. Froyd)
2856 * fixed a bug in computing method discriminating functions: It is
2857 now possible to define methods specialized on classes which have
2858 forward-referenced superclasses. (thanks to Gerd Moellmann)
2859 * fixed evaluation order in optional entries (reported by Gilbert
2861 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2862 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2863 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2864 required-by-AMOP NAME argument, as well as CLASS and
2865 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2866 * fixed bug 20: DEFMETHOD can define methods using names that are
2867 not the proper names of classes to designate class specializers.
2868 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2869 Unix file descriptors, instead of blowing up. (thanks to Antonio
2871 * Garbage collection refactoring: user-visible change is that a
2872 call to the GC function during WITHOUT-GCING will not do garbage
2873 collection until the end of the WITHOUT-GCING. If you were doing
2874 this you were probably losing anyway.
2875 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2876 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2877 TYPEP the latter but not the former.
2878 * The compiler issues a full WARNING for calls to undefined functions
2879 with names from the CL package.
2880 * MAP-INTO for a vector destination is open coded. (reported by
2881 Brian Downing on c.l.l)
2882 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2883 documentation string.
2884 * fixed some bugs revealed by Paul Dietz' test suite:
2885 ** COPY-ALIST now signals an error if its argument is a dotted
2887 ** Condition slots are now accessed more correctly in the presence
2888 of multiple initargs for a given slot.
2889 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2890 exclude restarts of the same name associated with a different
2892 ** DEFCLASS of forward-referenced classes with another
2893 forward-referenced class in the superclasses list no longer
2895 ** Condition slots are now initialized once each, not multiple
2896 times. (thanks to Gerd Moellmann)
2897 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2899 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2900 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2901 arguments contain duplicated elements.
2902 ** RESTART-CASE understands local macros.
2903 ** RESTART-CASE associates exactly its own restarts with a condition.
2904 ** ENDP in safe mode checks its argument to be of type LIST.
2905 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2906 in question is unbound.
2907 ** Optimized MAKE-INSTANCE functions no longer cause internal
2908 assertion failures in the presence of duplicate initargs.
2909 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2910 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2911 propagated even in the ctor optimized implementation of
2913 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2915 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2916 * a better implementation of SXHASH on (simple) bit vectors,
2917 measured both in execution speed and in distribution of results
2918 over the positive fixnums, has been installed. Likewise, a better
2919 implementation of EQUAL for simple bit vectors is now available.
2920 * fixed CEILING optimization for a divisor of form 2^k.
2921 * fixed bug 240 (emitting extra style warnings "using the lexical
2922 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2923 by Antonio Martinez)
2924 * fixed SXHASH, giving different results for NIL depending on type
2925 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2926 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2927 variable. (found by Rolf Wester)
2928 * SBCL does not ignore type declarations for special
2929 variables. (reported by rif on c.l.l 2003-03-05)
2930 * some bug fixes in contrib/sb-aclrepl/
2931 * fixed some bugs revealed by Paul Dietz' test suite:
2932 ** a bug in the CONS type specifier, whereby the CAR and CDR
2933 types got intertwined, has been fixed;
2934 ** the type system is now able to reason about the interaction
2935 between INTEGER and RATIO types more completely;
2936 ** APPEND, [N]REVERSE and NRECONC check that those their
2937 arguments, which must be proper lists, are really so;
2938 ** An array specialized to be unable to hold elements has been
2939 implemented, as required -- yes, really -- by ANSI;
2940 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2941 on malformed property lists;
2943 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2944 * incompatible packaging change: in line with Unix convention,
2945 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2946 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2947 * REQUIRE and PROVIDE are now optionally capable of doing something
2948 useful. See the documentation string for REQUIRE.
2949 * infrastructure for a managed SBCL contrib system: contributed
2950 modules in this release include:
2951 ** the ASDF system definition facility;
2952 ** an interface to the BSD Sockets API;
2953 ** an ACL-like convenience interface to the repl;
2954 (thanks to Kevin Rosenberg)
2955 ** an implementation of ROTATE-BYTE, with efficient implementation
2957 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2958 gives a closer approximation to the right answer than previously.
2959 (thanks to Raymond Toy)
2960 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2961 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2962 optional environment arguments, as required by ANSI.
2963 * fixed bugs in other functions taking environment objects, allowing
2964 calls with an explicit NIL environment argument to be compiled
2966 * fixed bug 228: primary return values from
2967 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2968 COMPILE or FUNCTION.
2969 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2970 :TYPE will no longer signal errors on innocuous objects.
2971 * fixed bug 231b: SETQ is better at respecting type declarations in
2972 the lexical environment.
2973 * fixed a bug in DEFCLASS: classes named by symbols with no or
2974 unprintable packages can now be defined.
2975 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2976 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2977 * fixed compiler failure related to checking types of functions.
2978 (reported by Robert E. Brown)
2979 * the compiler is now much more consistent in its error-checking
2980 treatment of bounding index arguments to sequence functions: in
2981 (SAFETY 3) code, errors will be signalled in almost all cases if
2982 invalid sequence bounding indices are passed to functions defined
2983 by ANSI to operate on sequences.
2984 * fixed a bug in the build procedure: documentation of SBCL-specific
2985 packages is now preserved and available in the final Lisp image.
2986 * lifted FDEFINITION lookup out of loops in the implementation of
2987 many list operations. (thanks to Robert E. Brown)
2988 * fixed a bug in the reader: the #n# reader macro now works for
2989 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2990 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2991 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2992 E. Brown and Tony Martinez respectively)
2993 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2994 SLOT-BOUNDP: in particular, it is now able to optimize them much
2995 better, and is now not vulnerable to having packages renamed.
2996 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2997 always called when a slot is not present in an instance. (thanks
2999 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3000 longer signal an error on unreading a character following EOF on
3001 the previous constituent stream. (thanks to Tony Martinez)
3002 * fixed some bugs revealed by Paul Dietz' test suite:
3003 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3004 not just nonnegative fixnums;
3005 ** the logical bit-array operators such as BIT-AND now accept an
3006 explicit NIL for their "opt-arg" argument (to indicate a
3007 freshly-consed result bit-array);
3008 ** ELT now signals an error on an invalid sequence index in safe
3010 ** the type system is now cleverer about negations of numeric
3011 types, and consequently understands the BIGNUM and RATIO types
3013 ** the type system is now cleverer about the interaction between
3014 INTEGER and RATIO types: while bugs still remain, many more
3015 cases are accurately computed;
3016 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3017 if it is in the last clause;
3018 ** CONSTANTLY now correctly returns a side-effect-free function in
3020 ** DECLARE is no longer treated as a special-operator; in
3021 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3022 * incremented fasl file version number due to the change in the
3023 DEFSTRUCT-SLOT-DESCRIPTION structure.
3025 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3026 * minor incompatible change: code processed by the "interpreter" or
3027 EVAL now has a compilation optimization policy of (DEBUG 2)
3028 (changed from (DEBUG 1)) to improve debuggability of interactive
3029 development, and to allow the use of the debug RETURN command in
3031 * an experimental implementation of the RETURN command for the
3032 debugger has been included. (thanks to Frederik Kuivinen)
3033 * fixed bug 62: constraints were not propagated into a loop.
3034 * fixed bug in embedded calls of SORT (reported and investigated by
3036 * fixed some bugs revealed by Paul Dietz' test suite:
3037 ** printing and reading of arrays with some dimensions having
3038 length 0 (thanks to Gerd Moellmann);
3039 ** BOA constructor with &AUX argument without a default value does
3040 not cause a type error;
3041 ** CONSTANTP now returns true for all self-evaluating objects.
3043 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3044 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3045 accessors that are related by inheritance, as specified in the
3046 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3047 Valtteri Vuorikoski)
3048 * The compiler is now able to inline functions that were defined in
3049 a complex lexical environment (e.g. inside a MACROLET).
3050 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3052 * Support for the upcoming FreeBSD-5.0 release has been included.
3053 (thanks to Dag-Erling Smorgrav)
3054 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3055 effect when it is not in a toplevel context.
3056 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3057 stand a better chance of being correct. (thanks to Gerd
3059 * fixed bug in COERCE, which now signals an error on coercing a
3060 rational to a bounded real type which excludes the expected
3062 * The compiler is now able to derive types more accurately from the
3063 COERCE and COMPILE functions.
3064 * fixed bug 223: functional binding is considered to be constant
3065 only for symbols in the CL package.
3066 * fixed bug 231: SETQ did not check the type of a variable being set
3067 (reported by Robert E. Brown)
3068 * A new optimization for MAKE-INSTANCE has been included, fixing
3069 various bugs (including relating to :ALLOCATION :CLASS slots and
3070 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3072 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3073 ** As required by ANSI, LOOP now disallows anonymous collection
3074 clauses such as COLLECT I in conjunction with aggregate boolean
3075 clauses such as THEREIS (= I 1);
3076 ** LOOP now signals an error when any variable is reused in the
3077 same loop (including the potentially useful construct analogous
3078 to WITH A = 1 WITH A = (1+ A);
3079 ** IT is only a special loop symbol within the first clause of a
3080 conditional loop clause;
3081 ** LOOP with a typed iteration variable over a hashtable now
3082 signals a type error iff it should.
3083 * fixed some other bugs revealed by Paul Dietz' test suite:
3084 ** FILE-STREAM now names the class previously known as FD-STREAM;
3085 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3086 argument) no longer signals an error;
3087 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3088 of the slot symbol, rather than using the current package
3089 ((:CONC-NAME "") continues to intern the slot's name in the
3091 * incremented fasl file version number, because of the incompatible
3092 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3093 of the new implementation of DEFINE-COMPILER-MACRO.
3095 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3096 * Support for building SBCL for MIPS platforms running in
3097 little-endian mode has now been checked in, and basic
3098 functionality on said platforms verified.
3099 * minor incompatible change: PCL now records the pathname of a file
3100 in which methods and the like are defined, rather than its
3102 * minor incompatible change: TRUENAME now considers the truename of
3103 a file naming a directory to be the pathname with :DIRECTORY
3104 component indicating that directory.
3105 * minor incompatible change: a NAMED clause in the extended form of
3106 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3107 reason for the previous behaviour is unclear.
3108 * more systematization and improvement of CLOS and MOP conformance
3109 in PCL (thanks to Gerd Moellman and Pierre Mai):
3110 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3112 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3113 primary methods with no specializers;
3114 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3116 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3117 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3118 FORWARD-REFERENCED-CLASSes; error reporting on
3119 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3121 ** SXHASH on CLOS instances now uses a slot internal to the
3122 instance to return different numbers on distinct instances,
3123 while preserving the same return value through invocations of
3125 ** DEFMETHOD signals errors when methods with longer incongruent
3126 lambda lists are added to generic functions;
3127 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3128 CLASS, as specified in AMOP;
3129 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3130 on the order returned by the primary method for classes of
3131 class STANDARD-CLASS;
3132 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3133 * fixed some bugs shown by Paul Dietz' test suite:
3134 ** DOLIST puts its body in TAGBODY;
3135 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3137 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3138 value producing form;
3139 ** if more variables are given to PROGV than values, extra
3140 variables are bound and made to have no value;
3141 ** NSUBSTITUTE on list arguments gets the right answer with
3143 ** ELT signals an error of type TYPE-ERROR when the index argument
3144 is not a valid sequence index;
3145 ** LOOP signals (at macroexpansion time) an error of type
3146 PROGRAM-ERROR when duplicate variable names are found;
3147 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3148 ** FUNCALL of special-operators now cause an error of type
3150 ** PSETQ now works as required in the presence of side-effecting
3151 symbol-macro places;
3152 ** NCONC accepts any object as its last argument;
3153 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3155 ** loop-for-as-package does not require a package to be explicitely
3157 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3159 * fixed bug 166: compiler preserves "there is a way to go"
3160 invariant when deleting code.
3161 * fixed bug 172: macro lambda lists with required arguments after
3162 &REST arguments now cause an error to be signalled. (thanks to
3164 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3165 bug. (thanks to Matthew Danish)
3166 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3168 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3169 arguments are lexically rebound. (thanks to Gerd Moellmann and
3171 * fixed bug 194: error messages are now more informative when there
3172 is no primary method applicable in a call to a generic
3173 function. (thanks to Gerd Moellmann)
3174 * fixed bug in command line argument checking (thanks to Julian
3176 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3177 (thanks to Matthew Danish)
3178 * incremented fasl file version number, because of the
3179 SXHASH-related changes in the layout of CLOS data structures
3181 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3182 * minor incompatible change: The runtime (the Unix executable named
3183 "sbcl") is now much pickier about the .core files it will load.
3184 Essentially it now requires .core files to descend from the same
3185 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3186 as the runtime does. (The intent is to prevent the crashes which
3187 can occur, and which can even be reported as mysterious failures,
3188 when people patch the sources or change the build parameters
3189 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3190 sbcl and .core files.)
3191 * fixed bug: VALUES-LIST is no longer optimized away.
3192 * fixed bug 142: The FFI conversion of C string values to Lisp
3193 string values no longer conses excessively. (thanks to Nathan
3194 Froyd porting Raymond Toy's fix to CMU CL)
3195 * began to systematize and improve MOP conformance in PCL (thanks to
3196 Nathan Froyd, Gerd Moellman and Pierre Mai):
3197 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3199 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3200 ** FINALIZE-INHERITANCE is now called on class finalization;
3201 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3202 argument precedence order.
3203 * fixed bug 202: The compiler no longer fails on functions whose
3204 derived types contradict their declared type.
3205 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3206 so it can be non-toplevel.
3207 * The fasl file version number has changed (because of the new
3208 implementation of DEFMACRO).
3209 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3210 safe code, that any length requirement by their type-specifier
3211 argument is valid. The exceptions to this are described in bug
3213 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3214 functions now check the types of their inputs as required by ANSI.
3215 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3216 introduced symbol is DECLAREd to be SPECIAL.
3217 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3218 * fixed bug: :COUNT argument to sequence functions may be negative
3219 * fixed bug: body of DO-SYMBOLS may contain declarations
3220 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3221 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3222 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3223 symbol macro only once
3224 * fixed printing of call frame when argument list is unavailable
3225 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3226 * compiler no longer signals WARNING on unknown keyword
3229 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3230 * A beta-quality port to the mips architecture running Linux,
3231 based on the old CMUCL backend, has been made. It has been tested
3232 on a big-endian kernel, and works sufficiently well to be able to
3233 rebuild itself; it has not been tested in little-endian mode.
3234 * fixed an inconsistency between gencgc.c and purify.c which made
3235 dumping/loading .core files unreliable
3236 * fixed bug 120a: The compiler now deals correctly with IFs where
3237 the consequent is the same as the alternative, instead of
3238 misderiving the return type. (thanks to Alexey Dejneka)
3239 * fixed bug 113: Logical pathnames are now dumpable (the logical
3240 host is resolved at load-time, throwing an error if it is not
3242 * fixed bug 174: FORMAT's error message is slightly clearer when a
3243 non-printing character is used in a format directive.
3244 * fixed several bugs in compiler checking of type declarations, i.e.
3245 violations of the Python "declarations are assertions" principle
3246 (thanks to Alexey Dejneka)
3247 * fixed several bugs in PCL's error checking (thanks to Gerd
3249 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3251 * fixed bug in compilation of functions as first class values
3252 (thanks to Antonio Martinez-Shotton)
3253 * The compiler's handling TYPE-ERRORs which it can prove will
3254 inevitably happen at runtime has been cleaned up and corrected
3255 in several ways. (thanks to Alexey Dejneka)
3256 * improved argument type checking for various basic arithmetic
3257 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3258 lavished on them in the past that they can be compiled in many
3259 ways in different special cases
3260 * fixed bug 181: compiler checks validity of user supplied type
3262 * cleaned up code flushing in optimization: Function calls which
3263 should signal errors for safety purposes (e.g. which ANSI says
3264 should signal errors when their arguments are of incorrect type)
3265 are no longer optimized away.
3266 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3267 * incremented fasl file version number, because changes in the
3268 implementation of sequence functions like COERCE caused
3269 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3270 in old inline expansions) to become undefined. (Actually these
3271 changes were later undone, so we might very well be binary
3272 compatible with 0.7.7 after all, but leaving the version number
3273 incremented seemed like the simplest and most conservative
3276 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3277 * An alpha-quality port to the parisc architecture running Linux,
3278 based on the old CMUCL backend, has been made. This, even more so
3279 than the other backends, should be considered still a work in
3280 progress; known problems include that the Linux kernel in 64-bit
3281 mode does not propagate the correct sigcontext structure to
3282 userspace, and consequently SBCL on a parisc64 kernel will not
3284 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3285 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3286 suppressed.) Also now that the compiler is looking at declarations
3287 in the environment, it checks optimization declarations as well,
3288 and suppresses inlining when (> DEBUG SPEED).
3289 * More fixes have been made to treatment of floating point exception
3290 treatment and other Unix signals. In particular, floating point
3291 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3292 * The detection and handling of control stack exhaustion (infinite
3293 or very deeply nested recursion) has changed. Stack exhaustion
3294 detection is now done by write-protecting pages at the OS level
3295 and applies at all optimization settings; when found, a
3296 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3297 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3298 be caught using IGNORE-ERRORS.
3299 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3300 correctly when transforming recursive LABELS functions to LETs.
3301 (thanks to Alexey Dejneka)
3302 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3303 that are names of constants or global variables.
3304 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3305 alien routines with docstrings.
3306 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3307 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3309 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3310 (thanks to Eric Marsden)
3311 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3312 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3313 * Bug fix: The SPARC backend can now compile functions involving
3314 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3315 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3316 to a functions expecting a PATHNAME-DESIGNATOR.
3317 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3318 lambda lists. (thanks to Alexey Dejneka)
3319 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3320 functions so that EXPORTing the name of the class doesn't cause
3321 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3322 to Antonio Martinez for reporting this)
3323 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3324 constant-folded without failing an assertion. (thanks to Einar
3325 Floystad Dorum for reporting this)
3326 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3327 types are now dealt with more correctly by the compiler.
3328 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3329 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3330 behaviour analogous to RENAME-FILE. This puts its behaviour more
3331 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3333 * The fasl file version number has changed again. (because of the
3334 bug fix involving the names of PCL MAKE-INSTANCE functions)
3336 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3337 * bug fix: Floating point exceptions are treated much more
3338 consistently on the x86/Linux and PPC/Linux platforms.
3339 * Array initialization with :INITIAL-ELEMENT is now much faster for
3340 cases when the compiler cannot open code the array creation, but
3341 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3342 array accesses have also seen a speed increase.
3343 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3344 specified. (This was at the root of some bad interactions between
3345 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3347 * bug fix: Internal error arguments for undefined functions are now
3348 computed correctly on the PPC/Linux platform.
3349 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3350 Raymond Toy's patch for CMU CL)
3351 * Support for the Solaris 9 operating environment has been included
3352 (thanks to Daniel Merritt)
3353 * A very ugly but hopefully complete draft of the missing FFI chapter
3354 of the manual has been created by reformatting the corresponding
3355 CMU CL manual chapter into (currently very ugly and incoherent)
3356 DocBook and bringing it up to date for SBCL behavior. Thus, the
3357 manual is now essentially complete, at least by my extreme
3358 once-and-only-once standards, whereby it's acceptable to refer to
3359 the doc strings of SB-EXT functions as the primary documentation.
3360 * The fasl file version number has changed again, due to cleanup of
3361 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3362 is no longer a static symbol.)
3364 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3365 * SBCL now builds with OpenMCL (version 0.12) as the
3366 cross-compilation host; also, more progress has been made toward
3367 bootstrapping under CLISP.
3368 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3370 * bug 158 fixed: The compiler can now deal with integer loop
3371 increments different from 1; fixing this turned out also to fix
3373 * bug 169 fixed: no more bogus warnings about using lexical bindings
3374 despite the presence of perfectly good SPECIAL declarations (thanks
3375 to David Lichteblau)
3376 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3377 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3378 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3380 * bug 180 fixed: Method combination specifications no longer ignore
3381 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3382 * bug fix: Structure type predicate functions now check their argument
3383 count as they should.
3384 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3385 correctly. (thanks to Pierre Mai)
3386 * minor incompatible change: The --noprogrammer option is deprecated
3387 in favor of the new --disable-debugger option, which is very similar.
3388 (The major difference is that it takes effect at a slightly different
3389 time at startup, causing handling of errors in --sysinit and
3390 --userinit files will be affected differently.) The
3391 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3392 been added to allow this functionality to be controlled from ordinary
3393 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3394 maintainers, who might want to run non-interactive scripts to
3395 build SBCL cores which will later be used interactively.)
3396 * minor incompatible change: The LOAD function no longer, when given
3397 a wild pathname to load, loads all files matching that pathname.
3398 Instead, an error of type FILE-ERROR is signalled.
3400 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3401 * bug 147 fixed: The compiler preserves its block link/count
3402 invariants more correctly now so that it doesn't crash. (thanks
3404 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3406 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3407 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3408 no-value-for-this-character as NIL (as specified by ANSI).
3409 * HOST-NAMESTRING on physical pathnames now returns a string that is
3410 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3411 (thanks to Christophe Rhodes)
3412 * The Alpha port handles icache flushing more correctly. (thanks to
3414 * More progress has been made toward bootstrapping under CLISP. (thanks
3415 to Christophe Rhodes)
3416 * The fasl file format has changed again, because dynamic loading
3417 on OpenBSD (which has non-ELF object files) motivated some cleanups
3418 in the way that foreign symbols are transformed and passed around.
3419 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3420 is no longer treated as whitespace by the reader, but instead as
3421 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3422 |A
\x7fB|, instead of A as it used to.
3424 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3425 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3426 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3427 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3428 this as of 0.7.1.45, but was left out of the previous news section
3429 (thanks to Dan Barlow)
3430 * SBCL now runs on the Solaris operating system on SPARC architectures
3431 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3432 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3433 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3434 * SPARC backend cleanups, allowing builds of cores optimized for V8
3435 and V9 SPARCS, and also emission of code targeted to a particular
3436 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3438 * SBCL is closer to bootstrapping under CLISP, thanks to various
3439 fixes by Christophe Rhodes.
3440 * The fasl file format has changed again, to allow the compiler's
3441 INFO database to support symbol macros.
3442 * The user manual (in doc/) is formatted into HTML more nicely.
3443 (thanks to coreythomas)
3444 * The system is smarter about SUBTYPEP relationships, especially
3445 those involving NOT types (including types such as ATOM which are
3446 represented internally using NOT types). Thus SUBTYPEP is less
3447 likely to return (VALUES NIL NIL) in general, and in particular
3448 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3450 * The fasl file format has changed again, because the internal
3451 representation of types now includes a new slot to support the new
3452 SUBTYPEP-of-NOT-types logic.
3453 * (not a change in the main branch of SBCL, but a related prototype
3454 which can hopefully be merged into the main branch of SBCL in the
3455 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3456 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3457 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3458 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3459 dispatch macro character combinations. (thanks to Alexey Dejneka)
3460 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3461 that superficially look like logical namestrings correctly.
3462 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3464 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3465 default supplied-p) for &optional and &key arguments. (thanks to
3468 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3469 * incompatible change: The compiler is now less aggressive about
3470 tail call optimization, doing it only when (> SPACE DEBUG) or
3471 (> SPEED DEBUG). (This is an incompatible change because there are
3472 programs which relied on the old CMU-CL-style behavior to optimize
3473 away their unbounded recursion which will now die of stack overflow.)
3474 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3475 for non-GENCGC systems has been increased to 20M (since that
3476 seems much closer to the likely performance optimum for modern
3477 systems than the old 4M value was)
3478 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3479 and *DEBUG-PRINT-LEVEL*
3480 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3481 of CMU CL's support for SPARC, and various endianness and other
3482 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3483 * new syntactic sugar for the Unix command line: --load foo.bar is now
3484 an alternate notation for --eval '(load "foo.bar")'.
3486 ** The system now detects stack overflow and handles it gracefully,
3487 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3488 optimization settings. (This is a good thing in general, and
3489 its introduction in this version should be particularly timely
3490 for anyone whose code fails because of suppression of tail
3492 ** The system now hunts for the C variable "environ" in a more
3493 devious way, to avoid segfaults when the C library version differs
3494 between compile time and run time. (thanks to Christophe Rhodes)
3495 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3496 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3497 ** The compiler no longer issues bogus style warnings for undefined
3498 classes in the same source file as the DEFCLASSes which defined
3499 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3501 ** fixes in CONDITION class precedence list for undefined function
3502 errors (thanks to Alexei Dejneka)
3503 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3504 correctly. (thanks to Dan Barlow)
3505 ** portability fixes aiming at bootstrapping under CLISP (thanks
3506 to Dave McDonald and Christophe Rhodes)
3507 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3508 ** fixes in type translation and and type inference (thanks to
3510 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3511 ** various fixes in the new ports (thanks to Dan Barlow)
3512 * several changes related to debugging:
3513 ** suppression of tail recursion, as noted above
3514 ** stack overflow detection, as noted above
3515 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3516 is now the default. (For some time encapsulation has been more
3517 reliable than the breakpoint-based :ENCAPSULATE NIL
3518 implementation, at least on X86 systems; and I just noticed that
3519 encapsulation also seems closer to the spirit of the ANSI
3522 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3524 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3525 up properly again. (There was a packaging bug in 0.7.0 which
3526 left their definitions in SB-SYS::LOAD-FOREIGN and
3527 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3528 vital for most things which interface to C-level interfaces,
3529 like extensions working with sockets or databases or
3530 Perl-compatible regexes or whatever, and the need to fix
3531 this bug is the main reason that 0.7.1 was released so
3533 ** DEFGENERIC is now choosier about the methods it redefines, so that
3534 reLOADing a previously-LOADed file containing DEFGENERICs does
3535 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3536 cycle now works as it should. (thanks to Alexey Dejneka)
3537 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3538 fixed by Christophe Rhodes. (He actually submitted this patch
3539 months ago, and I delayed until after 0.7.0.)
3540 ** Bug 111 (internal compiler confusion about runtime checks on
3541 FUNCTION types) was fixed by Alexey Dejneka.
3542 * Some internal cleanups (getting rid of variables which aren't
3543 needed now that the byte interpreter is gone) caused the fasl
3544 file format number to change again.
3546 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3547 * major incompatible change: The default fasl file extension, i.e. the
3548 default extension for files produced by COMPILE-FILE, has changed
3549 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3551 ** There are many changes in the implementation of the compiler.
3552 SBCL is now essentially a compiler-only implementation of ANSI
3553 Common Lisp. EVAL still "interprets" a few special cases, but
3554 almost all the interesting cases are handled by creating
3555 a LAMBDA expression, calling COMPILE on it, then calling
3556 FUNCALL on the result.
3557 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3558 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3559 Since the code is newer, there might still be some new bugs
3560 (though not as many as before Martin Atzmueller's fixes:-). But
3561 the new code is substantially simpler and clearer, and hopefully
3562 any remaining bugs will be simpler, less fundamental, and more
3563 fixable then the bugs in the old code.
3564 ** The revised compiler is still a little unsteady on its feet.
3566 *** The debugging information it produces (particularly the names
3567 of FUNCTION objects) is sometimes much less useful than what
3568 the old compiler produced.
3569 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3570 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3571 within a MACROLET) has been temporarily weakened.
3572 ** There are new compiler optimizations for various functions:
3573 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3574 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3575 *** the math functions TRUNCATE, FLOOR, and CEILING
3576 *** the function-of-all-trades COERCE
3577 Mostly these should be transparent, but there's one
3578 potentially-annoying problem (bug 117): when the compiler
3579 inline-expands a function and does type analysis on the result,
3580 it can create control paths which have type mismatches, and
3581 when it can't prove that those control paths aren't taken,
3582 it will issue WARNINGs about the type mismatches. This is
3583 a particular problem in practice for the new sequence functions.
3584 It's not clear how this should be fixed, and for now, a
3585 workaround is given in the entry for 117 in the BUGS file.
3586 ** (Because of the interaction between the two previous items --
3587 occasional inlining problems and new inline expansions -- some
3588 of the new sequence function optimizations won't really kick in
3589 completely until debugging information, and then inlining, are
3590 straightened out in some future version.)
3591 * minor incompatible changes:
3592 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3593 DIRECTORY no longer implicitly promotes NIL slots of its
3594 pathname argument to :WILD. In particular, when you ask for the
3595 contents of a directory (which you used to be able to do without
3596 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3597 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3598 ** changes in behavior that ANSI explicitly defines to be
3599 implementation dependent:
3600 *** The new compiler-only implementation still conforms with ANSI,
3601 but acts a little different than before. Besides the obvious
3602 changes in performance tradeoffs (that the cost per form passed
3603 to EVAL has gone up, and the cost per form executed by EVAL
3604 has gone down), the behavior of the system changes a little
3605 because there are no longer any interpreted function objects.
3606 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3607 e.g. doing COMPILE on the output of interactive DEFUN is
3609 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3611 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3612 has changed from (:CL) to NIL.
3613 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3614 e.g. #\Newline and #\Tab, don't have names specified in the
3615 ANSI Common Lisp standard, is now based on their ASCII symbolic
3616 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3617 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3618 are no longer used for output.
3619 ** changes in internal implementation constants:
3620 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3621 4 million. (If your application spends a lot of time GCing and
3622 you have a lot of RAM, you might want to experiment with
3623 increasing it even more.)
3624 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3625 However, almost all old code should still continue to work without
3626 immediate update, as SB-C-CALL is now a (deprecated) nickname
3628 ** Old operator names in the style DEF-FOO are now deprecated in
3629 favor of new corresponding names DEFINE-FOO, for consistency with
3630 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3631 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3632 internal symbols, but a few supported extensions like
3633 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3634 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3635 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3636 etc. as you get deeper into recursive calls to the debugger
3637 command loop, instead of the old "5]", "5]]", "5]]]"
3638 sequence. (I was motivated to do this when squabbles between
3639 ILISP and SBCL left me very deeply nested in the debugger. In the
3640 short term, this change will probably provoke more ILISP/SBCL
3641 squabbles, but hopefully it will be an improvement in the long run.)
3642 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3643 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3644 function, since declaiming return types involving aliens is
3645 (1) annoyingly messy to do by hand and (2) vital to efficient
3646 compilation of code which calls such functions.
3647 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3648 longer reexported by the SB-EXT package. They're solely useful
3649 for alien code, so it seems more logical that you should get
3650 them from the SB-ALIEN package, not in SB-EXT.
3651 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3652 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3653 features. Instead, the code that they used to control is always
3654 built into the system.
3655 * many other bug fixes
3656 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3657 advantage of the new EVAL-WHEN stuff and to clean them up in
3658 general, and they are now more ANSI-compliant in a number of
3659 ways. Martin Atzmueller is responsible for a lot of this.
3660 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3662 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3663 *** correct ERROR type for various file operations
3664 *** some fixes for Lisp streams
3665 *** DEFMETHOD syntax checking
3666 *** changing old weird representation of debug information as
3667 strings (which, among their other deficiencies, don't transform
3668 correctly when you rename packages, and don't change their
3669 print representation when you change things like *PACKAGE*
3670 and *PRINT-LENGTH*) to symbols and lists of symbols
3671 He also made several improvements and fixed several bugs in DESCRIBE.
3672 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3674 *** misbehavior of WRITE-STRING/WRITE-LINE
3675 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3676 and several other LOOP problems as well
3677 *** DIRECTORY when similar filenames are present
3678 *** DEFGENERIC with :METHOD options
3679 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3680 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3681 *** argument ordering in FIND with :TEST option
3682 *** mishandled package designator argument in APROPOS-LIST
3683 *** various problems in the backquote readmacro
3684 *** a bug in APROPOS
3685 *** probably some others that I'm not describing very well here,
3686 since the CVS log documents them by reference to sbcl-devel
3687 messages, and the SourceForge archives aren't working well.:-(
3688 ** Dan Barlow improved the Alpha port (and is making progress on the
3689 PPC port, for those of you who think different).
3690 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3691 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3692 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3693 continues to work on the SPARC port (for those of you in a position
3694 to look down upon our little PC-compatible boxes from a great height).
3695 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3696 and uses that copy, rather than the current dynamic value, when
3697 it's trying to decide whether to truncate output. Thus e.g.
3698 (let ((*print-lines* 50))
3699 (pprint-logical-block (stream nil)
3701 (let ((*print-lines* 8))
3702 (print (aref possiblybigthings i) stream)))))
3703 should now truncate the logical block only at 50 lines, instead of
3704 often truncating it at 8 lines, as it did before.
3705 * The doc/cmucl/ directory, containing old CMU CL documentation
3706 from the time of the fork, is no longer part of the base system.
3707 SourceForge has shut down its anonymous FTP service, and with it
3708 my original plan for distributing the old CMU CL documentation
3709 there. For now, if you need these files you can download an old
3710 SBCL source release and extract them from it.
3711 * The fasl file version number changed again, for dozens of reasons,
3712 some of which are apparent above.
3714 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3715 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3716 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3717 patch, so that DISASSEMBLE output is much nicer.
3718 * The code in the SB-PROFILE package now seems reasonably stable.
3719 I still haven't decided what the final interface should look like
3720 (I'd like PROFILE to interact cleanly with TRACE, since both
3721 facilities use function encapsulation) but if you have a need
3722 for profiling now, you can probably use it successfully with
3723 the current CMU-CL-style interface.
3724 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3725 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3726 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3727 link now returns the dangling link itself, and for similar
3728 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3729 link itself. (In these cases the old code signalled an error and
3730 looped endlessly, respectively.) Thus, DIRECTORY now works even
3731 in the presence of dangling and cyclic symbolic links.
3732 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3733 is now a supported extension again, since the consensus on
3734 sbcl-devel was that it can be useful for ordinary development
3735 work, not just for debugging SBCL itself.
3736 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3737 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3738 that currently-defined functions might be redefined later with
3739 different return types.
3740 * Hash tables can be printed readably, as inspired by CMU CL code
3741 of Eric Marsden and SBCL code of Martin Atzmueller.
3742 * better error handling in CLOS method combination, thanks to
3743 Martin Atzmueller porting Pierre Mai's CMU CL patches
3744 * more overflow fixes for >16Mbyte I/O buffers
3745 * A bug in READ has been fixed, so that now a single Ctrl-D
3746 character suffices to cause end-of-file on character streams.
3747 In particular, now you only need one Ctrl-D at the command
3748 line (not two) to exit SBCL.
3749 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3751 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3752 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3753 does the right thing.
3754 * The implementation of some type tests, especially for CONDITION
3755 types, is now tidier and maybe faster, due to CMU CL code
3756 originally by Douglas Crosher, ported by Martin Atzmueller.
3757 * Some math functions have been fixed, and there are new
3758 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3759 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3760 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3761 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3762 likely to be built into sbcl-0.7.0.)
3763 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3764 CL, ported to SBCL by Martin Atzmueller)
3765 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3766 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3767 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3768 * There's a new slam.sh hack to shorten the edit/compile/debug
3769 cycle for low-level changes to SBCL itself, and a new
3770 :SB-AFTER-XC-CORE target feature to control the generation of
3771 the after-xc.core file needed by slam.sh.
3772 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3773 COMPILE-FILE is no longer supported, so that now every function
3774 gets an entry point, so that block compilation looks a little
3775 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3776 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3777 returns the number of bytes consed since the system started,
3778 rather than the number consed since the first time the function
3779 was called. (The new definition parallels ANSI functions like
3780 CL:GET-INTERNAL-RUN-TIME.)
3781 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3782 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3783 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3784 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3785 * Fasl file version numbers are now independent of the target CPU,
3786 since historically most system changes which required version
3787 number changes have affected all CPUs equally. Similarly,
3788 the byte fasl file version is now equal to the ordinary
3791 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3792 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3793 is no longer recognized. I apologize for this, because it was
3794 listed in SB-EXT as a supported extension, but I found that
3795 its existing behavior was poorly specified, as well as incorrectly
3796 specified, and it looked like too much of a mess to straighten it
3797 out. I have enough on my hands trying to get ANSI stuff to work..
3798 * many patches ported from CMU CL by Martin Atzmueller, with
3799 half a dozen bug fixes in pretty-printing and the debugger, and
3800 half a dozen others elsewhere
3801 * fixed bug 13: Floating point infinities are now supported again.
3802 They might still be a little bit flaky, but thanks to bug reports
3803 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3804 as flaky as they were.
3805 * The --noprogrammer command line option is now supported. (Its
3806 behavior is slightly different in detail from what the old man
3807 page claimed it would do, but it's still appropriate under the
3808 same circumstances that the man page talks about.)
3809 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3810 are now supported, and enabled by default. Thus, the compiler can
3811 handle many floating point and complex operations much less
3812 inefficiently. (Thus e.g. you can implement a complex FFT
3814 * The compiler now detects type mismatches between DECLAIM FTYPE
3815 and DEFUN better, and implements CHECK-TYPE more correctly, and
3816 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3817 thanks to patches from Martin Atzmueller.
3818 * various fixes to make the cross-compiler more portable to
3819 ANSI-conforming-but-different cross-compilation hosts (notably
3820 Lispworks for Windows, following bug reports from Arthur Lemmens)
3821 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3822 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3823 fixed thanks to Pierre Mai's CMU CL patches.
3824 * fixes to keep the system from overflowing internal counters when
3825 it tries to use i/o buffers larger than 16M bytes
3826 * fixed bug 45a: Various internal functions required to support
3827 complex special functions have been merged from CMU CL sources.
3828 (When I was first setting up SBCL, I misunderstood a compile-time
3829 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3830 * improved support for type intersection and union, fixing bug 12
3831 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3832 more obscure bugs as well
3833 * some steps toward byte-compiling non-performance-critical
3834 parts of the system, courtesy of patches from Martin Atzmueller
3835 * Christophe Rhodes has made some debian packages of sbcl at
3836 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3837 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3838 stable, but are nonetheless usable. When he's ready, I'd be happy
3839 to add them to the SourceForge "File Releases" section. (And if
3840 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3841 * new fasl file format version number (because of changes in
3842 internal representation of (OR ..) types to accommodate the new
3843 support for (AND ..) types, among other things)
3845 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3846 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3848 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3850 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3851 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3852 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3853 * DESCRIBE now works on structure objects again.
3854 * Most function call argument type mismatches are now handled as
3855 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3856 know whether the function will be redefined before the call is
3857 executed. (The compiler could flag local calls with full WARNINGs,
3858 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3859 it doesn't keep track of enough information to know whether calls
3860 are local in this sense.)
3861 * Compiler output is now more verbose, with messages truncated
3862 later than before. (There should be some supported way for users
3863 to override the default verbosity, but I haven't decided how to
3864 provide it yet, so this behavior is still controlled by the internal
3865 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3866 src/compiler/ir1util.lisp.)
3867 * Fasl file format version numbers have increased again, because
3868 support for the Gray streams extension changes the layout of the
3869 system's STREAM objects.
3870 * The Gray subclassable streams extension now works, thanks to a
3871 patch from Martin Atzmueller.
3872 * The full LOAD-FOREIGN extension (not just the primitive
3873 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3874 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3875 but like most other programs, it defaults to copying the Unix
3876 environment from the original process instead of starting the
3877 new process in an empty environment.
3878 * Extensions which manipulate the Unix environment now support
3879 an :ENVIRONMENT keyword option which doesn't smash case or
3880 do other bad things. The CMU-CL-style :ENV option is retained
3881 for porting convenience.
3882 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3883 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3885 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3887 * A patch from Martin Atzmueller seems to have solved the SIGINT
3888 problem, and as far as we know, signal-handling now works cleanly.
3889 (If you find any new bugs, please report them!)
3890 * The system no longer defaults Lisp source file names to types
3891 ".l", ".cl", or ".lsp", but only to ".lisp".
3892 * The compiler no longer uses special default file extensions for
3893 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3894 seems to expect a single default extension for all compiled code,
3895 and there's no compelling reason to try to stretch the standard
3896 to allow two different extensions.) Instead, byte-compiled files
3897 default to the same extension as native-compiled files.
3898 * Fasl file format version numbers have increased again, because
3899 a rearrangement of internal implementation packages made some
3900 dumped symbols in old fasl files unreadable in new cores.
3901 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3902 many fewer weird special cases.
3903 * Bug #17 (differing COMPILE-FILE behavior between logical and
3904 physical pathnames) has been fixed, and some related misbehavior too,
3905 thanks to a patch from Martin Atzmueller.
3906 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3907 by Tim Moore, ported to SBCL by Martin Atzmueller.
3908 * Martin Atzmueller fixed several filesystem-related problems,
3909 including bug #36, in part by porting CMU CL patches, which were
3910 written in part by Paul Werkowski.
3911 * More compiler warnings in src/runtime/ are gone, thanks to
3912 more patches from Martin Atzmueller.
3913 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3916 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3918 * DESCRIBE now works on CONDITION objects.
3919 * The debugger now handles errors which arise when trying to print
3920 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3922 * The build system now uses an additional file, customize-target-features.lisp,
3923 to allow local modifications to the target *FEATURES* list. (The point of
3924 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3925 debugging features enabled, without having to worry about propagating it
3926 into everyone's system when I do a "cvs update".) When no
3927 customize-target-features.lisp file exists, the target *FEATURES* list
3928 should be constructed the same way as before.
3929 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3930 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3931 warn about behavior which is undefined under the ANSI spec. Note: This
3932 is specified by ANSI, but it's not very popular with programmers.
3933 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3934 macro in the SBCL sources for an example of a workaround which you
3935 might use to make portable ANSI-standard code which does what you want.
3936 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3937 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3938 when it's used as a non-toplevel form. (This is required in order
3939 to implement the DEFCONSTANT-EQX macro.)
3940 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3941 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3942 funny things, due to bugs in the implementation of EVAL-WHEN
3943 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3944 non-toplevel DEFCONSTANTs are uncommon.)
3945 * The core file version number and fasl file version number have been
3946 incremented, because the old noncompliant DEFCONSTANT behavior involved
3947 calling functions which no longer exist, and because I also took the
3948 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3949 * fixed bug 1 (error handling before read-eval-print loop starts), and
3950 redid debugger restarts and related debugger commands somewhat while
3952 ** The QUIT debugger command is gone, since it did something
3953 rather different than the SB-EXT:QUIT command, and since it never
3954 worked properly outside the main toplevel read/eval/print loop.
3955 Invoking the new TOPLEVEL restart provides the same functionality.
3956 ** The GO debugger command is also gone, since you can just invoke
3957 the CONTINUE restart directly instead.
3958 ** The TOP debugger command is also gone, since it's redundant with the
3959 FRAME 0 command, and since it interfered with abbreviations for the
3961 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3963 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3964 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3965 some porting work by Martin Atzmueller)
3966 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3967 with a boatload of patches which he ported to SBCL. Now that those
3969 ** The system tries to make sure that its low-priority messages
3970 are prefixed by semicolons, to help people who like to use
3971 syntax highlighting in their ILISP buffer. (This patch
3972 was originally due to Raymond Toy.)
3973 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3974 patches originally written by Raymond Toy.
3975 ** The compiler understands coercion between single-value and
3976 multiple-VALUES type expressions better, getting rid of some very
3977 weird behavior, thanks to patches originally by Robert MacLachlan
3978 and Douglas Crosher.
3979 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3980 lambda lists, thanks to a patch originally by Pierre Mai.
3981 ** The system no longer bogusly warns about "abbreviated type
3983 ** The compiler gets less confused by inlining and RETURN-FROM,
3984 thanks to some patches originally by Tim Moore.
3985 ** The system no longer hangs when dumping circular lists to fasl
3986 files, thanks to a patch originally from Douglas Crosher.
3987 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3988 undefined function error.
3989 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3990 in structure slot initforms) for now, documented workaround instead:-|
3991 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3992 * fixed bug 5: added stubs for various Gray stream functions called
3993 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3994 installed, at least appropriate type errors are generated
3995 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3996 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3997 consistently in DEFMETHOD forms.
3998 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3999 it doesn't seem to affect SBCL after all
4000 * The C runtime system now builds with better optimization and many
4001 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4003 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4005 * The system is now under CVS at SourceForge (instead of the
4006 CVS repository on my home machine).
4007 * The new signal handling code has been tweaked to treat register
4008 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4009 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4010 downstream, e.g. in debug-int.lisp, has implicit dependencies
4011 on the unsignedness of integer representation of machine words,
4012 and that caused the system to bomb out with infinite regress
4013 when trying to recover from type errors involving signed values,
4014 e.g. (BUTLAST '(1 2 3) -1).)
4015 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4016 (This was one of the bugs Peter Van Eynde reported back in July.)
4017 * The system now uses code inspired by Colin Walters' O(N)
4018 implementation of MAP (from the cmucl-imp@cons.org mailing
4019 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4020 inline the MAP operation, and there is more than one
4021 sequence argument to the MAP call (so that it can't just
4022 do ETYPECASE once and for all based on the type of the
4023 single sequence argument). (The old non-inline implementation
4024 of the general M-argument sequence-of-length-N case required
4025 O(M*N*N) time when any of the sequence arguments were LISTs.)
4026 * The QUIT :UNIX-CODE keyword argument has been renamed to
4027 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4028 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4030 * Raymond Wiker's patches to port dynamic loading from Linux to
4031 FreeBSD have been added.
4032 * The BUGS file is now more nearly up to date, thanks in large part
4033 to Martin Atzmueller's review of it.
4034 * The debugger now flushes standard output streams before it begins
4035 its output ("debugger invoked" and so forth).
4036 * The core version number and fasl file version number have both
4037 been incremented, because of incompatible changes in the layout
4039 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4040 used to not be called for a saved Lisp image.)
4041 * Martin Atzmueller's version of a patch to fix a compiler crash,
4042 as posted on sbcl-devel 13 September 2000, has been installed.
4043 * Instead of installing Martin Atzmueller's patch for the
4044 compiler transform for SUBSEQ, I deleted the compiler transform,
4045 and transforms for some similar consing operations.
4046 * A bug in signal handling which kept TRACE from working on OpenBSD
4048 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4049 away by constant folding
4050 * The system now defines its address space constants in one place
4051 (in the Lisp sources), and propagates them automatically elsewhere
4052 (through GENESIS and the sbcl.h file). Therefore, patching the
4053 address map is less unnecessarily tedious and error-prone. The
4054 Lisp names of address space constants have also been systematized.
4055 * CVS tags like dollar-Header-dollar have been removed from
4056 the sources, because they have never saved me trouble and
4057 they've been source of trouble working with patches and other
4058 diff-related operations.
4059 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4060 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4062 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4064 * The system has been ported to OpenBSD.
4065 * The system now compiles with a simple "sh make.sh" on the systems
4066 that it's supported on. I.e., now you no longer need to tweak
4067 text in the target-features.lisp-expr and symlinks in src/runtime/
4068 by hand, the make.sh takes care of it for you.
4069 * The system is no longer so grossly inefficient when compiling code
4070 involving vectors implemented as general (not simple) vectors (VECTOR T),
4071 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4072 which dares to use the various sequence functions on non-simple
4073 vectors, takes less of a performance hit.
4074 * There is now a primitive type predicate VECTOR-T-P
4075 to test for the (VECTOR T) type, so that e.g.
4076 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4077 can now be compiled with some semblance of efficiency. (The old code
4078 turned the type declaration into a full call to %TYPEP at runtime!)
4079 * AREF on (VECTOR T) is still not fast, since it's still compiled
4080 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4081 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4082 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4083 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4084 without making full calls to SUBTYPEP at runtime.
4085 (Some analogous efficiency issues for non-simple vectors specialized to
4086 element types other than T, or for non-simple multidimensional arrays,
4087 have not been addressed. They could almost certainly be handled the
4088 same way if anyone is motivated to do so.)
4089 * The changes in array handling break binary compatibility, so
4090 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4091 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4092 instead of (VALUES T T).
4093 * By following the instructions that Dan Barlow posted to sbcl-devel
4094 on 2 July 2000, I was able to enable primitive dynamic object
4095 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4096 functionality is not implemented (since it calls ld to resolve
4097 library references automatically, requiring RUN-PROGRAM for its
4098 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4099 try to resolve library references) is now supported.
4100 * The system now flushes the standard output streams when it terminates,
4101 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4102 them at several other probably-convenient times, e.g. in each pass of
4103 the toplevel read-eval-print loop, and after evaluating a form given
4104 as an "--eval" command-line option. (These changes were motivated by a
4105 discussion of stream flushing issues on cmucl-imp in August 2000.)
4106 * The source transform for TYPEP of array types no longer assumes
4107 that an array whose element type is a not-yet-defined type
4108 is implemented as an array of T, but instead punts, so that the
4109 type will be interpreted at runtime.
4110 * There is now some support for cross-compiling in make.sh: each of
4111 the phases of make.sh has its own script. (This should be transparent
4112 to people doing ordinary, non-cross-compile builds.)
4113 * Since my laptop doesn't have hundreds of megabytes of memory like
4114 my desktop machine, I became more motivated to do some items on
4115 my to-do list in order to reduce the size of the system a little:
4116 ** Arrange for various needed-only-at-cold-init things to be
4117 uninterned after cold init. To support this, those things have
4118 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4119 symbols with such names are now uninterned after cold init).
4120 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4121 fasl files for cold load.
4122 ** Remove the old compiler structure pooling code (which used to
4123 be conditional on the target feature :SB-ALLOC) completely.
4124 ** Redo the representation of some data in cold init to be more compact.
4125 (I also looked into supporting byte compiled code at bootstrap time,
4126 which would probably reduce the size of the system a lot, but that
4127 looked too complicated, so I punted for now.)
4128 * The maximum signal nesting depth in the src/runtime/ support code has
4129 been reduced from 4096 to 256. (I don't know any reason for the very
4130 large old value. If the new smaller value turns out to break something,
4131 I'll probably just bump it back up.)
4132 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4134 * Many, many bugs reported by Peter Van Eynde have been added to
4135 the BUGS list; some have even been fixed.
4136 * While enabling dynamic object file loading, I tried to make the
4137 code easier to understand, renaming various functions and variables
4138 with less ambiguous names, and changing some function calling
4139 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4140 * While trying to figure out how to do the OpenBSD port, I tried to
4141 clean up some of the code in src/runtime/. In particular, I dropped
4142 support for non-POSIX signal handling, added various comments,
4143 tweaked the code to reduce the number of compilation warnings, and
4144 renamed some files to increase consistency.
4145 * To support the new automatic configuration functionality in make.sh,
4146 the source file target-features.lisp-expr has been replaced with the
4147 source file base-target-features.lisp-expr and the machine-generated
4148 file local-target-features.lisp-expr.
4149 * fixed a stupid quoting error in make.sh so that using CMU CL
4150 "lisp -batch" as cross-compilation host works again
4152 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4154 * DESCRIBE no longer tries to call itself recursively to describe
4155 bound/fbound values, so that it no longer fails on symbols which are
4156 bound to themselves (like keywords, T, and NIL).
4157 * DESCRIBE now works on generic functions.
4158 * The printer now prints less-screwed-up representations of closures
4159 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4160 * A private symbol is used instead of the :EMPTY keyword previously
4161 used to mark empty slots in hash tables. Thus
4162 (DEFVAR *HT* (MAKE-HASH-TABLE))
4163 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4164 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4165 now does what ANSI says that it should. (You can still get
4166 similar noncompliant behavior if bang on the hash table
4167 implementation with all the symbols you get back from
4168 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4169 This breaks binary compatibility, since tests for equality to
4170 :EMPTY are wired into things like the macroexpansion of
4171 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4173 * There's now a minimal placeholder implementation for CL:STEP,
4174 as required by ANSI.
4175 * An obscure bug in the interaction of the normal compiler, the byte
4176 compiler, inlining, and structure predicates has been patched
4177 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4178 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4180 * Missing ordinary arguments in a macro call are now detected even
4181 when the macro lambda list contains &KEY or &REST.
4182 * The debugger no longer complains about encountering the top of the
4183 stack when you type "FRAME 0" to explicitly instruct it to go to
4184 the top of the stack. And it now prints the frame you request even
4185 if it's the current frame (instead of saying "You are here.").
4186 * As specified by ANSI, the system now always prints keywords
4187 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4189 * The default initial SIZE of HASH-TABLEs is now smaller.
4190 * Type information from CLOS class dispatch is now propagated
4191 into DEFMETHOD bodies, so that e.g.
4192 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4194 is now basically equivalent to
4195 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4196 (DECLARE (TYPE SINGLE-FLOAT X))
4198 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4199 operation, without having to do run-time type dispatch.
4200 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4201 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4202 * Since I don't seem to be making big file reorganizations very often
4203 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4204 is overflowing my ability to conveniently back them up), I've finally
4205 checked the system into CVS. (The CVS repository is on my home system,
4206 not at SourceForge -- putting it on SourceForge might come later.)
4207 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4208 high-level GC-NOTIFY-FOO functions send their output. (There's
4209 still very little control of where low-level verbose GC functions
4210 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4211 less than it used to -- the GC-NOTIFY-FOO functions are now under
4212 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4213 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4214 in only one place in the source code, and propagates it automatically
4215 everywhere that it's needed. Thus e.g. when I bump the version from
4216 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4217 * The C source files now include boilerplate legalese and documentation
4218 at the head of each file (just as the Lisp source files already did).
4219 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4220 to his page will be replaced with a link to his new CLiki service.
4222 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4224 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4225 * The build process now looks for GNU make under the default name "gmake",
4226 instead of "make" as it used to. If GNU make is not available as "gmake"
4227 on your system, you can change this default behavior by setting the
4228 GNUMAKE environment variable.
4229 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4230 can build without error under CMU CL.
4232 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4234 * There is now a partial SBCL user manual (with some new text and some
4235 text cribbed from the CMU CL manual).
4236 * The beginnings of a profiler have been added (starting with the
4237 CMU CL profiler and simplifying and cleaning up). Eventually the
4238 main interface should be through the TRACE macro, but for now,
4239 it's still accessed through vaguely CMU-CL-style functions and macros
4240 exported from the package SB-PROFILE.
4241 * Some problems left over from porting CMU CL to the new
4242 cross-compilation bootstrap process have been cleaned up:
4243 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4244 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4245 part when getting rid of the compiler *BACKEND* stuff.)
4246 ** The value of *NULL-TYPE* was screwed up, because it was
4247 being initialized before the type system knew the final
4248 definition of the 'NULL type. This screwed up several key
4249 optimizations in the compiler, causing inefficiency in all sorts
4250 of places. (I found it because I wanted to understand why
4251 GET-INTERNAL-RUN-TIME was consing.)
4252 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4253 PROCLAIM FTYPE information. Unfortunately this broke binary
4254 compatibility again, since now the forms output by DEFGENERIC
4255 to refer to functions which didn't exist in 0.6.3.
4256 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4257 can use the new (as of 0.6.3) transform for SOME into MAP into
4259 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4260 values to (INTEGER 0 1000000), so that the time code will no longer
4261 occasionally get blown up by Linux returning 1000000 microseconds
4262 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4263 its output conform to the ANSI spec. (Alas, this makes its output
4264 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4265 * A full call to MAP NIL with a single sequence argument no longer conses.
4266 * fixes to problems pointed out by Martin Atzmueller:
4267 * The manual page no longer talks about multiprocessing as though
4268 it were currently supported.
4269 * The ILISP support patches have been removed from the distribution,
4270 because as of version 5.10.1, ILISP now supports SBCL without us
4271 having to maintain patches.
4272 * added a modified version of Raymond Toy's recent CMU CL patch for
4273 EQUALP comparison of HASH-TABLE
4275 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4277 * The system still can't cross-compile itself with
4278 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4279 entails), but at least it can compile more of itself that way
4280 than it used to be able to, and various buglets which were uncovered
4281 by trying to cross-compile itself that way have now been fixed.
4282 * This release breaks binary compatibility again. This time
4283 at least I've incremented the FASL file format version to 2, so that the
4284 problem can be detected reliably instead of just causing weird errors.
4285 * various new style warnings:
4286 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4287 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4288 ** using the lexical binding of a variable named in the *FOO* style
4289 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4290 as specified by ANSI.
4291 * *RANDOM-STATE* is no longer automatically initialized from
4292 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4293 default behavior of the system is to repeat its behavior every time
4294 it's run. If you'd like to change this behavior, you can always
4295 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4296 old convention there was no comparably easy way to get the system to
4297 repeat its behavior every time it was run.
4298 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4299 FTYPE declarations has been removed, in favor of their ANSI
4300 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4301 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4302 the types of their sequence arguments can be determined at compile time.
4303 This is done through a new open code expansion for MAP which eliminates
4304 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4305 argument types can be determined at compile time.
4306 * The optimizer now transforms COERCE into an identity operation when it
4307 can prove that the coerced object is already of the correct type. (This
4308 can be a win for machine generated code, including the output of other
4309 optimization transforms, such as the MAP transform above.)
4310 * Credit information has been moved from source file headers into CREDITS.
4311 * Source file headers have been made more standard.
4312 * The CASE macro now compiles without complaining even when it has
4315 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4317 * (Note that the way that the PCL macroexpansions were rewritten
4318 to accommodate the change in DEFGENERIC below breaks binary
4319 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4320 not run under sbcl-0.6.2. Once we get out of alpha releases,
4321 i.e. hit release 1.0.0, we'll probably try to maintain binary
4322 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4323 and sbcl-1.4.4. Until then, however, it might be fairly common
4324 for maintenance releases to break binary compatibility.)
4325 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4327 * The reporting of unhandled signals has been changed to print some
4328 explanatory text as well as the report form. (Previously only
4329 the report form was printed.)
4330 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4331 it defines, so that the compiler no longer issues undefined function
4332 warnings for compiled-but-not-yet-loaded generic functions.
4333 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4334 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4335 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4336 section of the ANSI spec are supported.
4337 * The "" nickname for the "KEYWORD" package has been removed.
4338 The reader still handles symbol tokens which begin with a package marker
4339 as keywords, but it doesn't expose its mechanism for doing so in the
4340 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4341 * The system now issues STYLE-WARNINGs for contradictory TYPE
4342 proclamations. (Warnings for contradictory FTYPE proclamations would
4343 be nice too, but those can't be done usefully unless the type system
4344 is made smarter about FUNCTION types.)
4345 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4346 systematized, so that "*target-*.lisp is supposed to exist only on the
4347 target and imply that there's a related file which exists on the
4348 host, and *host-*.lisp is supposed to exist only on the host and imply
4349 that there's a related file which exists on the target. This involves a
4350 lot of renaming. Hopefully the acute confusion caused by the renaming
4351 will be justified by the reduction in chronic confusion..
4352 ** runtime-type.lisp -> early-target-type.lisp
4353 ** target-type.lisp -> late-target-type.lisp
4354 ** early-host-format.lisp -> early-format.lisp
4355 ** late-host-format.lisp -> late-format.lisp
4356 ** host-error.lisp -> misc-error.lisp
4357 ** early-error.lisp -> early-target-error.lisp
4358 ** late-error.lisp -> late-target-error.lisp
4359 ** host-defboot.lisp -> early-defboot.lisp
4360 ** code/misc.lisp -> code/target-misc.lisp
4361 ** code/host-misc.lisp -> code/misc.lisp
4362 ** code/numbers.lisp -> code/target-numbers.lisp
4363 ** code/early-numbers.lisp -> numbers.lisp
4364 ** early-host-type.lisp -> early-type.lisp
4365 ** late-host-type.lisp -> late-type.lisp
4366 ** host-typep.lisp -> typep.lisp
4367 ** load.lisp -> target-load.lisp
4368 ** host-load.lisp -> load.lisp
4369 ** host-disassem.lisp -> disassem.lisp
4370 ** host-insts.lisp -> insts.lisp
4371 ** byte-comp.lisp -> target-byte-comp.lisp
4372 ** host-byte-comp.lisp -> byte-comp.lisp
4373 ** host-signal.lisp -> signal.lisp
4374 ** host-defstruct.lisp -> defstruct.lisp
4375 ** late-target-type.lisp -> deftypes-for-target.lisp
4376 Furthermore, several other previously target-only files foo.lisp (e.g.
4377 hash-table.lisp and random.lisp) have been split into a target-and-host
4378 foo.lisp file and a target-only target-foo.lisp file, with their key type
4379 definitions in the target-and-host part, so that the cross-compiler will
4380 know more about target types.
4381 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4382 gone away. In their place are various *BACKEND-FOO* variables
4383 corresponding to the slots of the old structure.
4384 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4385 into the target SBCL has been fixed.
4386 * The chill.lisp system for loading cold code into a running SBCL
4388 * Support for the CMU CL "scavenger hook" extension has been removed.
4389 (It was undocumented and unused in the CMU CL sources that SBCL was
4390 derived from, and stale in sbcl-0.6.1.)
4391 * Various errors in the cross-compiler type system were detected
4392 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4393 (enabling various consistency checks). Many of them were fixed,
4394 but some hard problems remain, so the compiler is back to
4395 running without *TYPE-SYSTEM-INITIALIZED* for now.
4396 * As part of the cross-compiler type system cleanup, I implemented
4397 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4398 * I have started adding UNCROSS calls throughout the type system
4399 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4400 on cross-compiler input files will be able to go away, and various
4402 * CONSTANTP now returns true for quoted forms (as explicitly required
4405 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4407 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4408 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4409 (MAKE-LIST -1). In the longer term, it ought to become true
4410 that declarations are assertions even at SAFETY 1. For now, it's not
4411 quite true even at SAFETY 3, but it's at least more nearly true..
4412 (Note that this change seems to increases the size of the system by
4413 O(5%) and to decrease the speed of the compiler by 20% or more.)
4414 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4415 for the problem of printing dozens of lines of distracting information
4416 about low-level system machinery as part of the top stack frame
4417 on entry to the debugger when an undefined function was called.
4418 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4419 is not reset to COMMON-LISP-USER.
4420 * Compilation of stuff related to dyncount.lisp has been made conditional
4421 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4422 smaller. The various dyncount-related symbols have been moved into
4423 a new "SB-DYNCOUNT" package.
4424 * tty-inspect.lisp has been renamed to inspect.lisp.
4425 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4426 feature has gone away. (When we eventually port to other flavors of
4427 libc and/or Unix, we'll try to make the differences between flavors
4428 invisible at the user level.)
4429 * Various other *FEATURES* tags, and/or their associated conditionals,
4430 have been removed if obsolescent, or given better documentation, or
4431 sometimes given more-mnemonic names.
4433 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4435 * tidied up "make.sh" script
4436 * tidied up system directory structure
4437 * better "clean.sh" behavior
4438 * added doc/FOR-CMUCL-DEVELOPERS
4439 * many many small tweaks to output format, e.g. removing possibly-confusing
4440 trailing #\. character in DESCRIBE-INSTANCE
4441 * (EQUALP #\A 'A) no longer signals an error.
4442 * new hashing code, including EQUALP hashing
4443 * tidied up Lisp initialization and toplevel
4444 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4445 * command line argument processing
4446 * added POSIX-GETENV function to deal with Unix-ish environment variables
4447 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4448 terminating SBCL on EOF
4449 * non-verbose GC by default
4450 * There is no more "sbcl" shell script; the sbcl file is now the C
4451 runtime executable (just like CMU CL).
4452 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4454 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4455 symbol and package information as Lisp native symbol and package objects
4456 instead of strings naming symbols and strings naming packages. This way,
4457 whenever packages are renamed (as in warm init), debug information is
4458 transformed along with everything else.
4459 * tweaked the optimization policy declarations which control the building
4460 of SBCL itself. Now, among other things, the system no longer saves
4461 source location debugging information. (This helps two problems at once
4462 by reducing SBCL size and by keeping SBCL from trying to look for its
4463 sources -- which may not exist -- when reporting errors.)
4464 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4465 debugging and testing purposes
4466 * cleaned up printing, making the printer call PRINT-OBJECT for
4467 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4468 methods, giving nearly-ANSI behavior
4469 * converted almost all special variables to use *FOO* naming convention
4470 * deleted PARSE-TIME functionality, since it can be done portably
4471 * moved some files out of cold init into warm init
4472 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4474 * regularized formatting of source files
4475 * added an install.sh script
4476 * fixed ridiculous memory usage of cross-compiler by making
4477 compiler/alloc.lisp not try to do pooling unless it can hook
4478 itself into the GC of the cross-compilation host. Now the system
4479 builds nicely on my old laptop.
4480 * added :SB-ALLOC in target-features.lisp-expr
4481 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4482 was not implemented)
4483 * re-did condition handling and note reporting in the compiler. Notes
4484 are no longer handled by signalling conditions. Style warnings
4485 and warnings are handled more correctly and reported in such a way
4486 that it's easy to find one or the other in your output (so that you
4487 can e.g. figure out which of many problems caused COMPILE-FILE to
4489 * changed the severity of several compiler warnings from full WARNING
4490 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4491 compiler note reporting so that it doesn't use the condition system
4492 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4493 COMPILE-FILE command)
4494 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4495 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4496 no longer supported, since their implementation was deeply intertwingled
4497 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4498 * removed bogus "support" for compiler macros named (SETF FOO), and
4499 removed the compiler macro for SETF INFO (but only after making a fool
4500 of myself on the cmucl-imp mailing list by posting a bogus patch for
4501 DEFINE-COMPILER-MACRO..)
4502 * Compiled files containing forms which have side effects on the Lisp
4503 reader (such as DEFPACKAGE forms) are now handled more correctly.
4504 (Compiler queuing of top level lambdas has been suppressed by setting
4505 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4506 * deleted various currently-unused source files, e.g. gengc.lisp. They
4507 may be added back at some point e.g. when porting to other architectures,
4508 but until they are it's distracting to distribute them and to try to
4510 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4511 there were so many of them they're just distractions, and UNCROSS is
4512 known to be able to handle the current sources
4513 * moved PROFILE functionality into TRACE, so that it will be clear
4514 how the wrapping and unwrapping of functions when you profile them
4515 interacts with the wrapping and unwrapping of functions when you
4516 trace them. (Actually, the functionality isn't there yet, but at least
4517 the interface specification is there. Hopefully, the functionality will
4518 arrive with some future maintenance release.)
4519 * removed host-oops.lisp
4520 * changed signature of QUIT function to allow UNIX-CODE argument
4521 * fixed READ-SEQUENCE bug
4522 * tweaked verbose GC output so that it looks more like the progress
4523 output that ANSI specifies for functions like LOAD
4524 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4525 * added <http://sbcl.sourceforge.com> to the banner information printed by