1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
3 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
4 an executable in the search path, and does so in the child
5 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
6 been removed; it can be found in the CVS history, for somebody who
7 needs that search behavior (see the manual).
8 * minor incompatible change: compiler policy re. weakening type
9 checks has changed: now type checks are weakened on if SAFETY < 2
11 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
12 unparsing of directory pathnames as files. Analogously,
13 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
14 filename to parse into a directory pathname.
15 * enhancement: RUN-PROGRAM allows unicode arguments and environments
16 to be used (using the default stream external format), and allows
17 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
18 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
19 strings and vectors whose element-type or simplicity is not fully
20 known at compile-time.
21 * optimization: STRING-TRIM and related functions no longer allocate
22 a new string when no trimming needs to be performed. These functions
23 are also faster than before when the input string has been declared
25 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
26 * optimization: READ-LINE is significantly faster for files containing
28 * optimization: non-open coded uses of character comparison operators
29 (e.g. char=) no longer cons when called with more than one parameter
30 on platforms supporting dynamic-extent allocation.
31 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
32 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
33 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
35 * bug fix: some sequence functions elided bounds checking when
37 * bug fix: too liberal weakening of union-type checks when SPEED >
39 * bug fix: more bogus fixnum declarations in ROOM implementation
42 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
43 * new feature: MAKE-HASH-TABLE now experimentally accepts a
44 :SYNCHRONIZED argument, which makes the hash-table safe for
45 concurrent accesses (but not iteration.) See also:
46 SB-EXT:WITH-LOCKED-HASH-TABLE, and
47 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
48 * optimization: CONCATENATE on strings is an order of magnitue faster
49 in code compiled with (> SPEED SPACE).
50 * optimization: SUBSEQ is ~50% faster on lists.
51 * bug fix: bug 417 fixed -- source location reporting is now more robust.
52 * bug fix: SUBSEQ on a list will now correctly signal an error if if
53 END is smaller then START.
54 * bug fix: SB-PROFILE will no longer report extra consing for nested
55 calls to profiled functions.
56 * bug fix: ROOM implementation had bogus fixnum declarations which
57 could cause type-errors when calling ROOM on large images.
58 * bug fix: if file compilation is aborted, the partial fasl is now
59 deleted, and COMPILE-FILE returns NIL as the primary value.
60 * bug fix: number of thread safety issues relating to SBCL's internal
61 hash-table usage have been fixed.
62 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
63 be returned from its body when the values were being returned
64 using unknown-values return convection and the W-P-O was wrapped
65 inside an UNWIND-PROTECT.
66 * bug fix: sb-posix should now compile again under Windows, enabling
69 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
70 * incompatible change: hash-table accessor functions are no longer
71 automatically protected by locks. Concurrent accesses on the same hash-table
72 from multiple threads can give inconsistent results or even corrupt the
73 hash-table completely. Multi-threaded applications should do their own
74 locking at the correct granularity. In the current implementation it is
75 still safe to have multiple readers access the same table, but it's not
76 guaranteed that this property will be maintained in future releases.
77 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
78 and will signal an error at runtime.
79 * enhancement: SB-THREAD package now exports a semaphore interface.
80 * enhancement: CONS can now stack-allocate on x86 and
81 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
83 * enhancement: nested lists can now be stack allocated on
84 platforms providing stack allocation support.
85 * enhancement: dynamic-extent support has been extended to support
86 cases where there are multiple possible sources for the stack
88 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
89 if the mutex is uncontested on Linux.
90 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
91 as specified by the CLHS. (thanks to Tobias Rittweiler)
92 * bug fix: NaN comparison now works on x86-64.
93 * bug fix: CLOSE :ABORT T on a stream with pending output now
95 * bug fix: instances of non-standard metaclasses using standard
96 instance structure protocol sometimes missed the slot type checks
98 * bug fix: known functions can be inlined in high-debug code.
100 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
101 * minor incompatible change: the MSI installer on Windows no longer
102 associates .lisp and .fasl files with the installed SBCL.
103 * minor incompatible change: :UNIX is no longer present in *FEATURES*
104 on Windows. (thanks to Luis Oliviera)
105 * new feature: SB-CLTL2 contrib module now implements
106 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
107 * optimization: scavenging weak pointers is now more efficient,
108 requiring O(1) instead of O(N) per weak pointer to identify
109 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
110 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
111 method bodies using constant slot-names when the first argument is
112 a specializer parameter for the method.
113 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
114 names, when the class of the instance is a direct instance
115 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
116 as slow as the constant slot-name case.)
117 * optimization: member type construction is now O(N) instead
119 * optimization: UNION and NUNION are now O(N+M) for large
120 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
122 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
123 now more readable in environments like Slime which display it.
124 (thanks to Tobias C. Rittweiler)
125 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
126 was able to infer the correct argument type for the object on which
127 the CAS operation was being performed.
128 * bug fix: copy propagation interfered with parallel assignment
129 semantics in local calls. (reported by Paul Khuong)
130 * bug fix: the signed modular fixnum shift compiled to wrong code on
131 x86 and x86-64. (spotted by a slight modification to some of
134 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
135 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
136 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
137 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
139 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
140 and generic function now print the object identity for anonymous
141 instances. (thanks to Attila Lendvai)
142 * enhancement: as an extension to MOP, SBCL now supports SETF of
143 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
144 (thanks to Attila Lendvai)
145 * enhancement: sb-cover has an alternate annotation mode which puts
146 the form annotations at the car of the form, instead of annotating
147 it completely. The benefit of this mode is that it shows better
148 which forms were instrumented by the compiler.
149 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
150 by printer control variables. (reported by Dan Corkill)
151 * bug fix: system leaked memory when delayed output was performed by
152 the OS in smaller chunks then expected. (thanks to David Smith)
153 * bug fix: system leaked memory when file streams were not closed
155 * bug fix: large objects written to slow streams that were modified
156 after the write could end up with the modified state written to
157 the underlying file descriptor.
158 * bug fix: multiple threads operating in parallel on the same stream
159 could cause buffer-overflows.
160 * bug fix: source location information is stored correctly
161 (broken since 1.0.6). This bug would generally show up as the
162 Slime debugger higlighting the wrong form when the sldb-show-source
164 * bug fix: Branch forms are again annotated as branches in the sb-cover
166 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
167 on x86-64. (reported by Gregory Vanuxem)
168 * bug fix: EQUALP could return wrong results for structures with raw
169 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
170 word). (reported by Vjacheslav Fyodorov)
171 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
174 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
175 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
176 atomic compare-and-swap operations on threaded platforms.
177 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
178 allows assining a global minimum value to optimization qualities
179 (overriding proclamations and declarations).
180 * enhancement: closed over variables can be stack-allocated on x86
182 * performance bug fix: GETHASH and (SETF GETHASH) are once again
184 * optimization: slot definition lookup is now O(1). This speeds up
185 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
186 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
187 * optimization: ASSOC and MEMBER can now be open-coded for all
188 combinations of keyword arguments when second argument is constant
189 and SPEED >= SPACE. In other cases a specialized version is
191 * bug fix: using obsoleted structure instances with TYPEP and
192 generic functions now signals a sensible error.
193 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
194 (reported by Kristoffer Kvello)
195 * bug fix: backtrace construction is now more careful when making
196 lisp-objects from pointers on the stack, to avoid creating bogus
197 objects that can be seen by the GC.
198 * bug fix: defaulting of values in contexts expecting more than 7
199 variables now works on x86-64. (reported by Christopher Laux)
200 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
202 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
203 as the property-list of a symbol.
204 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
205 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
206 situations requested, are once again file-compileable. (reported
209 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
210 * MOP improvement: support for user-defined subclasses of
211 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
212 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
213 the expansion of DEFMETHOD, is responsible for generating a form
214 which creates a list of specializers when evaluated. Additional
215 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
216 debugging and introspective support.
217 * minor incompatible change: the (unsupported) spinlock interface
218 has changed: free spinlock now has the value NIL, and a held spinlock
219 has the owning thread as its value.
220 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
221 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
223 * enhancement: name of a socket-stream is now "a socket" instead of
225 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
226 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
227 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
228 for generic functions. (Thanks to Tobias C. Rittweiler)
229 * optimization: bignum printing speed has been improved by 20-40%
230 (depending on the bignum size.)
231 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
233 * bug fix: the cache used by the CLOS to store precomputed effective
234 methods, slot offsets, and constant return values is now thread and
236 * bug fix: generic function dispatch function updating is now thread
237 and interrupt safe (in the sense that the known issues have been
239 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
240 * bug fix: interning EQL-specializers is now thread and interrupt safe.
241 * bug fix: asdf systems with dependencies to the SB-POSIX or
242 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
243 * bug fix: interrupt safety of applicable method computation has been
246 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
247 * new contrib: sb-cover, an experimental code coverage tool, is included
249 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
250 significantly faster.
251 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
252 produce more efficient code on x86-64 if the compiler can determine
253 that (+ INDEX <constant>) does not require a bounds check and FOO
254 has an element type at least 8 bits wide.
255 * enhancement: a new, experimental synchronous timeout facility is
256 provided. Refer to SB-SYS:WITH-DEADLINE for details.
257 * enhancement: when a symbol name conflict error arises, the
258 conflicting symbols are always printed with a package prefix.
259 (thanks to Kevin Reid)
260 * enhancement: stepping is now once again supported on the SPARC and
261 MIPS platforms. (It is also now more likely to work on CheneyGC
263 * enhancement: sb-sprof can now also track and report accurate call
265 * bug fixes: the treatment of non-standard subclasses of
266 SB-MOP:SPECIALIZER is more correct.
267 * incompatible change: PURIFY no longer copies the data from the
268 dynamic space into the static and read-only spaces on platforms
269 that use the generational garbage collector
270 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
272 * bug fix: GC race condition occasionally resulting in crashes with
273 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
275 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
277 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
278 a STYLE-WARNING to be signalled (regression from 1.0.4.)
279 * bug fix: an asynchronous interrupt could previously leave the
280 system running with GC inhibited.
281 * bug fix: a DECLARE form evaluated at top-level now causes an error
282 rather than silently (or verbosely) returning NIL.
283 * bug fix: trying to dissassemble functions compiled at high DEBUG could
284 result in a error being signalled due to source form lookup errors.
285 (reported by Peter Graves)
287 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
288 * incompatible change: removed writer methods for host-ent-name,
289 host-ent-addresses -- changing the values did not update the DNS
291 * minor incompatible change: changed experimental JOIN-THREAD interface
292 * documentation: the manual now lists reader and writer methods
293 in class slot documentation sections. (thanks to Richard M Kreuter)
294 * documentation: unwinding from asyncronous events has been
295 documented as unsafe.
296 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
297 in multithreaded application code.
298 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
299 platforms. (thanks to James Anderson for the optimization hint)
300 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
302 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
303 variants no longer cons.
304 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
305 their NOT- variants no longer cons.
306 * optimization: EQUAL hash tables no longer use SXHASH for objects
307 of all data types, but instead use an EQL hash for types for which
308 EQUAL is the same as EQL.
309 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
310 are significantly faster.
311 * optimization: new STRING-OUTPUT-STREAM implementation is much
312 faster for large outputs and conses 30% less on average.
313 * enhancement: XREF information is now collected for references made
314 to global variables using SYMBOL-VALUE and a constant argument.
315 * enhancement: SIGINT now causes a specific condition
316 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
317 * bug fix: FILE-POSITION used to signal an error in some cases where
318 ANSI requires it to return NIL.
319 * bug fix: ADJUST-ARRAY is now interrupt-safe.
320 * bug fix: adding and removing fd-handlers is now interrupt-safe.
321 * bug fix: inlined calls to C now ensure 16byte stack alignment on
323 * bug fix: bad type declaration in the CLOS implementation has
324 been fixed. (reported by James Anderson)
325 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
326 (thanks to Lutz Euler)
327 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
328 required by the CLHS. (reported by Samium Gromoff)
329 * bug fix: dead unbound variable references now signal an error.
330 (reported by Marco Monteiro)
331 * bug fix: / with an unused value was being deleted in safe code.
332 (thanks to Marco Monteiro and Kevin Reid)
333 * bug fix: number of characters that can be written onto a single
334 line in a file is unlimited.
335 * bug fix: some GC deadlocks caused by asynchronous interrupts have
336 been fixed by inhibiting interrupts for when GC is disbled.
337 * bug fix: some interrupt safety issues with GETHASH, (SETF
338 GETHASH), CLRHASH and REMHASH have been fixed.
339 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
340 specifier no longer causes infinite recursion.
341 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
342 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
343 * bug fix: an &environment argument with the correct variable information
344 is passed to macros that are expanded during byte compilation.
345 (reported by Samium Gromoff)
346 * bug fix: a logic error was causing the hash function for bignums to
347 have excessive amounts of collisions. (reported by Faré Rideau)
348 * bug fix: modifying the contents of an array could change the return
349 value of SXHASH on that array, which is only allowed for strings
350 and bit vectors. (bug introduced in 0.9.16)
351 * improvement: the x86-64/darwin port now passes all tests (except
352 for the debugger tests) but should still be considered
353 experimental until this is fixed.
354 * improvement: a style-warning is signaled for CASE (etc) clauses with
355 duplicate keys. (thanks to Kevin Reid)
356 * improvement: macos/x86 and macos/x86-64 now use mach exception
357 handlers for dealing with illegal instructions (for trapping and
358 error handling) and memory protection violations (for GC).
360 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
361 * new platform: experimental support for x86-64/darwin (MacOS).
362 * incompatible change: the thread-safe (on most platforms) getaddrinfo
363 and getnameinfo sockets functions are used instead of gethostbyaddr
364 and gethostbyname, on platforms where the newer functions are available.
365 As a result, the ALIASES field of HOST-ENT will always be NIL on these
367 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
368 don't print the startup banner, but behave as if --noinform was passed
369 as a command line argument. (thanks to Kevin Reid)
370 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
371 * optimization: code using alien values with undeclared types is much faster.
372 * optimization: the compiler is now able to open code SEARCH in more cases.
373 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
374 * bug fix: using standardized COMMON-LISP special variables as loop
375 variables no longer signals bogus package lock violations. (reported
377 * bug fix: declaring local loop variables to be of a range-limited type
378 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
379 (reported by Andras Simon)
380 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
381 bugs remain on x86-64.)
382 * bug fix: the #= and ## reader macros now interact reasonably with
383 funcallable instances.
384 * bug fix: type-checks for function arguments were compiled using the
385 compiler policy settings of the wrong lexical environment (in 1.0.2
387 * bug fix: SHADOW accepts characters as string designators, as required
388 by the spec (thanks to Eric Marsden)
389 * bug fix: fixed GC safety issues when foreign functions are called with
390 non-base strings as arguments
391 * bug fix: more consistent error output for fatal-compiler-warnings, like
393 * bug fix: functions with proper names were showing up as NIL in
394 backtraces (reported by Edi Weitz, regression in 1.0.2)
396 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
397 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
398 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
399 produce more efficient code on the x86 if the compiler can determine
400 that (+ INDEX <constant>) does not require a bounds check and FOO
401 has an element type at least 8 bits wide.
402 * bug fix: references to undefined variables are handled the same way
403 in toplevel forms as in the normal compiler
404 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
406 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
407 (thanks to Jon Buller)
408 * improvement: faster compilation times for complex functions
409 * improvement: added readlink support to SB-POSIX (thanks to Richard
412 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
413 * improvement: experimental support for mach exception handling on
414 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
416 * improvement: support for GBK external format.
417 (thanks to Chun Tian (binghe))
418 * improvement: the debugger now displays variables that have been closed
419 over, in code compiled with (DEBUG 2) or higher
420 * improvement: support for executable cores on NetBSD (thanks to
422 * new feature: added a RESTART-FRAME debugger command
423 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
424 be used to provide an implementation for FILE-POSITION on Gray streams
425 (thanks to Eric Marsden)
426 * new feature: add syslog, openlog and closelog support to SB-POSIX
427 (thanks to Richard Kreuter)
428 * optimization: the function call overhead in code compiled with
429 a high DEBUG optimization setting is significantly
430 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
431 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
432 (thanks to Tony Martinez)
433 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
434 evaluated once (reported by Kevin Reid)
435 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
436 works again (thanks to Kilian Sprotte)
437 * bug fix: an error is signaled for tagbodies with duplicate tags
438 (thanks to Stephen Wilson)
439 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
440 * bug fix: Win32 port can now handle foreign code unwinding Lisp
441 stack frames from alien callbacks.
442 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
443 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
444 65k pages (thanks to David Woodhouse)
445 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
447 changes in sbcl-1.0.1 relative to sbcl-1.0:
448 * new platform: FreeBSD/x86-64, including support for threading.
449 * new feature: the compiler stores cross-referencing information
450 abount function calls (who-calls), macroexpansion (who-macroexpands)
451 and special variables (who-binds, who-sets, who-references) for code
452 compiled with (< SPACE 3). This information is available through the
453 sb-introspect contrib.
454 * new feature: users may subclass SEQUENCE, and have instances of
455 these classes interoperate with standard Common Lisp functions if
456 a number of methods are defined. (This feature is experimental
457 and the interface subject to change based on feedback from SBCL
458 users and the general community)
459 * improvement: sb-sprof traces call stacks to an arbitrary depth on
460 x86/x86-64, rather than the previous fixed depth of 8
461 * improvement: another pthread back-end of mutex "pthread-futex".
462 * bug fix: non-ascii command-line arguments are processed correctly
463 (thanks to Yaroslav Kavenchuk)
464 * bug fix: non-required arguments were not passed correctly when a method
465 defined using DEFMETHOD was called from a mop-generated method using
466 CALL-NEXT-METHOD (reported by Pascal Costanza)
467 * bug fix: recursion is now permitted in accessors through
468 SLOT-UNBOUND. (reported by Pascal Costanza)
469 * bug fix: an error was signaled at startup if the HOME environment
470 variable was defined, but had an empty value (reported by Peter Van Eynde)
471 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
472 signal a PROGRAM-ERROR, not a BUG.
473 * bug fix: SB-POSIX:READDIR works when built with large file support.
474 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
476 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
477 for better automatic indentation support. (thanks to Matt Pillsbury)
478 * optimization: loading generic functions no longer takes O(n^2) time,
479 proportional to the amount of methods in the generic function
480 (reported by Todd Sabin and Jeremy Brown)
481 * optimization: the FIND and POSITION family of sequence functions
482 are significantly faster on arrays whose element types have been
484 * improvements to the Windows port:
485 ** Intermittent heap corruption problems have been fixed. (thanks
486 to Alastair Bridgewater)
487 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
489 ** Lisp is able to unwind foreign exception frames from alien
490 callbacks. (thanks to Alastair Bridgewater)
492 changes in sbcl-1.0 relative to sbcl-0.9.18:
493 * improvement: experimental support for threading on FreeBSD/x86.
494 (thanks to NIIMI Satoshi)
495 * improvement: runtime option --dynamic-space-size can be used
496 to set the size of the dynamic space reserved on startup.
497 * improvement: floating point modes in effect are now saved in
498 core, and restored on startup.
499 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
500 startup, not time since first call to GET-INTERNAL-REAL-TIME.
501 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
502 threads are not running after *SAVE-HOOKS* have run.
503 * improvement: writes to CLOS instance slots are type-checked in code
504 compiled with (SAFETY 3)
505 * improvement: floating-point exception handling on FreeBSD (thanks to
507 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
508 (thanks to Zach Beane)
509 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
511 * improvement: added support for the Shift-JIS external format.
512 (contributed by NIIMI Satoshi)
513 * improvement: callbacks are supported on Linux/PPC. (thanks to
515 * bug fix: compiler bug triggered by a (non-standard) VALUES
516 declaration in a LET* was fixed. (reported by Kaersten Poeck)
517 * bug fix: file compiler no longer confuses validated and already
518 dumped structurres. (reported by Kaersten Poeck)
519 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
520 fill-pointer signals a type-error as required. (thanks to
522 * bug fix: disassemly of funcallable instances works.
523 * bug fix: single stepping on PPC.
524 * bug fix: fix thread-safety problems in the type system (generally
525 manifesting as nonsensical errors like "STRING is a bad type specifier
526 for sequences" or "The value 1 is not of type FIXNUM")
527 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
528 (reported by Josip Gracin)
529 * bug fix: an error is signaled for attempts to displace arrays with
530 incompatible element types (thanks to Mario Mommer)
531 * bug fix: more correct handling of wide characters in debug info
532 (bug reported by Attila Lendvai and fixed by Juho Snellman)
533 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
534 and don't cause extra consing
535 * optimization: MAP and MAP-INTO are significantly faster on vectors
536 whose elements types have been declared.
537 * Improvements to SB-SPROF:
538 ** Support for allocation profiling
539 ** Reduced profiling overhead, especially for long profiling runs
540 * Improvements to the Windows port:
541 ** floating point exceptions are now reported correctly.
542 ** stack exhaustion detection works partially.
543 ** more accurate GET-INTERNAL-REAL-TIME.
544 ** better RUN-PROGRAM behaviour in case of errors in the child
546 ** PROBE-FILE now simplifies pathnames correctly.
547 ** DIRECTORY now works correctly with :WILD-INFERIORS.
548 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
549 1st 1970 (although time timezone data will be unreliable).
550 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
552 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
553 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
554 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
556 * bug fix: two potential GC deadlocks affecting threaded builds.
557 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
558 returns T (reported by Anton Kazennikov)
559 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
560 with non-variable places
561 * bug fix: remove a race condition in the setting of
562 funcallable-instance functions, this should make threaded CLOS
563 code more stable against memory faults.
564 * bug fix: corruption of specials when unbinding is interrupted by an
565 asynchronous unwind (reported by Hannu Koivisto)
566 * improvement: the debugger will now also display local variables that
567 are only used once, for code compiled with a DEBUG optimization quality
570 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
571 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
572 * incompatible change: External-format support for FFI calls. The
573 SB-ALIEN:C-STRING no longer implies an ASCII
574 external-format. Instead the string is subject to external-format
575 conversion. Additionally return values of type SB-ALIEN:C-STRING
576 are by default of type (SIMPLE-ARRAY CHARACTER), not
577 SIMPLE-BASE-STRING. For an alien type that behaves like the old
578 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
579 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
580 * incompatible change: SB-EXT package no longer contains the
581 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
582 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
584 * incompatible change: the single-stepper is no longer available
585 on Alpha, Mips and Sparc platforms.
586 * minor incompatible change: the direct superclasses of
587 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
588 not (STANDARD-OBJECT FUNCTION). This makes the
589 class-precedence-lists of GENERIC-FUNCTION and
590 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
592 * minor incompatible change: the default stream external format on
593 non-unicode SBCL is no longer detected from the locale, but is always
595 * new feature: Add a version of evaluator that uses an interpreter instead
596 of the compiler. EVAL still uses the compiler by default, to switch it
597 to use the interpreter, set the value of the variable
598 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
599 * minor incompatible change: the single-stepper REPL has been merged
600 with the normal debugger (see the "Stepping" heading of the debugger help
601 for more details). The debugger command STEP will no longer switch
602 to the single-stepper REPL.
603 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
604 for a type now works.
605 * bug fix: loading of default sysinit file works. (thanks to Leonid
607 * bug fix: better detection of circularities in the file-compiler.
608 (reported by Marco Monteiro)
609 * bug fix: the CL pathname functions now work with files that have
610 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
611 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
612 systems with Unix98 pty semantics.
613 * bug fix: ASDF-INSTALL will now work with bsd tar.
614 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
616 * bug fix: timers expiring in dead threads no longer cause a
617 type-error (reported by Paul "Nonny Mouse").
618 * bug fix: thanks to more lightweight single-stepper instrumentation,
619 code compiled with (DEBUG 3) will compile and execute significantly faster,
620 and will have more accurate type-inferencing than before
621 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
622 whose bindings are modified
623 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
624 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
625 ** replace dummy implementations of CL:MACHINE-INSTANCE and
626 CL:SOFTWARE-VERSION with proper version
628 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
629 * feature: implemented the READER-METHOD-CLASS and
630 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
631 as specified by AMOP.
632 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
634 * optimization: faster LOGCOUNT implementation on x86 and x86-64
635 (thanks to Lutz Euler)
636 * optimization: hashing of general arrays and vectors has been
637 improved. (reported by Any Fingerhut)
638 * enhancement: SB-INTROSPECT is now able to find definitions of
639 profiled functions. (thanks to Troels Henriksen)
640 * enhancement: compiler-macro expansion applies now to FUNCALL forms
641 as well, allowing compiler-macros for SETF-functions to expand.
642 * enhancement: step-instrumentation no longer wraps known
643 single-value functions with multiple-value context, allowing
644 better type inference.
645 * fixed bug #337: use of MAKE-METHOD in method combination now works
646 even in the presence of user-defined method classes. (reported by
647 Bruno Haible and Pascal Costanza)
648 * fixed bug #339(c): if there are applicable methods not part of any
649 long-form method-combination group, call INVALID-METHOD-ERROR.
650 (reported by Bruno Haible)
651 * fixed bug #361: the :FUNCTION initarg in the protocol for
652 initialization of methods can now be used to override
653 internally-produced optimized functions. (reported by Bruno
655 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
656 system-provided lambda expression no longer cause warnings about
657 unbound #:|pv-table| symbols.
658 * bug fix: improved the handling of type declarations and the
659 detection of violations for keyword arguments with non-constant
661 * bug fix: potentially erronous calls to PATHNAME and
662 MERGE-PATHNAMES were being flushed in some cases.
663 (reported by Richard Kreuter)
664 * bug fix: compiled calls to TYPEP were mishandling obsolete
665 instances. (reported by James Bielman and Attila Lendvai)
666 * bug fix: format strings with ~> without matching ~< no longer
667 trigger an AVER, but signal an understandable error instead.
668 (reported by Antonio Martinez)
669 * bug fix: specifying an output-file without a pathname-type for
670 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
671 of input-file instead of "fasl". (reported by Robert Dodier)
672 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
673 (reported by James Y Knight).
674 * bug fix: compiler-macros-function did not consider the environment
675 argument for shadowing by local functions.
676 * bug fix: compiler-macros expansion was inhibited by local INLINE
678 * bug fix: inline expansions of known functions were subject to
679 step-instrumentation in high debug policies, leading to problems
681 * bug fix: compiler failed to differentiate between different CONS
683 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
684 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
685 * bug fix: #368: incorrect use of expressed vs. upgraded array
687 * bug fix: #306a: more precise unions of array types.
688 * thread-safety improvements:
689 ** CONDITION-WAIT could return early on Linux, if the thread was
690 interrupted and subsequently continued with SIGCONT.
691 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
693 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
694 * added support for the ucs-2 external format. (contributed by Ivan
696 * minor incompatible change: pretty printing of objects of type
697 (cons symbol) is, in the default pprint-dispatch-table, now
698 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
700 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
701 called later in a class's lifetime, possibly as late as when the
702 first instance of the class is created. Previously,
703 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
704 class became finalizeable.
705 * fixed bug: FILE-POSITION sometimes returned inconsistent results
706 for multibyte external-format streams. (thanks to "vbzoli")
707 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
708 with :ALLOCATION :CLASS inherited from superclasses of the
710 * fixed bug: anonymous classes can now be created using the :NAME
711 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
712 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
713 * fixed bug: core-files saved with :EXECUTABLE T can again be
714 executed when SBCL_HOME isn't set. (reported by James Knight)
715 * fixed bug: toplevel LOCALLY forms with declarations could
716 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
717 * fixed bug: printing from several different threads using different
718 values of *print-case* could cause invalid output, due to
719 some internal special variables of the printer not being bound
720 thread-locally (reported by Max Mikhanosha)
721 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
722 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
723 (reported by Pascal Costanza)
724 * fixed bug: FINALIZE-INHERITANCE is called from
725 REINITIALIZE-INSTANCE on classes when the class has previously
726 been finalized, as required by AMOP.
727 * minor code generation optimizations:
728 ** better register allocation in CLOS dispatching functions
729 ** overflow detection when coercing signed bytes to fixnums on x86-64
730 ** is now implemented with one IMUL instruction instead of three shifts
731 ** more efficient bit-vector access on x86 and x86-64
732 ** more efficient access to raw structure slots on x86 and x86-64
733 * fixed some bugs revealed by Paul Dietz' test suite:
734 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
737 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
738 * feature: thread support on Solaris/x86, and experimental thread support
740 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
742 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
743 default method for SLOT-UNBOUND.
744 * minor incompatible change: prevent the user from specializing the
745 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
746 somewhat counter to the intent of the protocol, I (CSR) think, and
747 additionally it just doesn't work in SBCL as currently
748 implemented, thanks to optimizations (that are always valid for
749 the other three SLOT-VALUEish functions, but not for the setter).
750 * minor incompatible change: the :SB-LDB feature is now enabled by
751 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
752 the low-level debugger.
753 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
754 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
755 down to an SB-SYS:FD-STREAM, allowing direct communication in
756 more cases. Temporary files and pipes are still used as a fallback
758 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
759 on the non-reentrant C functions localtime(3) and gmtime(3).
760 * bug fix: global optimization policy was not visible in LOCALLY and
762 * bug fix: class objects can be used as specializers in methods.
763 (reported by Pascal Costanza)
764 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
765 with a type error. (reported by blitz_ on #lisp)
766 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
767 failed with a type error. (reported by Pascal Bourguignon)
768 * bug fix: merging pathnames against defaults with :DIRECTORY
769 starting with '(:RELATIVE :BACK) should preserve the :BACK.
770 (reported by James Y Knight)
771 * bug fix: saving large (>2GB) cores on x86-64 now works
772 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
773 constant index and a (simple-array (signed-byte 32)) array
774 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
775 error (patch by Robert J. Macomber)
776 * feature: TIME now displays the amount of run-time spent in GC
777 * bug fix: The debugger now does a better job of respecting
778 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
779 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
780 workaround for bug 403.)
781 * bug fix: floating point exception handling now works on FreeBSD
782 (thanks to NIIMI Satoshi)
783 * fixed some bugs revealed by Paul Dietz' test suite:
784 ** MISC.641: LET-conversion were not supposed to work in late
786 * improvements to the Win32/x86 port:
787 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
788 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
790 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
791 * new feature: source path information is generated for macro-expansion
792 errors for use in IDE's like Slime (thanks to Helmut Eller)
793 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
794 * bug fix: compiler does not loop forever on an invalid type in
796 * improvement: compilation of most CLOS applications is significantly
798 * optimization: added a limited bytecode compiler for simple toplevel
799 forms, speeding up compilation and FASL loading
800 * bug fix: the statistical profiler now properly distinguishes anonymous
803 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
804 * minor incompatible change: in sbcl-0.9.11 (but not earlier
805 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
806 into subclasses of the finalized class. Now user calls to
807 FINALIZE-INHERITANCE finalize just the one class, and calls by the
808 system return before any subclasses are finalized.
809 * minor incompatible change: The reader no longer ignores errors
810 regarding non-existent packages in #+ and #- feature tests.
811 * new feature: command line options --no-sysinit, --no-userinit to
812 inhibit loading the corresponding init files
813 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
814 for sending data through UDP sockets (thanks to François-René Rideau)
815 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
816 error is signalled instead (thanks to François-René Rideau)
817 * minor incompatible change: Error signalling behaviour of lexical
818 operations violating package locks has changed slightly. Refer to
819 documentation on package locks for details.
820 * enhancement: EVAL can process IF-expressions without resorting to the
822 * bug fix: Some operations involving SIN, COS, or TAN weren't being
823 constant-folded properly due to missing out-of-line functions.
824 (reported by Mika Pihlajamäki)
825 * bug fix: LISTEN sometimes returned T even in cases where no data was
826 immediately available from the stream
827 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
828 were reversed. (reported by Levente Mészáros)
829 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
830 when given a vector displaced to an adjustable array. (reported
832 * bug fix: derivation of float boundaries from numbers outside the
833 appropriate float range (reported by John Wiseman)
834 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
836 * bug fix: precomputing cache entries for generic functions with
837 some subclasses of specializers as yet invalid does not attempt to
838 fill a cache line with a negative offset. (reported by Levente
840 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
841 allow condition class objects as arguments to DOCUMENTATION and
842 (SETF DOCUMENTATION); only find and set documentation for
843 structure names for the STRUCTURE doc-type. (suggested by Gary
845 * improvements to the Win32/x86 port:
846 ** (user-homedir-pathname) and default initialization file
847 locations now know about the user's "Documents and Settings"
848 directory (thanks to Yaroslav Kavenchuk)
849 ** run-program is implemented (thanks to Mike Thomas)
850 ** sockets support (thanks to Timothy Ritchey)
851 ** better backtrace support (thanks to Alastair Bridgewater)
852 ** sb-grovel supported
853 ** asdf-install and sb-posix work somewhat
854 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
855 * improvements to the Solaris/x86 port:
856 ** works on Solaris 11/Solaris Express
857 ** floating-point exception handling support
858 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
859 * fixed some bugs revealed by Paul Dietz' test suite:
860 ** REMOVE-METHOD returns its generic function argument even when
861 the method is not one of the generic functions' methods.
862 ** objects declared as MEMBER types can be admissible arguments to
864 ** printing characters should simply be printed by the FORMAT ~:C
866 ** compiler failure when compiling functions with hairy constant
867 defaults for optional parameters.
868 ** compiler produces wrong code when MAYBE-INLINE-expanding a
869 function, which is already optimized.
871 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
872 * new platform: experimental support for SBCL x86/Darwin, including
873 MacOS X 10.4.5 on Intel.
874 * new feature: Unicode character names are now known to the system
875 (through CHAR-NAME and NAME-CHAR).
876 * new feature: the filesystem location of SBCL's core file is
877 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
878 * minor incompatible change: the contrib modules SB-POSIX and
879 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
880 this change is to make it easier to distribute
881 location-independent binaries.
882 * bug fix: as implied by AMOP, standardized classes no longer have
883 slots named by external symbols of public packages. (reported by
885 * optimization: calling functions via constant symbols -- as in
886 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
887 via the function object as in (FUNCALL #'FOO).
888 * optimization: on x86 and x86-64, the calling convention for the
889 case of dealing with an unknown number of values has been altered
890 to be friendlier to the prediction heuristics implemented,
891 particularly with reference to CALL and RET pairing. (thanks to
892 Alastair Bridgewater)
893 * optimization: on x86, the code for access to thread-local symbol
894 values and binding/unbinding of thread-local symbols is smaller.
895 (thanks to Alastair Bridgewater)
896 * enhancement: CONSTANTP is now able to determine constantness of
897 more complex forms, including calls to constant-foldable standardized
898 functions and some special forms beyond QUOTE.
899 * fixed bug: occasional GC crashes on Solaris/x86
900 * optimization: x86-64 supports stack allocation of results of simple
901 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
902 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
903 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
904 (thanks to James Knight)
905 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
906 by the standard (thanks to Stephen Compall)
908 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
909 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
910 be used for bundling the runtime and the core file into one
911 executable binary. This feature is not currently supported on all SBCL
912 platforms. (thanks to James Bielman and NIIMI Satoshi)
913 * new feature: a generational or ephemeral garbage collector is now
914 the default on the PowerPC platform (both Linux and Darwin). The
915 old Cheney (stop and copy) collector is a build-time option.
916 (thanks to Cyrus Harmon, after Raymond Toy)
917 * minor incompatible change: the method by which SBCL finds its
918 contributed modules has changed; it no longer relies on symbolic
919 links from an $SBCL_HOME/systems directory, but searches directly
920 in the subdirectories of $SBCL_HOME.
921 * enhancement: the dynamic heap size on the Linux/PPC platform is
922 markedly larger, even using the older Cheney garbage collector.
923 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
924 certain complicated string types. (reported by Gary King)
925 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
926 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
927 * fixed bug: DOCUMENTATION on structures no longer signals an
928 error if no structure type of the right name exists. (reported by
930 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
931 applicable non-standard primary or :AROUND method on
932 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
933 no longer cause unbound variable errors. (reported by Kevin Reid)
934 * fixed bug: metacircle resolution in cases where methods have slots
935 added before the slots from STANDARD-METHOD. (reported by Jean
937 * fixed bug: the Power PC assembly code for calling into lisp
938 sometimes computed its return address wrongly (depending on the
939 15th bit of the address, fixed at link time). This used to
940 manifest itself in a segmentation violation while building PCL.
941 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
942 many others over the years)
943 * fixed bug: ROOM no longer reports silly numbers for stack usage if
944 the stack pointer (treated as a signed integer) is negative.
945 (thanks to Peter van Eynde)
947 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
948 * new platform: experimental support for the Windows operating
949 system has been added. (thanks to Alastair Bridgewater)
950 * fixed several bugs in and robustified the PPC FFI (including
951 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
952 * bug fix: allow non-simple string symbol names (reported by Paul
954 * bug fix: interrupt handling on NetBSD (thanks to Richard M
956 * bug fix: saving a core corrupted callbacks on x86/x86-64
957 * bug fix: closed a loophole in metacircularity detection and
958 grounding in the PCL implementation of CLOS.
959 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
960 (reported by tomppa on #lisp)
961 * bug fix: clear the direction flag on Lisp -> C transitions, as
962 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
963 (reported by Andrej Grozin and Hendrik Maryns)
964 * optimization: major improvements to GC efficiency on GENCGC platforms
965 * optimization: faster implementation of EQUAL
966 * optimization: emit more efficient opcodes for some common
967 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
968 * optimization: several other minor code-generation improvements on x86-64
969 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
971 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
972 * minor incompatible change: (SETF CLASS-NAME) and (SETF
973 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
974 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
975 by ANSI as a generic function, and for consistency (SETF
976 GENERIC-FUNCTION-NAME) is treated likewise.
977 * fixed bug #233.b: make constraint propagation notice when a variable
978 value is changed after it is referenced but before it is used
979 * fixed bug #296: no more arbitrary behaviour from filenames with
980 odd characters as --load arguments. (reported by Adam Warner)
981 * fixed bug #390: :CHARACTER-SET pathname components now work as
982 expected. (reported by Tim Daly Jr)
983 * fixed bug #391: complicated :TYPE intersections in slot
984 definitions no longer cause an error in PCL internals.
985 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
986 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
987 returning the number of octets which would be written to the
988 file-stream. (thanks to Robert J. Macomber)
989 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
990 arguments to the internal stream functions. (thanks to David
992 * bug fix: stack exhaustion is now handled on the x86/SunOS
993 platform. (thanks to Daisuke Homma)
994 * bug fix: more accurate ROOM results on GENCGC platforms
995 * optimization: improved type inference for arithmetic-for
996 index variables in LOOP
997 * optimization: faster floating-point SQRT on x86-64
998 * fixed some bugs revealed by Paul Dietz' test suite:
999 ** DOCUMENTATION returns NIL instead of "" for method combinations
1000 that don't have a docstring
1002 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1003 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1004 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1005 therefore are not customizeable by user code (as seems to be at
1006 least permitted and maybe required by AMOP). As a consolation,
1007 however, the SBCL implementation of these functions now calls
1008 REINITIALIZE-INSTANCE as specified by AMOP.
1009 * bug fix: slot-definition documentation is propagated as per ANSI
1010 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1011 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1012 argument, as do eslotd initialization methods. (from Pascal
1013 Costanza's "Closer" project)
1014 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1015 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1017 * bug fix: it is now possible to have more than one subclass of
1018 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1019 (reported by Bruno Haible, Pascal Costanza and others)
1020 * bug fix: the dependent update protocol now works for generic
1021 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1022 and Pascal Costanza)
1023 * bug fix: condition-class instances corresponding to
1024 DEFINE-CONDITION forms are now created eagerly. (reported by
1025 Kalle Olavi Niemitalo on comp.lang.lisp)
1026 * bug fix: floating point printing is more accurate in some
1027 circumstances. (thanks to Simon Alexander)
1028 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1029 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1030 * bug fix: callbacks on OS X now preserve stack-alignment.
1031 (thanks to Cyrus Harmon)
1032 * enhancement: the x86-64 disassembler is much better at
1033 disassembling SSE instructions. (thanks to Lutz Euler)
1034 * enhancement: CHECK-TYPE on a variable now allows type-inference
1035 (thanks to Vincent Arkesteijn)
1036 * optimization: improved performance of EUC-JP external format.
1037 (thanks to NIIMI Satoshi)
1038 * optimization: performance improvements to IO on file streams of
1039 :ELEMENT-TYPE CHARACTER
1040 * optimization: much faster memory allocation on x86-64
1041 * optimization: faster garbage collections (latency and throughput)
1043 * optimization: faster bignum subtraction and division on x86 and x86-64
1044 (thanks to Lutz Euler)
1045 * optimization: more accurate type-inference for loops that use a
1046 floating point index variable or a negative step.
1048 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1049 * bug fix: add a workaround to SBCL looping infinitely at startup on
1050 Linux kernels with apparently buggy implementations of personality().
1051 (thanks to Svein Ove Aas)
1052 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1053 (thanks to David Lichteblau)
1054 * bug fix: local bindings shadow global symbol macros for
1055 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1056 * optimization: non-open coded uses of numeric comparison operators
1057 (e.g. >) no longer cons when called with more than one parameter
1058 on platforms supporting dynamic-extent allocation.
1059 * enhancement: saving cores with foreign code loaded is now supported
1060 on MIPS/Linux in addition to the previously supported platforms.
1061 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1062 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1063 * bug fix: division by zero in sb-sprof when no samples were collected
1064 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1065 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1066 (thanks to Alexander Kjeldaas)
1067 * new feature: ignoring the return values of destructive functions
1068 like NREVERSE signals a compile-time style-warning.
1069 (thanks to Kevin Reid)
1070 * new feature: the alignment of alien structure fields can be
1071 explicitly specified. (thanks to Cyrus Harmon)
1072 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1073 and dump core on SIGQUIT
1075 ** incompatible change: threads do not inherit values of specials
1076 from their parents (see manual)
1077 ** bug fix: threads stacks belonging to dead threads are freed by the
1078 next exiting thread, no need to gc to collect thread stacks anymore
1079 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1080 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1081 ** bug fix: dynamic variable and thread start related gc lossage
1082 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1084 * fixed some bugs revealed by Paul Dietz' test suite:
1085 ** SUBTYPEP is slightly more accurate on heinously complicated
1086 CONS types where some of the members have uncertain (in the
1087 NIL, NIL sense) type relationships to each other.
1089 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1090 no longer purified unless :PURIFY T is explicitly specified.
1091 ** Non-purified cores are significantly smaller than before
1093 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1094 * new feature: timers based on Zach Beane's excellent timer package
1095 * added support for the following external formats: koi8-u,
1096 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1097 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1098 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1099 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1100 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1101 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1102 * incompatible change: a threaded SBCL will no longer revert to
1103 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1104 * bug fix: interrupts are disabled until startup is complete; no
1105 more sigsegvs when receiving a signal to soon
1106 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1108 * bug fix: add a workaround for the memory randomization features in
1109 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1110 workaround will only be in effect on systems with the proc filesystem
1112 * bug fix: printing objects of type HASH-TABLE signals a
1113 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1115 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1116 have been running for over 50 days. (reported by Gilbert Baumann)
1117 * bug fix: the logic for getting names of functions gets less
1118 confused when confronded with alternate-metaclass
1119 funcallable-instances. (reported by Cyrus Harmon)
1120 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1121 consistent, even on internal alternate-metaclass objects.
1122 * bug fix: flush closure information collected by physical
1123 environment analysis prepass before the main pass. (bug reported
1125 * bug fix: compiler pack phase does not modify a hashtable, which is
1126 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1127 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1128 may be expanded. (reported by Alexander <malishev> on c.l.l)
1129 * bug fix: no more style-warnings when compiling code using LAST
1130 under high-SPEED low-SPACE settings. (reported by David Wallin
1132 * fixed some incompatibilities between SBCL's MOP and the MOP
1134 ** the METAOBJECT class is now implemented;
1135 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1136 STANDARD-OBJECT, as required;
1137 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1138 now compatible (as required by VALIDATE-SUPERCLASS); there
1139 remains a consistency requirement over the presence or absence
1140 of the FUNCTION class in the superclass list of finalized
1141 classes; see the manual for more details;
1142 ** the :AROUND method for COMPUTE-SLOTS on
1143 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1144 requested slot ordering.
1146 ** bug fix: parent thread now can be gc'ed even with a live
1148 ** bug fix: binding a special with PROGV to no value is not
1149 the same as that symbol not having been bound (thanks to
1151 ** bug fix: binding specials is thread safe (thanks to
1153 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1154 and SB-SPROF do not die with 'no handler for signal XX in
1155 interrupt_handle_now(..)' anymore
1156 ** bug fix: WITH-TIMEOUT works with multiple running threads
1157 * fixed some bugs revealed by Paul Dietz' test suite:
1158 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1159 the :method-class keyword argument.
1161 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1162 * new port: the Solaris operating system on x86 processors is now
1163 mostly supported, though some rough edges in the environment
1164 remain. (thanks to Daisuke Homma)
1165 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1166 space for its heap. (thanks to Thiemo Seufer)
1167 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1168 closures. (thanks to Thiemo Seufer)
1169 * minor incompatible change: eof selects abort in the debugger.
1170 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1171 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1173 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1174 deprecated, and will go away in a future revision of SBCL.
1175 * minor incompatible change: GC-ON and GC-OFF are no longer
1176 implemented with a counter, it does not matter how many times gc
1177 is switched on or off
1178 * bug fix: discriminating functions for generic function classes
1179 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1180 make invalid assumptions about method precedence order. (reported
1182 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1183 platforms. (thanks to Thiemo Seufer)
1184 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1185 * bug fix: correct pathname printing with printer escaping is on.
1186 (thanks to Kevin Reid)
1187 * bug fix: complex VOP definitions in "user-space" no longer trigger
1188 package locks. (reported by Zach Beane)
1189 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1190 for funcallable-instance functions, and is no different from
1192 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1193 characters in the current readtable. (reported by Nicholas Neuss)
1194 * bug fix: SUBTYPEP on various CONS types returns more a more
1195 accurate acknowledgment of its certainty.
1196 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1197 lists in some cases. This partially fixes bug 384.
1198 * flush all standard streams before prompting in the REPL and the
1200 * bug fix: signal handling and triggering gc do not conflict
1201 directly anymore, in particular a high frequency sb-sprof does
1202 not prevent gc from running
1203 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1204 approximation for timezone and DST information between the
1205 universal time epoch and the smallest negative 32-bit time_t.
1206 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1207 year 1899 so long as the encoded time is non-negative.
1208 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1209 the correct answer for non-constant rotate amounts when performing
1210 an inline 32-bit rotation.
1212 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1213 there is only one thread in the session
1214 ** bug fix: memory leak for streams created in one thread and
1215 written to in another
1216 ** bug fix: lockup when compiled with gcc4
1217 ** bug fix: race that allows the gc to be triggered when gc is
1219 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1220 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1222 ** bug fix: finished threads are gc'ed properly
1223 * fixed some bugs revealed by Paul Dietz' test suite:
1224 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1225 has arguments with a different set of applicable methods from
1226 the orignal arguments.
1227 ** The type error thrown by MAP now has a correct expected-type
1229 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1230 name a compiled function.
1231 ** Three MISC tests where a large bignum was improperly coerced to
1232 a float that couldn't represent that bignum during type
1233 derivation were fixed.
1234 ** SUBTYPEP can now handle types involving the negation of a
1235 list-form FUNCTION type.
1236 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1237 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1238 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1240 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1241 * New feature: Experimental support for bivalent streams: streams
1242 opened with :element-type :default now allow character and binary
1243 (unsigned-byte 8) I/O
1244 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1245 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1247 * bug fix: no more highly sporadic "couldn't check whether ~S is
1248 readable" when reading a stream and an interrupt hits in the middle
1249 of a select system call
1250 * compiler better recognizes complex arrays (reported by Eduardo
1252 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1253 for variables on SBCL built with linkage-tables. (reported by Luis
1255 * various error reporting improvements.
1256 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1257 (Thanks to Thiemo Seufer)
1258 * bug fix: nested reader invokations work correctly
1259 * bug fix: it is possible to have simultaneous references to foreign
1260 code and foreign data with the same name.
1262 ** added x86-64 support
1263 ** incompatible change: the threading api now works with thread
1264 objects instead of thread ids
1265 ** bug fix: threads are protected from signals and interruption when
1266 starting up or going down
1267 ** bug fix: a race where an exiting thread could lose its stack to gc
1268 ** bug fix: don't halt on infinite error in threads if possible
1269 ** fixed numerous gc deadlocks introduced in the pthread merge
1270 ** bug fix: fixed thread safety issues in read and print
1271 ** bug fix: debugger doesn't hang on session lock if interrupted at
1272 an inappropriate moment
1273 ** bug fix: run-program is now thread safe(r)
1274 ** bug fix: inner with-recursive-lock no longer releases the mutex
1275 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1276 * fixed some bugs revealed by Paul Dietz' test suite:
1277 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1278 EXPECTED-TYPE slots filled.
1279 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1281 ** MAKE-CONDITION accepts classes as type-designators.
1282 ** COMPILE may never return NIL.
1283 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1284 range before calling Unix time functions
1286 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1287 * numerous signal handling fixes to increase stability
1288 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1289 * minor incompatible change: we now correctly canonize default
1290 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1291 as per the MOP, rather than the historical (INITARG INITFUNCTION
1292 INITFORM). (reported by Bruno Haible)
1293 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1294 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1295 for more information.
1296 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1297 pathname is a directory pathname.
1298 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1299 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1301 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1302 :METACLASS argument in addition to a class name. (reported by
1303 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1304 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1305 (reported by Bruno Haible; patch from Gerd Moellmann)
1306 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1308 * bug fix: more cleanups to the floating point exception handling on
1309 x86-64 (thanks to James Knight)
1310 * bug fix: the compiler does not try to constant fold calls of
1311 COERCE and BIT-* functions when they return freshly constructed
1312 objects. (reported by Nikodemus Siivola and Paul Dietz)
1313 * optimization: improved the MIPS versions of generic (in the
1314 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1315 * optimization: direct conversion of (unsigned-byte 32) to floats on
1316 the PowerPC platform.
1317 * optimization: structure instances with raw slots now use less
1318 memory, and probably show better memory locality. (thanks to
1320 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1321 are treated as requests for stack allocation on the x86-64,
1322 Alpha-32, PPC and SPARC platforms.
1323 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1324 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1325 (thanks to Tony Martinez)
1327 ** gcing a dead thread can no longer lead to lockups
1328 ** threads block signals until they are set up properly
1329 ** errno is no longer shared by threads
1330 ** interrupt-thread restores the eflags register on x86
1331 ** fixed some lockups due to gc/thread interaction
1332 * fixed some bugs revealed by Paul Dietz' test suite:
1333 ** invalid dotted lists no longer raise a read error when
1334 *READ-SUPPRESS* is T
1335 ** signal an error if a symbol that names a declaration is used
1336 as the name of a type, or vice versa
1337 ** allow using the (declare (typespec var*)) abbreviation for
1338 (declare (type typespec var*)) with all type specifiers
1339 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1340 ** always return NIL from PPRINT-POP when OBJECT is NIL
1341 ** don't signal errors when pretty-printing malformed LABELS,
1342 FLET or MACROLET forms
1343 ** declarations in a DOLIST body are also in scope for the
1345 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1347 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1350 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1351 * fixed cross-compiler leakages that prevented building a 32-bit
1352 target with a 64-bit host compiler.
1353 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1354 opened with :IF-EXISTS :OVERWRITE.
1355 * fixed bug 281: error for an invalid qualifier in a short-form method
1356 combination method is not signalled until the faulty method is called.
1357 * bug fix: iteration variable type inferrer failed to deal with open
1358 intervals. (reported by Alan Shields)
1359 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1360 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1361 or more runtime options were provided to the sbcl binary.
1362 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1363 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1364 * optimization: there's now a fast-path for fixnum arguments in the
1365 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1366 * optimization: the code generated on x86-64 is more compact thanks
1367 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1368 * bug fix: floating-point exception handling now partly works on
1369 x86-64. (Thanks to James Knight)
1370 * improvement to the MIPS backend from Thiemo Seufer: C-style
1371 64-bit long long arguments and return values to alien functions
1373 * fixed some bugs revealed by Paul Dietz' test suite:
1374 ** the type-error signalled from WARN has a filled-in DATUM slot.
1375 ** the type-error required when a stream is not associated with
1376 a file has the stream as its datum.
1377 ** type-errors on single-floats on x86-64 no longer have
1378 :INVALID-OBJECT as the datum
1379 ** the type-errors signalled for invalid function names now have
1380 a correct expected type
1381 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1382 for structures defined with a :TYPE. Documentation strings for
1383 typed structures are no longer immediately discarded
1384 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1385 broadcast streams and synonym streams. FILE-LENGTH now also works
1386 on broadcast streams.
1388 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1389 * incompatible change: the --noprogrammer option, deprecated since
1390 version 0.7.5, has been removed. Please use the equivalent
1391 --disable-debugger option instead.
1392 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1394 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1395 inoperational for a while now) has been completely removed.
1396 * null lexical environments are now printed as #<NULL-LEXENV>,
1397 significantly reducing the amount of clutter in typical backtraces.
1398 * documentation on weak pointers, finalization, and after GC hooks
1399 has been added to the manual.
1400 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1401 as other specialized array types, is much faster. SUBSEQ and
1402 COPY-SEQ on such arrays have also been sped up.
1403 * optimization: EQL is now more efficient when at least other argument
1404 is known to be of type (OR FIXNUM (NOT NUMBER)).
1405 * fixed bug: the runtime is now less vulnerable to changes in the
1406 size of the SBCL object on OS X, and virtual memory is reserved for
1407 all spaces that need to be at a fixed address.
1408 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1409 * fixed bug: finalizers and after GC hooks that cause consing are now
1411 * fixed bug: compiler error messages and summaries are now printed to
1412 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1413 * fixed inference of the upper bound of an iteration variable.
1414 (reported by Rajat Datta).
1415 * fixed bug 211e: calling local functions with duplicated constant
1416 keyword argument no longer causes a bogus style warning about an
1418 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1419 ftype declaration to be disregarded. (reported by Dave Roberts)
1420 * fixed bug 373: caused by erronous compilation of references to alien
1421 variables in the runtime on ppc/darwin.
1422 * fixed bug 376: CONJUGATE type deriver.
1423 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1424 (reported by Baughn on #lisp)
1425 * a cross-compiler bug on non-x86 platforms has been identified and
1426 fixed. (thanks to Bruno Haible)
1427 * improvements to the MIPS runtime code for increased stability.
1428 (thanks to Thiemo Seufer)
1429 * increased the maximimum compact environment size to allow
1430 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1431 * partial workaround for bug 135: don't name the function we're
1432 calling for hairy cases of EVAL, so as not to accumulate one environment
1433 entry per call to EVAL. (reported by Kevin Reid)
1434 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1435 * optimization: type testing for non-vector arrays should be faster.
1436 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1438 * bug fix: setting 31st element of a bit vector to zero did not work
1440 * bug fix: redefining a class definition which failed due to a
1441 previous accessor / function clash now works (but see BUGS entry
1442 #380 for more problems in this area). (thanks to Zach Beane)
1443 * the long form of DEFINE-METHOD-COMBINATION disables method group
1444 checking when given a single method group with pattern *. (thanks
1446 * on x86 compiler supports stack allocation of results of simple
1447 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1448 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1449 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1450 * fixed some bugs related to Unicode integration:
1451 ** the restarts for recovering from input and output encoding
1452 errors only appear when there is in fact such an error to
1454 * fixed some bugs revealed by Paul Dietz' test suite:
1455 ** MISC.549 and similar: late transformation of unsafe type
1456 assertions into derived types caused unexpected code
1458 ** SCALE-FLOAT type deriver is less wrong.
1459 ** type derivers for EXP, LOG and similar functions compute result
1460 types for complex arguments better.
1461 ** (MISC.563) CONJUGATE type deriver works for very restricted
1463 ** out-of-line type testers for character strings are available.
1464 ** EQUAL compiler transform understands specialness of objects
1466 ** accessing double-floats stored on the stack now works on x86-64.
1467 ** debugger internals could sometimes create invalid lispobjs,
1468 resulting in GC crashes.
1469 ** MISC.548: type check weakening can convert required type into
1471 ** initialization forms for bindings are not in scope of free special
1474 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1475 * incompatible change: thread support for non-NPTL systems has
1476 been removed - locking is buggy and unreliable. A threaded
1477 SBCL build will now warn at startup and refuse to create
1478 new threads, unless futex support is detected in the kernel
1479 * incompatible change: the top level REPL now has only an ABORT
1480 restart associated with it, not TOPLEVEL and ABORT as it used to.
1481 TOP and TOPLEVEL are now available as debugger commands for
1482 returning to the top level.
1483 * incompatible change: forms evaluated in the REPL now use the
1484 global optimization policy.
1485 * incompatible change: user- and system-initialization files are
1486 no longer processed with LOAD, but by READ and EVAL; hence the
1487 global optimization policy, startup package, readtable, etc,
1489 * The .fasl file format number has been incremented because of
1490 various incompatible changes.
1491 * internal entry point details and argument counts no longer appear
1492 in backtraces unless explicitly requested by setting
1493 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1494 * built-in and standard functions no longer have names like "top
1495 level local call to FOO".
1496 * fixed bug 32: functions defined in non-null lexical environments
1497 now have more legible printed representation
1498 * fixed bug 33: functions defined in non-null lexical environemnts
1499 are now more amenable to inspection by INSPECT.
1500 * workaround for bug 354: XEPs no longer appear in backtraces unless
1501 explicitly requested.
1502 * fixed bug: receiving the signal which results from attempting to
1503 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1504 SIGBUS on other *BSDs) on architectures where the C stack is also
1505 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1506 to debugger confusion or wild instability. (reported by Cheuksan
1508 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1509 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1510 (reported by Lutz Euler)
1511 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1512 compiled code. (reported by Paul Dietz)
1513 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1514 specialized parameter to a method to something that is not TYPEP
1515 the specializer is now possible.
1516 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1517 face of package deletion.
1518 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1519 pathname. (reported by David Tolpin on c.l.l)
1520 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1521 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1522 STRING-STREAM and FILE-STREAM, respectively.
1523 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1524 than previously; in addition, its test suite is now run on build.
1525 (thanks to Luke Gorrie)
1526 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1527 * optimization: added a immediate representation for single-floats
1529 * fixed some bugs revealed by Paul Dietz' test suite:
1530 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1531 ** attempting to create a package with a colliding nickname causes
1532 correctable errors to be signalled.
1533 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1534 ** several x86-64 backend bugs related to sign-extension of immediate
1537 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1538 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1539 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1541 * fixed loading of multiply forward-referenced layouts.
1542 (thanks to Cheuksan Wang)
1543 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1544 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1545 (see COMPILE-FILE documentation for details of :PRINT option)
1546 * fixed bugs 19 and 317: fixed-format floating point printing is
1547 more accurate. This also fixes a bug reported by Adam Warner
1548 related to the ~@F format directive.
1549 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1551 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1552 dispatch table if the from-char is a dispatch macro character.
1553 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1554 by the wordsize no longer ignore the last word. (reported by Lutz
1556 * fixed bug in type checking of dynamic-extent variables. (reported
1558 * optimization: sequence traversal functions use their freedom to
1559 coerce function designators to functions.
1560 * optimization: code with many calls to CLOS methods specialized on
1561 CLOS classes has had redundant type checks removed.
1562 * optimization: added declarations to speed up operations that access
1563 the internal character database (for example STRING-UPCASE)
1564 * optimization: comparison operations between floats and sufficiently small
1565 fixnums no longer create extra rationals
1566 * fixed some bugs related to Unicode integration:
1567 ** portions of multibyte characters at the end of buffers for
1568 character-based file input are correctly transferred to the
1569 start of the buffer at the next read.
1570 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1571 passing it through to OPEN.
1572 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1573 argument given to internal calls to OPEN.
1574 ** the built-in comment readers (introduced by character sequences
1575 ";" and "#|") are more forgiving to encoding errors; they will
1576 STYLE-WARN and then attempt to resync the stream at a character
1577 boundary. (thanks to Teemu Kalvas)
1578 * fixed some bugs in the x86-64 port:
1579 ** Negative short int return values from c-calls are sign-extended
1581 ** The stack is aligned to 16-bytes for c-calls, as required by
1582 the ABI. (thanks to Cheuksan Wang)
1583 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1584 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1585 for enums. (thanks to Vincent Arkesteijn)
1586 ** Multiple small optimizations and bugfixes for floating point
1588 * fixed some bugs revealed by Paul Dietz' test suite:
1589 ** Space, Tab, Linefeed, Return and Page have the invalid
1590 secondary constituent character trait.
1591 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1593 ** WITH-INPUT-FROM-STRING should only update the index place on
1595 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1596 works more reliably.
1597 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1598 with one-dimensional arrays with fill pointers.
1599 ** TYPE-OF failed on a complex with an integer realpart and a
1601 ** compiler failure during type inference for the code of form
1602 (IF (EQL X (THE ...)) ...) (MISC.535).
1604 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1605 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1606 platform. The port supports 61-bit fixnums, large memory spaces
1607 and reloading shared object files.
1608 * enhancement: saving cores with foreign code loaded is now
1609 supported on ppc/Darwin in addition to the previously supported
1611 * enhancement: the statistical profiler now walks deeper into the
1612 call stack for more meaningful call-graphs and accrued time
1613 reports (x86/x86-64 only). It also now reports time spent in
1615 * enhancement: it is now possible to trace most individual methods
1616 of a generic function in addition to tracing the generic function
1618 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1620 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1621 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1622 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1623 produces an error. (thanks to Vincent Arkesteijn)
1624 * bug fix: NAMESTRING on pathnames with :WILD components in their
1625 directories works correctly. (thanks to Artem V. Andreev)
1626 * fixed bug 125: compiler preserves identity of closures. (reported
1628 * bug fixed: functions with &REST arguments sometimes failed with
1629 "Undefined function" when compiled with (DEBUG 3). (reported by
1631 * bug fix: overflow during compiling of setting element of a bit
1632 vector with constant index and value. (reported by Timmy Douglas)
1633 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1634 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1635 * on x86 compiler partially supports stack allocation of dynamic-extent
1637 * GO and RETURN-FROM do not check the extent of their exit points
1638 when compiled with SAFETY 0.
1639 * fixed some bugs related to Unicode integration:
1640 ** encoding and decoding errors are now much more robustly
1641 handled; it should now be possible to recover even from invalid
1642 input or output to the terminal. (thanks to Teemu Kalvas)
1643 ** provided a first cut at implementing STRING-TO-OCTETS and
1644 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1645 ** altered the SB-MD5 contributed module slightly, changing the
1646 interface just enough for it to be supportable for builds where
1647 lisp characters are not eight bits.
1648 * fixed some bugs revealed by Paul Dietz' test suite:
1649 ** the FORMATTER-generated functions for ~V[ conditionals require
1650 the correct number of arguments.
1651 ** READ-FROM-STRING returns the mandated second value when applied
1652 to displaced strings.
1653 ** the #\Rubout and #\Backspace characters are treated as invalid
1654 constituent characters by the tokenizer.
1656 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1657 * new feature: reloading changed shared object files with
1658 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1659 * new feature: references to foreign variables and functions
1660 can now be compiled and loaded before the corresponding shared
1661 object file is loaded, as long as the foreign definitions are
1662 available at runtime.
1663 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1664 supported. (thanks to Dan Debertin)
1665 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1666 just keywords. (thanks to Vincent Arkesteijn)
1667 * fixed bug #331: structure-class instances corresponding to
1668 DEFSTRUCT forms are now created eagerly.
1669 * fixed bug #345: backtraces from calls to undefined functions work
1670 on x86 as well. Related bug #61 is now also partially fixed on x86
1671 (backtraces from throws to unknown catch tags.)
1672 * bug fix: lambda-list parsing is now stricter vrt. order and number
1673 of lambda-list keywords.
1674 * bug fix: as specified by AMOP, an error is signalled if a
1675 class-option appears multiple times in a DEFCLASS form. (reported
1677 * bug fix: RANDOM can be compiled when the compiler derives the type
1678 of its numeric argument as a disjoint set of small integers.
1679 (reported by Paul Dietz)
1680 * bug fix: starting a core saved with shared objects loaded when
1681 those objects are not available no longer causes threaded SBCL to
1682 hang. (reported by Sean Ross)
1683 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1685 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1686 argument, which must always be NIL. (reported by Kalle Niemitalo)
1687 * bug fix: printing 1.0d+23 no longer results in an error.
1688 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1689 * bug fix: structure slot setters preserve evaluation order. (thanks
1691 * fixed some bugs related to Unicode integration:
1692 ** RUN-PROGRAM can allow its child to take input from a Lisp
1693 stream. (reported by Stefan Scholl)
1694 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1695 locales. (reported by Ken Causey)
1696 * fixed some bugs revealed by Paul Dietz' test suite:
1697 ** INCF, DECF and REMF evaluate their place form as specified in
1699 ** FORMATTER expands ~{ ~} iteration directives with V or #
1700 parameters correctly.
1701 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1702 consequent uses no arguments correctly.
1703 ** the system has a partial understanding of the (COMPLEX RATIO)
1705 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1708 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1709 * new feature: a build-time option (controlled by the :SB-UNICODE
1710 keyword feature, enabled by default) for building the system with
1711 support for the entire 21-bit character space defined by the
1713 * new feature: the system now has rudimentary external-format
1714 support; the primary user-visible change at this time is that
1715 characters with the high bit set (such as lower-case-e-acute) will
1716 print correctly to a terminal in a UTF-8 environment.
1717 * minor incompatible change: BASE-CHAR no longer names a class;
1718 however, CHARACTER continues to do so, as required by ANSI.
1719 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1720 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1721 has been moved to the SB-EXT package (temporarily re-exported from
1723 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1724 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1726 * fixed bug #351: better error-handlind and reporting for malformed
1728 * fixed bug #350: bignum-printing is now more memory-efficient,
1729 allowing printing of very large bignums, eg. (expt 2 10000000).
1730 (reported by Bruno Haible)
1731 * fixed bug #302: better primitive-type selection for intersection
1733 * fixed bug #308: non-graphic characters now all have names, as
1734 required. (reported by Bruno Haible)
1735 * bug fix: redefining a class with different superclasses now correctly
1736 removes it from the direct-subclasses of its previous superclasses.
1737 (reported by David Morse)
1738 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1739 new value now works. (reported by Bruno Haible)
1740 * bug fix: correct canonicalization of multiple non-standard slot
1741 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1742 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1743 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1744 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1745 now exists, an signals an error.
1746 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1747 during macroexpansion for non-iteration variables as well. (reported
1748 by Bruno Haible for CMUCL)
1749 * bug fix: Cyclic structures and unprintable objects in compiler
1750 messages no longer cause errors. (reported by Bruno Haible)
1751 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1752 and READ-FROM-STRING all now return a primary value of NIL if
1753 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1754 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1755 (reported by Bruno Haible for CMUCL)
1756 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1757 do not designate a condition. (reported by Bruno Haible for
1759 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1760 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1761 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1762 permitting resolution in favour of any of the conflicting symbols.
1763 (reported by Bruno Haible for CMUCL)
1764 * bug fix: EQUAL compiler optimizations is less aggressive on
1765 strings which can potentially compare true despite having distinct
1766 specialized array element types.
1767 * bug fix: unit enumerations can be defined without dividing by
1768 zero. (thanks to Vincent Arkesteijn)
1769 * FORMAT compile-time argument count checking has been enhanced.
1770 (report from Bruno Haible for CMUCL)
1771 * a partial workaround for the bug 262: the compiler does not try to
1772 inline-expand a local function doing RETURN-FROM from a deleted
1773 BLOCK. (thanks to Peter Denno for the bug report and to David
1774 Wragg for the simple test case)
1775 * fixed some bugs revealed by Paul Dietz' test suite:
1776 ** RENAME-PACKAGE allows all package designators as new package
1778 ** constraint propagation and lambda variable substitution are
1779 more cautious in dealing with partially deleted code.
1780 ** compiler sometimes lost reoptimization passes.
1781 ** CERROR, given a condition as condition designator, passes its
1782 remaining arguments to the continue format control without
1784 ** Case-altering FORMAT directives work correctly on non-ASCII
1786 ** The REAL type specifier handles bounds outside the floating
1787 point ranges without signalling FLOATING-POINT-OVERFLOW.
1788 ** Functions with IR1-transformations can create intercomponent
1789 references to global functions.
1790 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1792 ** FORMAT ~R treats a nil value for its first parameter correctly.
1794 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1795 * enhancement: saving cores with foreign code loaded is now
1796 supported on x86/NetBSD and sparc/Linux in addition to the previously
1797 supported platforms.
1798 * bug fix: on some platforms repeated installations caused multiple
1799 copies of HTML documentation to be installed -- should not happen
1800 any more. (reported by Stefan Scholl)
1801 * bug fix: parsing self-recursive alien record types multiple times
1802 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1803 original patch by Helmut Eller for CMUCL)
1804 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1805 (thanks to Richard Kreuter)
1806 * bug fix: defining classes whose accessors are methods on existing
1807 generic functions in other (locked) packages no longer signals
1808 bogus package lock violations. (reported by François-René Rideau)
1809 * bug fix: special variables as DEFMETHOD parameters no longer have
1810 associated bogus type declarations. (reported by David Wragg and
1812 * bug fix: read-write consistency on streams of element-type
1813 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1814 * bug fix: redefiniton of the only method of a generic function with
1815 no DEFGENERIC no longer emits a full WARNING. In addition,
1816 redefinition of generic functions with no DEFGENERIC to an
1817 incompatible lambda list now signals an error. (thanks to Zach
1819 * bug fix: DEFGENERIC now works even when there's a function of the
1820 same name in an enclosing lexical environment. (thanks to Zach
1822 * fixed compiler failure, caused by instrumenting code during
1823 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1824 * optimization: added loop analysis and improved register allocation
1825 * fixed some bugs revealed by Paul Dietz' test suite:
1826 ** POSITION on displaced vectors with non-zero displacement
1827 returns the right answer.
1828 ** (SIMPLE-STRING) is a valid type specifier for sequence
1830 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1832 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1833 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1835 ** The justification version of the FORMAT ~< directive treats
1836 non-zero minpad parameter correctly.
1838 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1839 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1840 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1841 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1842 the supported interface.
1843 * new feature: Single-stepping of code compiled with DEBUG 2 or
1844 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1845 * new feature: saving cores with foreign code loaded is now
1846 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1847 Timothy Moore's work for CMUCL)
1848 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1849 parameters to * instead of NIL if no initform is supplied.
1850 (reported by Johan Bockgaard)
1851 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1852 correctly when FUNCALL appears as the car of the form. Note:
1853 despite this FUNCALL forms are not currently subject to
1854 compiler-macro expansion. (port of Raymond Toy's fix for the
1855 same from CMUCL, reported by Johan Bockgaard)
1856 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1857 (thanks for Teemu Kalvas)
1858 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1859 case instead of terminating the iteration (thanks for Julian
1860 Squires, Sean Champ and Raymond Toy)
1861 * bug fix: incorrect expansion of defgeneric that caused a style
1862 warning. (thanks for Zach Beane)
1863 * bug fix: slot accessor effective method computation works properly
1864 for classes with multiple non-standard applicable methods on
1865 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1866 * on x86 compiler supports stack allocation of results of LIST and
1867 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1868 CMUCL implementation by Gerd Moellmann)
1869 * fixed some bugs revealed by Paul Dietz' test suite:
1870 ** FORMAT strings with both the ~<~:;~> form of the justification
1871 directive and pretty-printing directives cause an error.
1873 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1874 * incompatible change: the internal functions
1875 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1876 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1877 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1878 instead of the old functions.
1879 * new feature: on platforms where "dladdr" is available foreign
1880 function names now appear in backtraces. (based on Helmut Eller's
1882 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1883 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1885 * documentation: ASDF manual is now distributed with SBCL.
1886 * bug fix: SBCL can now load its contributed modules with REQUIRE
1887 even if the system-provided entries have been removed from
1888 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1890 * bug fix: backtraces involving undefined functions or assembly
1891 routines are more informative. (thanks to Brian Downing)
1892 * bug fix: mutually referent alien structures now work correctly.
1893 (reported by Rick Taube)
1894 * bug fix: structures defined by WITH-ALIEN can be referred to
1895 within other definitions in the same WITH-ALIEN.
1896 * bug fix: division operators (MOD, TRUNCATE and the like) with
1897 constant zero divisors and integer dividends no longer generate
1899 * bug fix: provide default methods for INPUT-STREAM-P and
1900 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1901 * bug fix: improve the detection and resolution of MOP metacycles
1902 (where effective method computation depends on the generic function
1903 having its effective method computed). (reported by Bruno Haible)
1904 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1905 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1906 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1907 * bug fix: pathnames with relative directory components can be
1908 represented relative to default pathnames.
1909 * optimization: in taking the GCD of bignums, reduce the two bignums
1910 to approximately the same size (using Euclid's algorithm) before
1911 applying the more sophisticated binary GCD. (thanks to Juho
1913 * optimization: COUNT on bitvectors now operates word-at-a-time.
1914 * optimization: ASH with a positive, but not necessarily constant,
1915 (leftwards) shift, when in a modular context, is compiled to a
1917 * fixed some bugs revealed by Paul Dietz' test suite:
1918 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1919 if the corresponding argument is NIL.
1920 ** FORMAT directives accepting a minpad parameter treat negative
1922 ** literal commas as character parameters to FORMAT directives are
1924 ** literal spaces directly after ~<Newline> directives within a
1925 format-logical-block (~:< ~@:>) do not induce :FILL-style
1926 conditional newlines.
1927 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1928 as specified: it no longer includes conditional newlines).
1929 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1931 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1932 :PER-LINE-PREFIX argument does not evaluate to a string.
1934 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1935 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1936 the manual for details; for now, package locks can be disabled by
1937 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1938 if no major problems are found then it is likely that they will be
1939 compiled in unconditionally.
1940 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1941 unsupported operators on all platforms. To load a shared library
1942 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1943 object file, link it into a shared library outside of SBCL and
1944 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1946 * bug workaround: ROOM T and the GC were not working together
1947 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1948 reporting facility aren't preserved. That reporting has been
1949 disabled, so now until and unless someone figures out how to make it
1950 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1951 * minor incompatible change: as threatened around sbcl-0.8.0, the
1952 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1953 The symbols are also exported from SB-PCL for backwards
1954 compatibility, but more so than before SB-PCL should be treated as
1955 an implementation-internal package.
1956 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1958 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1959 It is known as of this release not to work on the Alpha, however.
1960 * fixed bug #167: errors signalled due to illegal syntax in method
1961 bodies are now more legible.
1962 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1963 designators and can hence be used with TYPEP.
1964 * fixed bug #333: CHECK-TYPE now ensures that the type error
1965 signalled, if any, has the right object to be accessed by
1966 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1967 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1968 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1970 * fixed bug #334: programmatic addition of slots using specialized
1971 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1972 and :ALLOCATION :CLASS slots.
1973 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1974 float exponents. (rereported by Peter Seibel)
1975 * fixed a bug: #\Space (and other whitespace characters) are no
1976 longer considered to be macro characters in standard syntax by
1977 GET-MACRO-CHARACTER.
1978 * fixed bug: initialization of condition class metaobjects no longer
1979 causes an instance of the condition to be created. (reported by
1981 * fixed bug: it is now possible to have slots such that
1982 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1983 description is neither :INSTANCE nor :CLASS.
1984 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1985 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1986 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1987 non-standard SLOT-DEFINITION-ALLOCATION values.
1988 * fixed bugs: various race conditions handling exiting threads.
1989 CL-PPCRE's thread torture test now passes.
1990 * fixed arguably-a-bug: GC time proportional to number of threads in
1991 system even when most of them are idle
1992 * optimization: improved performance of BIT and SBIT on bit-vectors.
1993 * .fasl file incompatibility: The fasl file version number has
1994 been incremented because of changes associated with package locks.
1996 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1997 * minor incompatible change: the system no longer provides
1998 optimization hints (conditions which are TYPEP
1999 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2000 mode; these hints are still emitted when higher SPEED optimization
2002 * new contrib module: a sampling profiler (profiling by statistical
2003 sampling, rather than by instrumenting functions) is available as
2004 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2005 * the behaviour of the standard function ED is now customizeable by
2006 third parties through a hook variable: see ED's documentation
2007 string for information on the protocol.
2008 * the compiler no longer emits efficiency notes for (FUNCALL X)
2009 when the type of X is uncertain under default optimization
2011 * fixed bug 276: mutating a binding of a specialized parameter to a
2012 method to something that is not TYPEP the specializer is now
2014 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2015 exhibit the required behaviour on the x86 platform. (thanks to
2016 Peter van Eynde, Eric Marsden and Bruno Haible)
2017 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2018 even for difficult arguments. (reported by Peter Graves)
2019 * fixed bug 141a: the backquote printer now descends quoted
2021 * fixed another bug in backquote printing: no more destructive
2022 modification of the form's list structure. (reported by Brian
2024 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2025 move between its address being taken and the call to
2026 interrupt_thread, fixing a crashing race condition.
2027 * the SB-POSIX contrib implementation has been adjusted so that it
2028 no longer exhibits ridiculously poor performance when constructing
2029 instances corresponding to C structs.
2031 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2032 * minor incompatible change: the sb-grovel contrib now treats C
2033 structures as alien (in the sense of SB-ALIEN) objects rather than
2034 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2035 has implications for memory management of client code
2036 (sb-grovel-returned objects must now be manually managed) and for
2037 type safety (alien objects now have full types).
2038 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2039 used to control emission of compiler diagnostics, rather than the
2040 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2041 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2042 quality should be considered deprecated.
2043 * install.sh now installs the user manual as well
2044 * (not quite a new documentable feature, but worth considering in
2045 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2046 beginnings of a semantically meaningful condition hierarchy is
2047 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2049 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2050 designator as the defaults argument.
2051 * fixed bug: Displaced arrays whose displaced-to array has become
2052 too small now cause ARRAY-DIMENSION to signal an error, providing
2053 for safer bounds-checking. (reported by Bruno Haible)
2054 * fixed bug: DEFCLASS slot definitions with identical :READER and
2055 :WRITER names now signal a reasonable error. (reported by Thomas
2057 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2059 * fixed bug: Invalid slot specification errors now print correctly.
2060 (thanks to Zach Beane)
2061 * fixed bug 320: Shared to local slot value transfers in class
2062 redefinitions now happen corrently from superclasses as
2063 well. (reported by Bruno Haible)
2064 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2066 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2067 improper lists correctly. (reported by Bruno Haible)
2068 * fixed bug 313: source-transform for <fun-name> was erroneously
2069 applied to a call of a value of a variable with name <fun-name>.
2070 (reported by Antonio Menezes Leitao)
2071 * fixed bug 307: The obsolete instance protocol ensures that
2072 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2073 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2074 remove dead unknown-values globs from the middle of the stack.
2075 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2076 easier to resume long computations after using *BREAK-ON-SIGNALS*
2077 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2078 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2079 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2080 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2081 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2082 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2084 * fixed a (fairly theoretical) bug in string printing: if
2085 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2086 does not have array-element-type equal to the most general string
2088 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2089 function name. (thanks to Nikodemus Siivola)
2090 * fixed bug: compile-time format string checker failed on
2091 non-closed ~{. (reported by Thomas F Burdick)
2092 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2093 don't warn on legal loop code involving more than one aggregate
2094 boolean. (thanks to Nikodemus Siivola)
2095 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2096 in keyword position to MAKE-PACKAGE was not being reported as
2098 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2099 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2101 * fixed bugs 280 and 312: the checking for multiple definitions in a
2102 file is less likely to become confused by uses of inline
2104 * fixed bug: the #S reader macro performs the keyword coercion
2105 specified for slot names. (reported by Kalle Niemitalo)
2106 * fixed bug: lambda lists may contain symbols whose names start with
2107 & but are not lambda-list-keywords; their occurrence triggers a
2109 * fixed bug 321: define-method-combination argument lambda lists do
2110 not cause invalid code to be generated when &optional or &aux
2111 variables have default values. (reported by Bruno Haible)
2112 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2113 have CLOS classes; this appears also to have fixed reported
2114 instability in stack exhaustion detection.
2115 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2116 MUFFLE-WARNING when no associated restart is present is now
2118 * optimization: rearranged the expansion of various defining macros
2119 so that each expands into only one top-level form in a
2120 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2122 * optimization: used a previously little-used slot in symbols to
2123 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2125 * fixed some bugs revealed by Paul Dietz' test suite:
2126 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2128 ** Fixed an optimization bug related to inheritance of initforms
2129 from local to shared slots.
2130 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2131 ** CAST optimizer forgot to flush argument derived type.
2132 ** print/read consistency on floats is now orders of magnitude
2133 more likely. (thanks also to Bruno Haible for a similar report
2135 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2136 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2137 for objects of type REAL. Make it so.
2138 ** SXHASH is specified (infelicitously) to respect similarity,
2139 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2140 it so. (thanks to Markus Ziegler)
2141 ** on the Alpha, the compiler succeeds in compiling functions
2142 returning a known number of arguments greater than 63.
2143 ** fixed handling of invalid NIL arguments in keyword position
2145 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2147 ** provided a readably-printable representation for RANDOM-STATE
2149 ** ensured that pathnames, where they have a namestring, always
2150 print using #P"..." syntax.
2152 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2153 * Support for the forthcoming 2.0 version of the NetBSD kernel
2154 running on x86 hardware has been added. (thanks to Perry
2155 E. Metzger most immediately, and others for their past work)
2156 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2157 current release version; SBCL's OpenBSD support had been broken
2158 since about the time of OpenBSD's switch to ELF binary format.)
2159 * [placeholder for DX summary]
2160 ** user code with &REST lists declared dynamic-extent, under high
2161 speed or space and low safety and debug optimization policy.
2162 * The manual has been converted to Texinfo format and the debugger
2163 chapter from the cmucl manual has been added.
2164 * A facility has been added to extract documentation strings from
2165 sbcl and store them as Texinfo-formatted snippets for inclusion in
2166 the manual (via Texinfo's @include directive)
2167 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2168 the test case to Dave Roberts)
2169 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2170 pointers. (reported by Sean Ross)
2171 * bug fix: PROFILE output is printed nicely even for large numerical
2172 values. (thanks to Zach Beane)
2173 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2174 greater than 32 handle EOF correctly.
2175 * bug fix: on X86 an immediate argument of the IMUL instruction is
2176 correctly printed in disassembly. (thanks to Lutz Euler)
2177 * bug fix: class slots in redefined classes preserve their old
2178 values. (thanks to Bruno Haible and Nikodemus Siivola)
2179 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2180 succeeds. (reported by Marco Baringer)
2181 * fixed some bugs revealed by Paul Dietz' test suite:
2182 ** READ-SEQUENCE now works on ECHO-STREAMs.
2183 ** RATIONALIZE works more according to its specification. (thanks
2186 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2187 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2188 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2189 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2190 should matter to you only if you rebind the printer control
2191 variables and then find you want different bindings in the
2192 debugger than in the ordinary execution of your program.)
2193 * The runtime build system has been tweaked to support building
2194 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2195 assembler and linker. (thanks to Nikodemus Siivola)
2196 * Unbound, undefined, undeclared variables now trigger full
2197 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2198 more often programmer error than deliberate exploitation of undefined
2200 * optimization: the hash algorithm for strings has changed to one
2201 that is less vulnerable to spurious collisions. (thanks to Juho
2203 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2204 needless bounds checking. (thanks to Juho Snellman)
2205 * optimization: implemented multiplication as a modular
2206 (UNSIGNED-BYTE 32) operation on the PPC backend.
2207 * fixed some bugs revealed by Paul Dietz' test suite:
2208 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2210 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2211 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2212 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2214 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2216 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2218 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2219 their output stream on EOF from read.
2220 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2221 have been read to end-of-file.
2222 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2224 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2225 description of determination of which consecutive characters
2227 ** Printing the "Space" character with escaping on now yields "#\\ ",
2228 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2229 ** Reading floating-point numbers with *READ-BASE* set to a number
2230 less than 10 works correctly.
2231 ** Reading floating-point numbers with *READ-BASE* set to a number
2232 more than 10 works correctly.
2233 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2234 the readtable currently in effect.
2236 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2237 * minor incompatible change: parsing of namestrings on a physical
2238 (Unix) host has changed; numbers after the final #\. in a
2239 namestring are no longer interpreted as a version field. This is
2240 intented to be largely invisible to the user, except that the
2241 meaning of the namestring "*.*.*" has changed: it now refers to a
2242 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2243 should usually be replaced by
2244 (make-pathname :name :wild :type :wild :version :wild)
2245 with the added benefit that this is more likely to be portable.
2246 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2247 option to OPEN now signals an error if the file being opened
2248 exists; this may have an impact on existing code.
2249 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2250 signals received, on the PowerPC platforms (both Linux and
2251 Darwin). (thanks to Pierre Mai for pointing out the location of
2253 * several fixes on OS X: The system now builds and runs cleanly on
2254 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2255 (thanks to Brian Mastenbrook)
2256 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2257 second-resolution: integer multiples of 1/3600 between -24 and 24.
2258 (thanks to Vincent Arkesteijn)
2259 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2260 type when called with 1 argument; PEEK-CHAR checked type of
2261 PEEK-TYPE only after having read first character from a
2262 stream. (reported by Peter Graves)
2263 * bug fix: the garbage collector now has much better locality
2264 behaviour, and in particular no longer treats all memory as being
2265 exhausted when many small objects point to each other in a deeply
2267 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2268 recognized as being TYPEP their class.
2269 * bug fix: the PUSHNEW documentation string has been corrected.
2270 (thanks to Vincent Arkesteijn)
2271 * bug fix: defaulting of the value for the last of an atypically
2272 large number of multiple values being bound was not being
2273 performed correctly on the Alpha or PPC platforms
2274 * optimization: implemented multiplication as a modular
2275 (UNSIGNED-BYTE 32) operation on the x86 backend.
2276 * optimization: SEARCH on simple-base-strings can now be open-coded.
2277 (see also contrib/compiler-extras.lisp for inspiration for
2278 teaching the compiler about the Boyer-Moore algorithm).
2279 * value, returned by MAX (and MIN) called with several EQUALP, but
2280 not EQL, arguments now does not depend on compiler settings.
2281 * fixed some bugs revealed by Paul Dietz' test suite:
2282 ** in stack analysis liveness information is propagated from
2283 non-local entry points.
2284 ** pathwise CAST removing failed when the CAST node did not start
2286 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2287 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2289 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2290 host is already defined.
2291 ** RENAME-FILE works on streams instead of signalling an internal
2293 ** PEEK-CHAR uses the current readtable when determining whether
2294 or not a character is whitespace.
2295 ** MERGE-PATHNAMES handles the case when the pathname does not
2296 specify a name while the default-pathname specifies a version.
2297 ** Pathnames now stand a better chance of respecting print/read
2299 ** Attempting to use standardized file system operators with a
2300 pathname with invalid :DIRECTORY components signals a
2302 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2304 ** DIRECTORY on logical pathnames is more correct.
2305 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2306 signal a TYPE-ERROR if their argument is not a stream.
2307 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2308 designator argument does not designate a stream.
2309 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2310 examining the synonym.
2311 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2313 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2314 element-type larger than ([UN]SIGNED-BYTE 32).
2316 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2317 * When built with the :SB-FUTEX feature, threaded builds now take
2318 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2319 for faster/more reliable mutex and condition variable support.
2320 * Incompatible change (but one you probably shouldn't have been using
2321 anyway): the interface and code for arbitrating between multiple
2322 threads in the same user session has been redesigned.
2323 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2324 error when called without an explicit environment argument.
2325 (thanks to Nikodemus Siivola)
2326 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2327 confused about their position. (thanks to Adam Warner and Gerd
2329 * bug fix: Pretty printing backquoted forms with unquotations in the
2330 argument list position of various code constructs such as LAMBDA
2331 now works correctly. (reported by Paul Dietz)
2332 * bug fix: Pretty printing unquotations no longer loses all
2333 stream position information.
2334 * optimization: performance of string output streams is now less
2335 poor for multiple small sequence writes.
2336 * optimization: performance of CSUBTYPEP in the presence of complex
2337 expressions involving CONS and NOT many times has been improved.
2338 (reported by Paul Dietz)
2339 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2341 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2342 streams for the REPL, for the debugger, and so on. (thanks to
2344 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2345 * fixed some bugs revealed by Paul Dietz' test suite:
2346 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2347 is ignored if it is too small, rather than propagating through
2348 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2349 ** extremely complex negations of CONS types were not being
2350 sufficiently canonicalized, leading to inconsistencies in
2352 ** VALUES tranformer lost derived type.
2354 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2355 * fixed a bootstrapping bug: the build process no longer assumes
2356 that the various BOOLE-related constants have the same value in
2357 host and target lisps. (noted by Paul Dietz' test suite on an
2358 SBCL binary built from CLISP)
2359 * The system can now be dynamically linked on the MIPS platform,
2360 which enables dynamic loading of foreign code from Lisp. (thanks
2361 to Ralf Baechle for discussions on the MIPS ABI)
2362 * The system now records debugging information for its own source
2363 files in a filesystem-position-independent manner, relative to
2364 the "SYS" logical host.
2365 * fixed a compiler bug: MV-LET convertion did not check references
2366 to the "max args" entry point. (reported by Brian Downing)
2367 * tweaked disassembly notes to be less confident about proclaiming
2368 some instruction as an LRA. (thanks to Brian Downing)
2369 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2370 now each have their own history, command character, and other
2371 characteristics. (thanks to David Lichteblau)
2372 * fixed some bugs revealed by Paul Dietz' test suite:
2373 ** compiler failure in compiling LOGAND expressions including a
2375 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2376 shift greater than 32.
2377 ** FLUSH-DEST did not mark blocks for type check regeneration.
2378 ** HANDLER-CASE failed to accept declarations in handler clauses
2379 in some circumstances.
2381 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2382 * New code in contrib/sb-introspect (still probably not entirely
2383 stable yet) provides some support for smart Lisp development
2384 environments like SLIME.
2385 * The conditions signalled for errors occurring when loading .fasl
2386 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2387 in a way which should help ASDF recover gracefully.
2388 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2389 stuff has been cleaned up. If you code contrib/ stuff, this might
2390 affect you, and you can look at contrib/README, contrib/STANDARDS,
2391 and/or the 0.8.4.27 diff to check.
2392 * In full calls the compiler now does not generate checks for declared
2393 argument types for all arguments.
2394 * various threading fixes
2395 ** and some experimental patches which didn't make it into
2396 the main tree for this release, but which are shipped in
2397 contrib/experimental-thread.patch as a possible fix for some
2398 failures (deadlock, spinning...) in GC-intensive multithreaded
2400 * fixed PPC build problem (source code incompatibility of different
2401 library versions): added offsetof() hackery which attempts to divine
2402 where glibc maintainers put uc_mcontext today
2403 * fixed bug 282: compiler does not trust type assertions while passing
2404 arguments to a full call.
2405 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2406 &OPTIONAL VALUES type parameter.
2407 * fix bug 214: algorithm for noting rejected templates is now more
2408 similar to that of template seletion. (also reported by rydis on
2410 * fixed bug 141b: printing backquoted information readably and prettily
2411 inserts a space where necessary.
2412 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2413 (CHARACTER 10) are now reported as errors, rather than propagated
2414 as unknown types. (reported by piso on #lisp)
2415 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2416 with values NIL and :ERROR. (thanks to Milan Zamazal)
2417 * fixed bug 191c: CLOS now does proper keyword argument checking as
2418 described in CLHS 7.6.5 and 7.6.5.1.
2419 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2420 longer raise an error; further, using a list as a for-as-arithmetic
2421 counter now raises a meaningful error.
2422 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2423 now understood by sequence creation functions such as MAKE-SEQUENCE
2425 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2426 a STRING-INPUT-STREAM.
2427 * compiler enhancement: SIGNUM is now better able to derive the type
2429 * type declarations inside WITH-SLOTS are checked. (reported by
2431 * fixed some bugs revealed by Paul Dietz' test suite:
2432 ** incorrect optimization of TRUNCATE for a positive first
2433 argument and negative second.
2434 ** compiler failure in let-convertion during flushing dead code.
2435 ** compiler failure while deriving type of TRUNCATE on an
2436 interval, containing 0.
2437 ** ASH of a negative bignum by a negative bignum count now returns
2439 ** intersection of CONS types now canonicalizes properly, fixing
2440 inconsistencies in SUBTYPEP.
2442 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2443 * incompatible change: The --disable-debugger command line
2444 option now clobbers the debugger at a more fundamental
2445 level, by redefining #'INVOKE-DEBUGGER instead of by
2446 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2447 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2448 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2449 enter the debugger REPL and then suffer recursive errors
2450 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2451 while under the new system, BREAK in --disable-debugger mode
2452 terminates the system just as an unhandled error would.
2453 * fixed compiler performance when processing loops with a step >1;
2454 * bug fix: DOCUMENTATION now retrieves generic function
2455 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2456 support has been systematized, and now supports the methods
2457 specified by ANSI, along with a default method and a method for
2458 slot documentation. (reported by Nathan Froyd)
2459 * bug fix: effective methods associated with a generic function are
2460 no longer cached over a change of that generic function's method
2461 combination. (reported by Andreas Fuchs)
2462 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2463 names a non-existent directory. (thanks to Andreas Fuchs)
2464 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2465 the PPC platform, lead to stack corruption; this has been fixed.
2466 (reported by Rainer Joswig)
2467 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2468 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2469 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2470 arithmetic in general, is now much more reliable on the PPC
2472 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2473 platform now returns the right answer.
2474 * optimization: restored some effective method precomputation in
2475 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2476 precomputation is now tunable.
2477 * optimization: compiler-internal data structure use has been
2478 reviewed, and changes have been made that should improve the
2479 performance of the compiler by about 20%.
2480 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2481 simple-base-strings and simple-bit-vectors is improved.
2482 * optimization: the optimization of 32-bit logical and arithmetic
2483 functions introduced in version 0.8.3 on the x86 has been
2484 implemented on the mips, ppc and sparc platforms; an
2485 implementation of the same facility, but for 64-bit arithmetic,
2486 has been added for the alpha.
2487 * microoptimization: the compiler is better able to make use of the
2488 x86 LEA instruction for multiplication by constants.
2489 * bug fix: in some situations compiler did not report usage of
2490 generic arithmetic in (SPEED 3) policy.
2491 * bug 145b fix: compiler used wrong type specifier while converting
2492 MEMBER-types to numeric.
2493 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2495 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2496 index argument now works properly on the Alpha platform.
2497 * bug fix: floating point exception treatment on the Alpha platform
2499 * bug fix: FILE-POSITION works much better on string input and
2500 output streams. (thanks to Nikodemus Siivola)
2501 * bug fix: many threading/garbage collection symptoms sorted.
2502 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2503 might be pseudo-atomic.
2504 * internal change: Stopping for GC is now done with signals not
2505 ptrace. GC is now done in whichever thread wanted it, instead of
2507 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2509 * bug fix: result form in DO is not contained in the implicit
2511 * incompatible change: ICR structure is changed; the value part of
2512 CONTINUATION is now called LVAR; corresponding functions are
2513 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2514 * added type deriver for ISQRT (thanks to Robert E. Brown).
2515 * bug fix: better support for loading from the command line when an
2516 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2518 * fixed some bugs revealed by Paul Dietz' test suite:
2519 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2520 ** ROUND and FROUND now give the right answer when given very
2521 small float arguments.
2522 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2524 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2525 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2526 ** LCM should return a non-negative integer.
2527 ** PARSE-INTEGER returned the index of a terminator instead of the
2528 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2529 ** PARSE-INTEGER returned an incorrect index being applied to a
2531 ** LCM with two arguments of 0 returns 0 rather than signalling
2533 ** unsigned addition of a 32-bit constant with the high bit set no
2534 longer causes an internal compiler error on the x86.
2535 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2536 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2537 with negative last argument.
2538 ** byte specifiers with zero size and position no longer cause
2539 an error during type derivation.
2540 ** bignum multiplication on the Alpha platform now returns the
2542 * porting: The system now builds on SuSE AMD64, although it still
2543 generates a 32-bit binary.
2544 * .fasl file incompatibility: The fasl file version number has
2545 been incremented (because of the changes to internal compiler
2546 data structures referred to above).
2548 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2549 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2550 more accurately, on the Darwin kernel running on PowerPC hardware.
2551 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2552 * Compiler code deletion notes now signal a condition of type
2553 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2554 an associated MUFFLE-WARNING restart.
2555 * The compiler now performs limited argument count validation of
2556 constant format strings in FORMAT, and where appropriate in ERROR,
2557 CERROR and WARN. (thanks to Gerd Moellmann)
2558 * New ASDF-INSTALL contrib can be used for automatic download and
2559 installation of third-party Lisp code from CCLAN or other sites
2561 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2562 which forces another thread to execute a function supplied by the
2564 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2565 now accept and act upon their :ELEMENT-TYPE keyword argument.
2566 (reported by Martin Atzmueller, Edi Weitz)
2567 * bug fix: FILE-POSITION now accepts position designators up to
2568 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2569 is the greater. (thanks to Patrik Nordebo)
2570 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2571 Kalle Olavi Niemitalo)
2572 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2573 printable. (reported by Eric Marsden)
2574 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2575 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2576 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2577 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2578 (thanks to Stig Sandoe)
2579 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2580 UNREAD-CHAR now correctly includes the unread character in the
2581 target sequence. (thanks to Gerd Moellmann)
2582 * bug fix in threaded builds: the system can now be suspended and
2583 resumed by shell job control with minimal disruption.
2584 * bug fixes in times and timezones >2038AD
2585 * better handling of "where is GNU make?" problem in build scripts
2586 (thanks to Nikodemus Siivola)
2587 * new optimization: inside a named function any reference to a
2588 function with the same name is considered to be a self-reference;
2589 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2590 optimization quality.
2591 * new optimization on x86: logical functions and + now have
2592 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2593 used when the result is truncated to 32 bits.
2594 * VALUES declaration is partially enabled.
2595 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2596 * fixed some bugs revealed by Paul Dietz' test suite:
2597 ** The system now obeys the constraint imposed by
2598 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2599 types form a lattice under type intersection.
2600 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2601 ** ASSOC now ignores NIL elements in an alist.
2602 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2603 and (1+ MOST-POSITIVE-FIXNUM) answers.
2604 ** The addition of a method with invalid qualifiers to a generic
2605 function does not cause an error to be signalled immediately;
2606 a warning is signalled, and the error is generated only on
2607 calling the generic function.
2608 * changed .fasl file version number, in part to add type codes for
2609 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2610 obscure ANSI requirements
2612 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2613 * fixed bug 148: failure to inline-expand a local function left
2614 garbage, confusing the compiler.
2615 * fixed bugs 3cd: structure slot readers perform type check if the
2616 slot can have an invalid value (i.e. it is either not initialized
2617 or can be written with a less specific slot writer).
2618 * bug fix: the compiler now traps array references to elements off
2619 the end of an array; previously, the bounds checking in some
2620 circumstances could go off-by-one.
2621 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2623 * type declarations for array element types now obey the description
2624 on the CLHS page "Declaration TYPE", as per discussions on
2625 sbcl-help around 2003-05-08. This means that a declaration
2626 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2627 declaration, all references to BAR will be asserted or assumed
2628 (with THE, so dependent on compiler policy) to involve objects of
2629 type FOO. Note that no such declaration is implied in
2630 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2631 * declared types of functions from the "Conditions"
2632 chapter. (reported by Paul Dietz)
2633 * bug fix: CERROR accepts a function as its first argument.
2634 * bug fix: NTH an NTHCDR accept a bignum as index
2635 arguments. (reported by Adam Warner)
2636 * optimization: character compare routines now optimize comparing
2637 against a constant character. (reported by Gilbert Baumann)
2638 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2639 argument now works properly on the MIPS platform.
2640 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2641 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2642 declared to have a functional type, does not cause an error
2644 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2645 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2646 * fixed bug: if last continuation of a deleted block has a
2647 destination, this destination should be deleted too. (reported by
2649 * fixed a bug in the bootstrap process: the host compiler's values
2650 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2651 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2652 test case from Patrik Nordebo)
2653 * improved the ability of the disassembler on the PPC platform to
2654 provide helpful disassembly notes.
2655 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2656 the class in more cases than previously.
2657 * bug fix: FILE-POSITION now understands :START and :END for
2658 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2659 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2660 8) data. (thanks to David Lichteblau)
2661 * bug fix: it is possible to add a method to a generic function
2662 without lambda list.
2663 * bug fix: reader failed to signal END-OF-FILE inside an
2664 object representation. (reported by Nikodemus Siivola)
2665 * fixed some bugs revealed by Paul Dietz' test suite:
2666 ** LAST and [N]BUTLAST should accept a bignum.
2667 ** condition slot accessors are methods.
2668 ** (VECTOR NIL) is a subtype of STRING.
2670 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2671 * minor incompatible change: some nonsensical specialized lambda
2672 lists (used in DEFMETHOD) which were previously ignored now signal
2674 * minor incompatible change: the system is now aware of the types of
2675 variables in the COMMON-LISP package, and will signal errors for
2676 most violations of these type constraints (where previously they
2677 were silently accepted).
2678 * minor incompatible change: COMPILE-FILE now uses the freedom
2679 afforded (ANSI 3.2.2.3) to use derived function types for
2680 functions defined in the same file. This also permits the system
2681 to warn on static type mismatches and function
2682 redefinition. (Currently it does not work with high DEBUG level.)
2683 * minor incompatible change: VALUES declaration is disabled.
2684 * When issuing notes, the compiler now signals a condition of type
2685 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2686 restart for use in user handlers. It is expected that the
2687 COMPILER-NOTE condition will eventually become a condition
2688 supertype to a hierarchy of note types, which will then be
2689 handleable in a similar fashion. However, at the moment, no such
2690 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2691 but it's an implementation detail, not a classification for the
2693 * Changes in type checking closed the following bugs:
2694 ** type checking of unused values (192b, 194d, 203);
2695 ** template selection based on unsafe type assertions (192c, 236);
2696 ** type checking in branches (194bc).
2697 * A short form of VALUES type specifier has ANSI meaning (it has
2698 increased the number of situations when SBCL cannot perform type
2700 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2701 keywords or constants is permissible.
2702 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2703 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2704 classes. (thanks to Antonio Martinez)
2705 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2706 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2707 argument to a pretty-print stream. Instead, it leaves any such
2708 operations to DESCRIBE-OBJECT methods.
2709 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2711 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2712 lambda lists. (thanks to David Lichteblau)
2713 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2714 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2715 (thanks to Gerd Moellmann)
2716 * fixed bug 246: increased compilation speed of long
2717 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2719 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2721 * DEFCONSTANT now throws a condition of type
2722 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2723 constant to a non-EQL value; CONTINUE and ABORT restarts
2724 respectively change and preserve the value.
2725 * fixed bug 63: The code walker, part of the implementation of CLOS,
2726 is now better at handling symbol macros.
2727 * bug fix: There is no longer an internal implementation type named
2728 CL:LENGTH. (reported by Raymond Toy)
2729 * bug fix: In macro-like defining macros/special operators the
2730 implicit block does not enclose the lambda list.
2731 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2732 enclosed in parentheses) are not suitable as type specifiers, and
2733 their use properly signals an error now.
2734 * bug fix: An argument count mismatch for a type specifier in code
2735 being compiled no longer causes an unhandled error at compile
2736 time, but signals a compile-time warning.
2737 * fixed simple vector readable printing
2738 * bug fix: DESCRIBE takes more care over whether the class
2739 precedence list slot of a class is bound before accessing it.
2740 (reported by Markus Krummenacker)
2741 * bug fix: FORMATTER can successfully compile pretty-printer format
2742 strings which use variants of the ~* directive inside.
2743 * bug fix: SEARCH now applies its TEST predicate to the elements of
2744 the arguments in the correct order. (thanks to Wolfhard Buss)
2745 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2746 in type inference. (thanks to Robert E. Brown)
2747 * bug fix: Reading in symbols with an explicit package name of ""
2748 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2749 * fixed some bugs revealed by Paul Dietz' test suite:
2750 ** NIL is now allowed as a structure slot name.
2751 ** Arbitrary numbers, not just REALs, are allowed in certain
2752 circumstances in LOOP for-as-arithmetic clauses.
2753 ** Multiple class redefinitions before slot access no longer
2754 causes a type error.
2755 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2756 association between the name and a class.
2757 ** Generic functions with non-standard method-combination and over
2758 five methods all of which return constants no longer return NIL
2759 after the first few invocations. (thanks to Gerd Moellmann)
2760 ** CALL-NEXT-METHOD with no arguments now passes the original
2761 values of the arguments, even in the presence of assignment.
2762 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2764 ** DISASSEMBLE works with closures and funcallable instances.
2765 ** ADD-METHOD now returns the generic function, not the new method.
2766 ** FIND-METHOD signals an error if the lengths of the specializers
2767 is incompatible with the generic function, even if the ERRORP
2769 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2770 which its argument is a member.
2771 ** DEFCLASS only redefines the class named by its class-name
2772 argument if that name is the proper name of the class;
2773 otherwise, it creates a new class.
2774 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2775 of the UNBOUND-SLOT condition to the name of the slot.
2776 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2777 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2778 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2779 SLOT-MAKUNBOUND in the specified fashion.
2781 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2782 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2783 cross-compilation host. As a consequence, we can now bootstrap our
2784 way up to SBCL starting with a bare gcc toolchain and human-readable
2785 source code (first the source to CLISP, then the source to SBCL).
2786 * A contributed module containing a partial implementation of the
2787 simple-streams interface has been included. (thanks to Rudi
2789 * A contributed module implementing the RFC1321 Message Digest
2790 Algorithm, known as MD5, has been included.
2791 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2792 no longer has any effect, as the code controlled by this feature
2793 has been deleted. (As far as we know, no-one has ever built using
2794 this feature, and its semantics were confused in any case).
2795 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2796 work on conditions (as required by the ANSI specification),
2797 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2798 expected behaviour on conditions. Users should note, however,
2799 that such behaviour is not required by the ANSI specification,
2800 and so use of this behaviour may render their code unportable.
2801 * fixed some bugs revealed by Paul Dietz' test suite:
2802 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2804 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2805 ** SXHASH on condition objects no longer returns NIL.
2806 ** :ALLOCATION :CLASS slots are better treated; their values are
2807 updated on class redefinition, and initforms inherited from
2808 superclasses are applied.
2809 ** REMOVE-METHOD returns its generic function argument even when
2810 no method was removed.
2811 ** SHARED-INITIALIZE now initializes the values of the requested
2812 slots, including those with :ALLOCATION :CLASS.
2813 ** ALLOCATE-INSTANCE now works on structure classes defined via
2814 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2816 ** SLOT-EXISTS-P now works on conditions, as well as structures
2818 ** MAKE-LOAD-FORM now has the required methods on
2819 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2820 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2821 keyword, and now implements the SLOT-NAMES argument.
2822 ** methods with &OPTIONAL arguments no longer allow too many
2823 arguments to be passed in the call without error.
2824 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2825 option is consistent with the required arguments of the generic
2826 function lambda list.
2827 * bug fix: REQUIRE accepts a string designator. (Thanks to
2829 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2830 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2831 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2832 * bug fix: The compiler now checks for duplicated variables in macro
2834 * bug fix: SETQ on globals returns the correct value.
2835 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2836 not a COMPILER-ERROR (followed by some other strange error on
2837 choosing the CONTINUE restart).
2838 * bug fix: make.sh and friends are now more consistent in the way that
2839 they look for GNU "make".
2841 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2842 * experimental native threads support (on x86 Linux >=2.4 only).
2843 This is not compiled in by default: you need to add :SB-THREAD to
2844 the target features. See the "Beyond ANSI" chapter of the manual
2846 * fix for longstanding nonANSIism: The old distinction between
2847 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2848 The return value from CL:FIND-CLASS is now a CLOS class, and
2849 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2850 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2851 * An interface to the MetaObject Protocol, as described in Kiczales,
2852 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2853 MIT Press, 1991, is available from the SB-MOP package.
2854 * incompatible change: the SB-PCL package should now be considered
2855 a private implementation detail, and no longer a semi-private MOP
2857 * minor incompatible change: due to rearrangement for threads, the
2858 control stack and binding stack are now allocated at arbitrary
2859 addresses instead of being hardcoded per-port. Users affected by
2860 this probably have to be doing advanced things with shared
2861 libraries, and will know who they are.
2862 * minor incompatible change: Previously, all --eval forms used were
2863 processed with READ before any of them were processed with EVAL.
2864 Now each --eval form is processed with both READ and EVAL before
2865 the next --eval form is processed. (Thus package operations like
2866 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2867 work as the user might reasonably expect.)
2868 * minor incompatible change: *STANDARD-INPUT* is now only an
2869 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2871 * minor incompatible change: Y-OR-N-P is now character-oriented,
2872 not line oriented. Also, YES-OR-NO-P now works without errors.
2873 (thanks to Antonio Martinez)
2874 * sb-aclrepl module improvements: an integrated inspector, added
2875 repl features, and a bug fix to :trace command.
2876 * Known functions, which cannot be open coded by the backend, are
2877 considered to be able to check types of their arguments. (fixing
2878 a bug report by Nathan J. Froyd)
2879 * fixed a bug in computing method discriminating functions: It is
2880 now possible to define methods specialized on classes which have
2881 forward-referenced superclasses. (thanks to Gerd Moellmann)
2882 * fixed evaluation order in optional entries (reported by Gilbert
2884 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2885 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2886 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2887 required-by-AMOP NAME argument, as well as CLASS and
2888 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2889 * fixed bug 20: DEFMETHOD can define methods using names that are
2890 not the proper names of classes to designate class specializers.
2891 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2892 Unix file descriptors, instead of blowing up. (thanks to Antonio
2894 * Garbage collection refactoring: user-visible change is that a
2895 call to the GC function during WITHOUT-GCING will not do garbage
2896 collection until the end of the WITHOUT-GCING. If you were doing
2897 this you were probably losing anyway.
2898 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2899 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2900 TYPEP the latter but not the former.
2901 * The compiler issues a full WARNING for calls to undefined functions
2902 with names from the CL package.
2903 * MAP-INTO for a vector destination is open coded. (reported by
2904 Brian Downing on c.l.l)
2905 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2906 documentation string.
2907 * fixed some bugs revealed by Paul Dietz' test suite:
2908 ** COPY-ALIST now signals an error if its argument is a dotted
2910 ** Condition slots are now accessed more correctly in the presence
2911 of multiple initargs for a given slot.
2912 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2913 exclude restarts of the same name associated with a different
2915 ** DEFCLASS of forward-referenced classes with another
2916 forward-referenced class in the superclasses list no longer
2918 ** Condition slots are now initialized once each, not multiple
2919 times. (thanks to Gerd Moellmann)
2920 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2922 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2923 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2924 arguments contain duplicated elements.
2925 ** RESTART-CASE understands local macros.
2926 ** RESTART-CASE associates exactly its own restarts with a condition.
2927 ** ENDP in safe mode checks its argument to be of type LIST.
2928 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2929 in question is unbound.
2930 ** Optimized MAKE-INSTANCE functions no longer cause internal
2931 assertion failures in the presence of duplicate initargs.
2932 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2933 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2934 propagated even in the ctor optimized implementation of
2936 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2938 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2939 * a better implementation of SXHASH on (simple) bit vectors,
2940 measured both in execution speed and in distribution of results
2941 over the positive fixnums, has been installed. Likewise, a better
2942 implementation of EQUAL for simple bit vectors is now available.
2943 * fixed CEILING optimization for a divisor of form 2^k.
2944 * fixed bug 240 (emitting extra style warnings "using the lexical
2945 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2946 by Antonio Martinez)
2947 * fixed SXHASH, giving different results for NIL depending on type
2948 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2949 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2950 variable. (found by Rolf Wester)
2951 * SBCL does not ignore type declarations for special
2952 variables. (reported by rif on c.l.l 2003-03-05)
2953 * some bug fixes in contrib/sb-aclrepl/
2954 * fixed some bugs revealed by Paul Dietz' test suite:
2955 ** a bug in the CONS type specifier, whereby the CAR and CDR
2956 types got intertwined, has been fixed;
2957 ** the type system is now able to reason about the interaction
2958 between INTEGER and RATIO types more completely;
2959 ** APPEND, [N]REVERSE and NRECONC check that those their
2960 arguments, which must be proper lists, are really so;
2961 ** An array specialized to be unable to hold elements has been
2962 implemented, as required -- yes, really -- by ANSI;
2963 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2964 on malformed property lists;
2966 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2967 * incompatible packaging change: in line with Unix convention,
2968 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2969 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2970 * REQUIRE and PROVIDE are now optionally capable of doing something
2971 useful. See the documentation string for REQUIRE.
2972 * infrastructure for a managed SBCL contrib system: contributed
2973 modules in this release include:
2974 ** the ASDF system definition facility;
2975 ** an interface to the BSD Sockets API;
2976 ** an ACL-like convenience interface to the repl;
2977 (thanks to Kevin Rosenberg)
2978 ** an implementation of ROTATE-BYTE, with efficient implementation
2980 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2981 gives a closer approximation to the right answer than previously.
2982 (thanks to Raymond Toy)
2983 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2984 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2985 optional environment arguments, as required by ANSI.
2986 * fixed bugs in other functions taking environment objects, allowing
2987 calls with an explicit NIL environment argument to be compiled
2989 * fixed bug 228: primary return values from
2990 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2991 COMPILE or FUNCTION.
2992 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2993 :TYPE will no longer signal errors on innocuous objects.
2994 * fixed bug 231b: SETQ is better at respecting type declarations in
2995 the lexical environment.
2996 * fixed a bug in DEFCLASS: classes named by symbols with no or
2997 unprintable packages can now be defined.
2998 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2999 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3000 * fixed compiler failure related to checking types of functions.
3001 (reported by Robert E. Brown)
3002 * the compiler is now much more consistent in its error-checking
3003 treatment of bounding index arguments to sequence functions: in
3004 (SAFETY 3) code, errors will be signalled in almost all cases if
3005 invalid sequence bounding indices are passed to functions defined
3006 by ANSI to operate on sequences.
3007 * fixed a bug in the build procedure: documentation of SBCL-specific
3008 packages is now preserved and available in the final Lisp image.
3009 * lifted FDEFINITION lookup out of loops in the implementation of
3010 many list operations. (thanks to Robert E. Brown)
3011 * fixed a bug in the reader: the #n# reader macro now works for
3012 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3013 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3014 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3015 E. Brown and Tony Martinez respectively)
3016 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3017 SLOT-BOUNDP: in particular, it is now able to optimize them much
3018 better, and is now not vulnerable to having packages renamed.
3019 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3020 always called when a slot is not present in an instance. (thanks
3022 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3023 longer signal an error on unreading a character following EOF on
3024 the previous constituent stream. (thanks to Tony Martinez)
3025 * fixed some bugs revealed by Paul Dietz' test suite:
3026 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3027 not just nonnegative fixnums;
3028 ** the logical bit-array operators such as BIT-AND now accept an
3029 explicit NIL for their "opt-arg" argument (to indicate a
3030 freshly-consed result bit-array);
3031 ** ELT now signals an error on an invalid sequence index in safe
3033 ** the type system is now cleverer about negations of numeric
3034 types, and consequently understands the BIGNUM and RATIO types
3036 ** the type system is now cleverer about the interaction between
3037 INTEGER and RATIO types: while bugs still remain, many more
3038 cases are accurately computed;
3039 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3040 if it is in the last clause;
3041 ** CONSTANTLY now correctly returns a side-effect-free function in
3043 ** DECLARE is no longer treated as a special-operator; in
3044 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3045 * incremented fasl file version number due to the change in the
3046 DEFSTRUCT-SLOT-DESCRIPTION structure.
3048 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3049 * minor incompatible change: code processed by the "interpreter" or
3050 EVAL now has a compilation optimization policy of (DEBUG 2)
3051 (changed from (DEBUG 1)) to improve debuggability of interactive
3052 development, and to allow the use of the debug RETURN command in
3054 * an experimental implementation of the RETURN command for the
3055 debugger has been included. (thanks to Frederik Kuivinen)
3056 * fixed bug 62: constraints were not propagated into a loop.
3057 * fixed bug in embedded calls of SORT (reported and investigated by
3059 * fixed some bugs revealed by Paul Dietz' test suite:
3060 ** printing and reading of arrays with some dimensions having
3061 length 0 (thanks to Gerd Moellmann);
3062 ** BOA constructor with &AUX argument without a default value does
3063 not cause a type error;
3064 ** CONSTANTP now returns true for all self-evaluating objects.
3066 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3067 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3068 accessors that are related by inheritance, as specified in the
3069 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3070 Valtteri Vuorikoski)
3071 * The compiler is now able to inline functions that were defined in
3072 a complex lexical environment (e.g. inside a MACROLET).
3073 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3075 * Support for the upcoming FreeBSD-5.0 release has been included.
3076 (thanks to Dag-Erling Smorgrav)
3077 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3078 effect when it is not in a toplevel context.
3079 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3080 stand a better chance of being correct. (thanks to Gerd
3082 * fixed bug in COERCE, which now signals an error on coercing a
3083 rational to a bounded real type which excludes the expected
3085 * The compiler is now able to derive types more accurately from the
3086 COERCE and COMPILE functions.
3087 * fixed bug 223: functional binding is considered to be constant
3088 only for symbols in the CL package.
3089 * fixed bug 231: SETQ did not check the type of a variable being set
3090 (reported by Robert E. Brown)
3091 * A new optimization for MAKE-INSTANCE has been included, fixing
3092 various bugs (including relating to :ALLOCATION :CLASS slots and
3093 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3095 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3096 ** As required by ANSI, LOOP now disallows anonymous collection
3097 clauses such as COLLECT I in conjunction with aggregate boolean
3098 clauses such as THEREIS (= I 1);
3099 ** LOOP now signals an error when any variable is reused in the
3100 same loop (including the potentially useful construct analogous
3101 to WITH A = 1 WITH A = (1+ A);
3102 ** IT is only a special loop symbol within the first clause of a
3103 conditional loop clause;
3104 ** LOOP with a typed iteration variable over a hashtable now
3105 signals a type error iff it should.
3106 * fixed some other bugs revealed by Paul Dietz' test suite:
3107 ** FILE-STREAM now names the class previously known as FD-STREAM;
3108 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3109 argument) no longer signals an error;
3110 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3111 of the slot symbol, rather than using the current package
3112 ((:CONC-NAME "") continues to intern the slot's name in the
3114 * incremented fasl file version number, because of the incompatible
3115 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3116 of the new implementation of DEFINE-COMPILER-MACRO.
3118 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3119 * Support for building SBCL for MIPS platforms running in
3120 little-endian mode has now been checked in, and basic
3121 functionality on said platforms verified.
3122 * minor incompatible change: PCL now records the pathname of a file
3123 in which methods and the like are defined, rather than its
3125 * minor incompatible change: TRUENAME now considers the truename of
3126 a file naming a directory to be the pathname with :DIRECTORY
3127 component indicating that directory.
3128 * minor incompatible change: a NAMED clause in the extended form of
3129 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3130 reason for the previous behaviour is unclear.
3131 * more systematization and improvement of CLOS and MOP conformance
3132 in PCL (thanks to Gerd Moellman and Pierre Mai):
3133 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3135 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3136 primary methods with no specializers;
3137 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3139 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3140 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3141 FORWARD-REFERENCED-CLASSes; error reporting on
3142 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3144 ** SXHASH on CLOS instances now uses a slot internal to the
3145 instance to return different numbers on distinct instances,
3146 while preserving the same return value through invocations of
3148 ** DEFMETHOD signals errors when methods with longer incongruent
3149 lambda lists are added to generic functions;
3150 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3151 CLASS, as specified in AMOP;
3152 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3153 on the order returned by the primary method for classes of
3154 class STANDARD-CLASS;
3155 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3156 * fixed some bugs shown by Paul Dietz' test suite:
3157 ** DOLIST puts its body in TAGBODY;
3158 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3160 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3161 value producing form;
3162 ** if more variables are given to PROGV than values, extra
3163 variables are bound and made to have no value;
3164 ** NSUBSTITUTE on list arguments gets the right answer with
3166 ** ELT signals an error of type TYPE-ERROR when the index argument
3167 is not a valid sequence index;
3168 ** LOOP signals (at macroexpansion time) an error of type
3169 PROGRAM-ERROR when duplicate variable names are found;
3170 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3171 ** FUNCALL of special-operators now cause an error of type
3173 ** PSETQ now works as required in the presence of side-effecting
3174 symbol-macro places;
3175 ** NCONC accepts any object as its last argument;
3176 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3178 ** loop-for-as-package does not require a package to be explicitely
3180 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3182 * fixed bug 166: compiler preserves "there is a way to go"
3183 invariant when deleting code.
3184 * fixed bug 172: macro lambda lists with required arguments after
3185 &REST arguments now cause an error to be signalled. (thanks to
3187 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3188 bug. (thanks to Matthew Danish)
3189 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3191 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3192 arguments are lexically rebound. (thanks to Gerd Moellmann and
3194 * fixed bug 194: error messages are now more informative when there
3195 is no primary method applicable in a call to a generic
3196 function. (thanks to Gerd Moellmann)
3197 * fixed bug in command line argument checking (thanks to Julian
3199 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3200 (thanks to Matthew Danish)
3201 * incremented fasl file version number, because of the
3202 SXHASH-related changes in the layout of CLOS data structures
3204 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3205 * minor incompatible change: The runtime (the Unix executable named
3206 "sbcl") is now much pickier about the .core files it will load.
3207 Essentially it now requires .core files to descend from the same
3208 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3209 as the runtime does. (The intent is to prevent the crashes which
3210 can occur, and which can even be reported as mysterious failures,
3211 when people patch the sources or change the build parameters
3212 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3213 sbcl and .core files.)
3214 * fixed bug: VALUES-LIST is no longer optimized away.
3215 * fixed bug 142: The FFI conversion of C string values to Lisp
3216 string values no longer conses excessively. (thanks to Nathan
3217 Froyd porting Raymond Toy's fix to CMU CL)
3218 * began to systematize and improve MOP conformance in PCL (thanks to
3219 Nathan Froyd, Gerd Moellman and Pierre Mai):
3220 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3222 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3223 ** FINALIZE-INHERITANCE is now called on class finalization;
3224 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3225 argument precedence order.
3226 * fixed bug 202: The compiler no longer fails on functions whose
3227 derived types contradict their declared type.
3228 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3229 so it can be non-toplevel.
3230 * The fasl file version number has changed (because of the new
3231 implementation of DEFMACRO).
3232 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3233 safe code, that any length requirement by their type-specifier
3234 argument is valid. The exceptions to this are described in bug
3236 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3237 functions now check the types of their inputs as required by ANSI.
3238 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3239 introduced symbol is DECLAREd to be SPECIAL.
3240 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3241 * fixed bug: :COUNT argument to sequence functions may be negative
3242 * fixed bug: body of DO-SYMBOLS may contain declarations
3243 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3244 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3245 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3246 symbol macro only once
3247 * fixed printing of call frame when argument list is unavailable
3248 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3249 * compiler no longer signals WARNING on unknown keyword
3252 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3253 * A beta-quality port to the mips architecture running Linux,
3254 based on the old CMUCL backend, has been made. It has been tested
3255 on a big-endian kernel, and works sufficiently well to be able to
3256 rebuild itself; it has not been tested in little-endian mode.
3257 * fixed an inconsistency between gencgc.c and purify.c which made
3258 dumping/loading .core files unreliable
3259 * fixed bug 120a: The compiler now deals correctly with IFs where
3260 the consequent is the same as the alternative, instead of
3261 misderiving the return type. (thanks to Alexey Dejneka)
3262 * fixed bug 113: Logical pathnames are now dumpable (the logical
3263 host is resolved at load-time, throwing an error if it is not
3265 * fixed bug 174: FORMAT's error message is slightly clearer when a
3266 non-printing character is used in a format directive.
3267 * fixed several bugs in compiler checking of type declarations, i.e.
3268 violations of the Python "declarations are assertions" principle
3269 (thanks to Alexey Dejneka)
3270 * fixed several bugs in PCL's error checking (thanks to Gerd
3272 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3274 * fixed bug in compilation of functions as first class values
3275 (thanks to Antonio Martinez-Shotton)
3276 * The compiler's handling TYPE-ERRORs which it can prove will
3277 inevitably happen at runtime has been cleaned up and corrected
3278 in several ways. (thanks to Alexey Dejneka)
3279 * improved argument type checking for various basic arithmetic
3280 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3281 lavished on them in the past that they can be compiled in many
3282 ways in different special cases
3283 * fixed bug 181: compiler checks validity of user supplied type
3285 * cleaned up code flushing in optimization: Function calls which
3286 should signal errors for safety purposes (e.g. which ANSI says
3287 should signal errors when their arguments are of incorrect type)
3288 are no longer optimized away.
3289 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3290 * incremented fasl file version number, because changes in the
3291 implementation of sequence functions like COERCE caused
3292 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3293 in old inline expansions) to become undefined. (Actually these
3294 changes were later undone, so we might very well be binary
3295 compatible with 0.7.7 after all, but leaving the version number
3296 incremented seemed like the simplest and most conservative
3299 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3300 * An alpha-quality port to the parisc architecture running Linux,
3301 based on the old CMUCL backend, has been made. This, even more so
3302 than the other backends, should be considered still a work in
3303 progress; known problems include that the Linux kernel in 64-bit
3304 mode does not propagate the correct sigcontext structure to
3305 userspace, and consequently SBCL on a parisc64 kernel will not
3307 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3308 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3309 suppressed.) Also now that the compiler is looking at declarations
3310 in the environment, it checks optimization declarations as well,
3311 and suppresses inlining when (> DEBUG SPEED).
3312 * More fixes have been made to treatment of floating point exception
3313 treatment and other Unix signals. In particular, floating point
3314 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3315 * The detection and handling of control stack exhaustion (infinite
3316 or very deeply nested recursion) has changed. Stack exhaustion
3317 detection is now done by write-protecting pages at the OS level
3318 and applies at all optimization settings; when found, a
3319 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3320 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3321 be caught using IGNORE-ERRORS.
3322 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3323 correctly when transforming recursive LABELS functions to LETs.
3324 (thanks to Alexey Dejneka)
3325 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3326 that are names of constants or global variables.
3327 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3328 alien routines with docstrings.
3329 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3330 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3332 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3333 (thanks to Eric Marsden)
3334 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3335 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3336 * Bug fix: The SPARC backend can now compile functions involving
3337 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3338 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3339 to a functions expecting a PATHNAME-DESIGNATOR.
3340 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3341 lambda lists. (thanks to Alexey Dejneka)
3342 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3343 functions so that EXPORTing the name of the class doesn't cause
3344 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3345 to Antonio Martinez for reporting this)
3346 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3347 constant-folded without failing an assertion. (thanks to Einar
3348 Floystad Dorum for reporting this)
3349 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3350 types are now dealt with more correctly by the compiler.
3351 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3352 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3353 behaviour analogous to RENAME-FILE. This puts its behaviour more
3354 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3356 * The fasl file version number has changed again. (because of the
3357 bug fix involving the names of PCL MAKE-INSTANCE functions)
3359 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3360 * bug fix: Floating point exceptions are treated much more
3361 consistently on the x86/Linux and PPC/Linux platforms.
3362 * Array initialization with :INITIAL-ELEMENT is now much faster for
3363 cases when the compiler cannot open code the array creation, but
3364 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3365 array accesses have also seen a speed increase.
3366 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3367 specified. (This was at the root of some bad interactions between
3368 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3370 * bug fix: Internal error arguments for undefined functions are now
3371 computed correctly on the PPC/Linux platform.
3372 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3373 Raymond Toy's patch for CMU CL)
3374 * Support for the Solaris 9 operating environment has been included
3375 (thanks to Daniel Merritt)
3376 * A very ugly but hopefully complete draft of the missing FFI chapter
3377 of the manual has been created by reformatting the corresponding
3378 CMU CL manual chapter into (currently very ugly and incoherent)
3379 DocBook and bringing it up to date for SBCL behavior. Thus, the
3380 manual is now essentially complete, at least by my extreme
3381 once-and-only-once standards, whereby it's acceptable to refer to
3382 the doc strings of SB-EXT functions as the primary documentation.
3383 * The fasl file version number has changed again, due to cleanup of
3384 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3385 is no longer a static symbol.)
3387 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3388 * SBCL now builds with OpenMCL (version 0.12) as the
3389 cross-compilation host; also, more progress has been made toward
3390 bootstrapping under CLISP.
3391 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3393 * bug 158 fixed: The compiler can now deal with integer loop
3394 increments different from 1; fixing this turned out also to fix
3396 * bug 169 fixed: no more bogus warnings about using lexical bindings
3397 despite the presence of perfectly good SPECIAL declarations (thanks
3398 to David Lichteblau)
3399 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3400 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3401 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3403 * bug 180 fixed: Method combination specifications no longer ignore
3404 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3405 * bug fix: Structure type predicate functions now check their argument
3406 count as they should.
3407 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3408 correctly. (thanks to Pierre Mai)
3409 * minor incompatible change: The --noprogrammer option is deprecated
3410 in favor of the new --disable-debugger option, which is very similar.
3411 (The major difference is that it takes effect at a slightly different
3412 time at startup, causing handling of errors in --sysinit and
3413 --userinit files will be affected differently.) The
3414 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3415 been added to allow this functionality to be controlled from ordinary
3416 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3417 maintainers, who might want to run non-interactive scripts to
3418 build SBCL cores which will later be used interactively.)
3419 * minor incompatible change: The LOAD function no longer, when given
3420 a wild pathname to load, loads all files matching that pathname.
3421 Instead, an error of type FILE-ERROR is signalled.
3423 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3424 * bug 147 fixed: The compiler preserves its block link/count
3425 invariants more correctly now so that it doesn't crash. (thanks
3427 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3429 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3430 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3431 no-value-for-this-character as NIL (as specified by ANSI).
3432 * HOST-NAMESTRING on physical pathnames now returns a string that is
3433 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3434 (thanks to Christophe Rhodes)
3435 * The Alpha port handles icache flushing more correctly. (thanks to
3437 * More progress has been made toward bootstrapping under CLISP. (thanks
3438 to Christophe Rhodes)
3439 * The fasl file format has changed again, because dynamic loading
3440 on OpenBSD (which has non-ELF object files) motivated some cleanups
3441 in the way that foreign symbols are transformed and passed around.
3442 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3443 is no longer treated as whitespace by the reader, but instead as
3444 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3445 |A
\7fB|, instead of A as it used to.
3447 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3448 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3449 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3450 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3451 this as of 0.7.1.45, but was left out of the previous news section
3452 (thanks to Dan Barlow)
3453 * SBCL now runs on the Solaris operating system on SPARC architectures
3454 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3455 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3456 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3457 * SPARC backend cleanups, allowing builds of cores optimized for V8
3458 and V9 SPARCS, and also emission of code targeted to a particular
3459 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3461 * SBCL is closer to bootstrapping under CLISP, thanks to various
3462 fixes by Christophe Rhodes.
3463 * The fasl file format has changed again, to allow the compiler's
3464 INFO database to support symbol macros.
3465 * The user manual (in doc/) is formatted into HTML more nicely.
3466 (thanks to coreythomas)
3467 * The system is smarter about SUBTYPEP relationships, especially
3468 those involving NOT types (including types such as ATOM which are
3469 represented internally using NOT types). Thus SUBTYPEP is less
3470 likely to return (VALUES NIL NIL) in general, and in particular
3471 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3473 * The fasl file format has changed again, because the internal
3474 representation of types now includes a new slot to support the new
3475 SUBTYPEP-of-NOT-types logic.
3476 * (not a change in the main branch of SBCL, but a related prototype
3477 which can hopefully be merged into the main branch of SBCL in the
3478 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3479 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3480 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3481 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3482 dispatch macro character combinations. (thanks to Alexey Dejneka)
3483 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3484 that superficially look like logical namestrings correctly.
3485 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3487 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3488 default supplied-p) for &optional and &key arguments. (thanks to
3491 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3492 * incompatible change: The compiler is now less aggressive about
3493 tail call optimization, doing it only when (> SPACE DEBUG) or
3494 (> SPEED DEBUG). (This is an incompatible change because there are
3495 programs which relied on the old CMU-CL-style behavior to optimize
3496 away their unbounded recursion which will now die of stack overflow.)
3497 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3498 for non-GENCGC systems has been increased to 20M (since that
3499 seems much closer to the likely performance optimum for modern
3500 systems than the old 4M value was)
3501 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3502 and *DEBUG-PRINT-LEVEL*
3503 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3504 of CMU CL's support for SPARC, and various endianness and other
3505 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3506 * new syntactic sugar for the Unix command line: --load foo.bar is now
3507 an alternate notation for --eval '(load "foo.bar")'.
3509 ** The system now detects stack overflow and handles it gracefully,
3510 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3511 optimization settings. (This is a good thing in general, and
3512 its introduction in this version should be particularly timely
3513 for anyone whose code fails because of suppression of tail
3515 ** The system now hunts for the C variable "environ" in a more
3516 devious way, to avoid segfaults when the C library version differs
3517 between compile time and run time. (thanks to Christophe Rhodes)
3518 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3519 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3520 ** The compiler no longer issues bogus style warnings for undefined
3521 classes in the same source file as the DEFCLASSes which defined
3522 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3524 ** fixes in CONDITION class precedence list for undefined function
3525 errors (thanks to Alexei Dejneka)
3526 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3527 correctly. (thanks to Dan Barlow)
3528 ** portability fixes aiming at bootstrapping under CLISP (thanks
3529 to Dave McDonald and Christophe Rhodes)
3530 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3531 ** fixes in type translation and and type inference (thanks to
3533 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3534 ** various fixes in the new ports (thanks to Dan Barlow)
3535 * several changes related to debugging:
3536 ** suppression of tail recursion, as noted above
3537 ** stack overflow detection, as noted above
3538 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3539 is now the default. (For some time encapsulation has been more
3540 reliable than the breakpoint-based :ENCAPSULATE NIL
3541 implementation, at least on X86 systems; and I just noticed that
3542 encapsulation also seems closer to the spirit of the ANSI
3545 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3547 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3548 up properly again. (There was a packaging bug in 0.7.0 which
3549 left their definitions in SB-SYS::LOAD-FOREIGN and
3550 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3551 vital for most things which interface to C-level interfaces,
3552 like extensions working with sockets or databases or
3553 Perl-compatible regexes or whatever, and the need to fix
3554 this bug is the main reason that 0.7.1 was released so
3556 ** DEFGENERIC is now choosier about the methods it redefines, so that
3557 reLOADing a previously-LOADed file containing DEFGENERICs does
3558 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3559 cycle now works as it should. (thanks to Alexey Dejneka)
3560 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3561 fixed by Christophe Rhodes. (He actually submitted this patch
3562 months ago, and I delayed until after 0.7.0.)
3563 ** Bug 111 (internal compiler confusion about runtime checks on
3564 FUNCTION types) was fixed by Alexey Dejneka.
3565 * Some internal cleanups (getting rid of variables which aren't
3566 needed now that the byte interpreter is gone) caused the fasl
3567 file format number to change again.
3569 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3570 * major incompatible change: The default fasl file extension, i.e. the
3571 default extension for files produced by COMPILE-FILE, has changed
3572 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3574 ** There are many changes in the implementation of the compiler.
3575 SBCL is now essentially a compiler-only implementation of ANSI
3576 Common Lisp. EVAL still "interprets" a few special cases, but
3577 almost all the interesting cases are handled by creating
3578 a LAMBDA expression, calling COMPILE on it, then calling
3579 FUNCALL on the result.
3580 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3581 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3582 Since the code is newer, there might still be some new bugs
3583 (though not as many as before Martin Atzmueller's fixes:-). But
3584 the new code is substantially simpler and clearer, and hopefully
3585 any remaining bugs will be simpler, less fundamental, and more
3586 fixable then the bugs in the old code.
3587 ** The revised compiler is still a little unsteady on its feet.
3589 *** The debugging information it produces (particularly the names
3590 of FUNCTION objects) is sometimes much less useful than what
3591 the old compiler produced.
3592 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3593 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3594 within a MACROLET) has been temporarily weakened.
3595 ** There are new compiler optimizations for various functions:
3596 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3597 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3598 *** the math functions TRUNCATE, FLOOR, and CEILING
3599 *** the function-of-all-trades COERCE
3600 Mostly these should be transparent, but there's one
3601 potentially-annoying problem (bug 117): when the compiler
3602 inline-expands a function and does type analysis on the result,
3603 it can create control paths which have type mismatches, and
3604 when it can't prove that those control paths aren't taken,
3605 it will issue WARNINGs about the type mismatches. This is
3606 a particular problem in practice for the new sequence functions.
3607 It's not clear how this should be fixed, and for now, a
3608 workaround is given in the entry for 117 in the BUGS file.
3609 ** (Because of the interaction between the two previous items --
3610 occasional inlining problems and new inline expansions -- some
3611 of the new sequence function optimizations won't really kick in
3612 completely until debugging information, and then inlining, are
3613 straightened out in some future version.)
3614 * minor incompatible changes:
3615 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3616 DIRECTORY no longer implicitly promotes NIL slots of its
3617 pathname argument to :WILD. In particular, when you ask for the
3618 contents of a directory (which you used to be able to do without
3619 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3620 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3621 ** changes in behavior that ANSI explicitly defines to be
3622 implementation dependent:
3623 *** The new compiler-only implementation still conforms with ANSI,
3624 but acts a little different than before. Besides the obvious
3625 changes in performance tradeoffs (that the cost per form passed
3626 to EVAL has gone up, and the cost per form executed by EVAL
3627 has gone down), the behavior of the system changes a little
3628 because there are no longer any interpreted function objects.
3629 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3630 e.g. doing COMPILE on the output of interactive DEFUN is
3632 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3634 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3635 has changed from (:CL) to NIL.
3636 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3637 e.g. #\Newline and #\Tab, don't have names specified in the
3638 ANSI Common Lisp standard, is now based on their ASCII symbolic
3639 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3640 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3641 are no longer used for output.
3642 ** changes in internal implementation constants:
3643 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3644 4 million. (If your application spends a lot of time GCing and
3645 you have a lot of RAM, you might want to experiment with
3646 increasing it even more.)
3647 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3648 However, almost all old code should still continue to work without
3649 immediate update, as SB-C-CALL is now a (deprecated) nickname
3651 ** Old operator names in the style DEF-FOO are now deprecated in
3652 favor of new corresponding names DEFINE-FOO, for consistency with
3653 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3654 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3655 internal symbols, but a few supported extensions like
3656 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3657 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3658 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3659 etc. as you get deeper into recursive calls to the debugger
3660 command loop, instead of the old "5]", "5]]", "5]]]"
3661 sequence. (I was motivated to do this when squabbles between
3662 ILISP and SBCL left me very deeply nested in the debugger. In the
3663 short term, this change will probably provoke more ILISP/SBCL
3664 squabbles, but hopefully it will be an improvement in the long run.)
3665 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3666 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3667 function, since declaiming return types involving aliens is
3668 (1) annoyingly messy to do by hand and (2) vital to efficient
3669 compilation of code which calls such functions.
3670 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3671 longer reexported by the SB-EXT package. They're solely useful
3672 for alien code, so it seems more logical that you should get
3673 them from the SB-ALIEN package, not in SB-EXT.
3674 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3675 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3676 features. Instead, the code that they used to control is always
3677 built into the system.
3678 * many other bug fixes
3679 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3680 advantage of the new EVAL-WHEN stuff and to clean them up in
3681 general, and they are now more ANSI-compliant in a number of
3682 ways. Martin Atzmueller is responsible for a lot of this.
3683 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3685 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3686 *** correct ERROR type for various file operations
3687 *** some fixes for Lisp streams
3688 *** DEFMETHOD syntax checking
3689 *** changing old weird representation of debug information as
3690 strings (which, among their other deficiencies, don't transform
3691 correctly when you rename packages, and don't change their
3692 print representation when you change things like *PACKAGE*
3693 and *PRINT-LENGTH*) to symbols and lists of symbols
3694 He also made several improvements and fixed several bugs in DESCRIBE.
3695 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3697 *** misbehavior of WRITE-STRING/WRITE-LINE
3698 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3699 and several other LOOP problems as well
3700 *** DIRECTORY when similar filenames are present
3701 *** DEFGENERIC with :METHOD options
3702 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3703 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3704 *** argument ordering in FIND with :TEST option
3705 *** mishandled package designator argument in APROPOS-LIST
3706 *** various problems in the backquote readmacro
3707 *** a bug in APROPOS
3708 *** probably some others that I'm not describing very well here,
3709 since the CVS log documents them by reference to sbcl-devel
3710 messages, and the SourceForge archives aren't working well.:-(
3711 ** Dan Barlow improved the Alpha port (and is making progress on the
3712 PPC port, for those of you who think different).
3713 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3714 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3715 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3716 continues to work on the SPARC port (for those of you in a position
3717 to look down upon our little PC-compatible boxes from a great height).
3718 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3719 and uses that copy, rather than the current dynamic value, when
3720 it's trying to decide whether to truncate output. Thus e.g.
3721 (let ((*print-lines* 50))
3722 (pprint-logical-block (stream nil)
3724 (let ((*print-lines* 8))
3725 (print (aref possiblybigthings i) stream)))))
3726 should now truncate the logical block only at 50 lines, instead of
3727 often truncating it at 8 lines, as it did before.
3728 * The doc/cmucl/ directory, containing old CMU CL documentation
3729 from the time of the fork, is no longer part of the base system.
3730 SourceForge has shut down its anonymous FTP service, and with it
3731 my original plan for distributing the old CMU CL documentation
3732 there. For now, if you need these files you can download an old
3733 SBCL source release and extract them from it.
3734 * The fasl file version number changed again, for dozens of reasons,
3735 some of which are apparent above.
3737 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3738 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3739 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3740 patch, so that DISASSEMBLE output is much nicer.
3741 * The code in the SB-PROFILE package now seems reasonably stable.
3742 I still haven't decided what the final interface should look like
3743 (I'd like PROFILE to interact cleanly with TRACE, since both
3744 facilities use function encapsulation) but if you have a need
3745 for profiling now, you can probably use it successfully with
3746 the current CMU-CL-style interface.
3747 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3748 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3749 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3750 link now returns the dangling link itself, and for similar
3751 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3752 link itself. (In these cases the old code signalled an error and
3753 looped endlessly, respectively.) Thus, DIRECTORY now works even
3754 in the presence of dangling and cyclic symbolic links.
3755 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3756 is now a supported extension again, since the consensus on
3757 sbcl-devel was that it can be useful for ordinary development
3758 work, not just for debugging SBCL itself.
3759 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3760 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3761 that currently-defined functions might be redefined later with
3762 different return types.
3763 * Hash tables can be printed readably, as inspired by CMU CL code
3764 of Eric Marsden and SBCL code of Martin Atzmueller.
3765 * better error handling in CLOS method combination, thanks to
3766 Martin Atzmueller porting Pierre Mai's CMU CL patches
3767 * more overflow fixes for >16Mbyte I/O buffers
3768 * A bug in READ has been fixed, so that now a single Ctrl-D
3769 character suffices to cause end-of-file on character streams.
3770 In particular, now you only need one Ctrl-D at the command
3771 line (not two) to exit SBCL.
3772 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3774 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3775 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3776 does the right thing.
3777 * The implementation of some type tests, especially for CONDITION
3778 types, is now tidier and maybe faster, due to CMU CL code
3779 originally by Douglas Crosher, ported by Martin Atzmueller.
3780 * Some math functions have been fixed, and there are new
3781 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3782 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3783 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3784 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3785 likely to be built into sbcl-0.7.0.)
3786 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3787 CL, ported to SBCL by Martin Atzmueller)
3788 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3789 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3790 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3791 * There's a new slam.sh hack to shorten the edit/compile/debug
3792 cycle for low-level changes to SBCL itself, and a new
3793 :SB-AFTER-XC-CORE target feature to control the generation of
3794 the after-xc.core file needed by slam.sh.
3795 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3796 COMPILE-FILE is no longer supported, so that now every function
3797 gets an entry point, so that block compilation looks a little
3798 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3799 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3800 returns the number of bytes consed since the system started,
3801 rather than the number consed since the first time the function
3802 was called. (The new definition parallels ANSI functions like
3803 CL:GET-INTERNAL-RUN-TIME.)
3804 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3805 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3806 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3807 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3808 * Fasl file version numbers are now independent of the target CPU,
3809 since historically most system changes which required version
3810 number changes have affected all CPUs equally. Similarly,
3811 the byte fasl file version is now equal to the ordinary
3814 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3815 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3816 is no longer recognized. I apologize for this, because it was
3817 listed in SB-EXT as a supported extension, but I found that
3818 its existing behavior was poorly specified, as well as incorrectly
3819 specified, and it looked like too much of a mess to straighten it
3820 out. I have enough on my hands trying to get ANSI stuff to work..
3821 * many patches ported from CMU CL by Martin Atzmueller, with
3822 half a dozen bug fixes in pretty-printing and the debugger, and
3823 half a dozen others elsewhere
3824 * fixed bug 13: Floating point infinities are now supported again.
3825 They might still be a little bit flaky, but thanks to bug reports
3826 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3827 as flaky as they were.
3828 * The --noprogrammer command line option is now supported. (Its
3829 behavior is slightly different in detail from what the old man
3830 page claimed it would do, but it's still appropriate under the
3831 same circumstances that the man page talks about.)
3832 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3833 are now supported, and enabled by default. Thus, the compiler can
3834 handle many floating point and complex operations much less
3835 inefficiently. (Thus e.g. you can implement a complex FFT
3837 * The compiler now detects type mismatches between DECLAIM FTYPE
3838 and DEFUN better, and implements CHECK-TYPE more correctly, and
3839 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3840 thanks to patches from Martin Atzmueller.
3841 * various fixes to make the cross-compiler more portable to
3842 ANSI-conforming-but-different cross-compilation hosts (notably
3843 Lispworks for Windows, following bug reports from Arthur Lemmens)
3844 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3845 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3846 fixed thanks to Pierre Mai's CMU CL patches.
3847 * fixes to keep the system from overflowing internal counters when
3848 it tries to use i/o buffers larger than 16M bytes
3849 * fixed bug 45a: Various internal functions required to support
3850 complex special functions have been merged from CMU CL sources.
3851 (When I was first setting up SBCL, I misunderstood a compile-time
3852 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3853 * improved support for type intersection and union, fixing bug 12
3854 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3855 more obscure bugs as well
3856 * some steps toward byte-compiling non-performance-critical
3857 parts of the system, courtesy of patches from Martin Atzmueller
3858 * Christophe Rhodes has made some debian packages of sbcl at
3859 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3860 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3861 stable, but are nonetheless usable. When he's ready, I'd be happy
3862 to add them to the SourceForge "File Releases" section. (And if
3863 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3864 * new fasl file format version number (because of changes in
3865 internal representation of (OR ..) types to accommodate the new
3866 support for (AND ..) types, among other things)
3868 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3869 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3871 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3873 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3874 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3875 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3876 * DESCRIBE now works on structure objects again.
3877 * Most function call argument type mismatches are now handled as
3878 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3879 know whether the function will be redefined before the call is
3880 executed. (The compiler could flag local calls with full WARNINGs,
3881 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3882 it doesn't keep track of enough information to know whether calls
3883 are local in this sense.)
3884 * Compiler output is now more verbose, with messages truncated
3885 later than before. (There should be some supported way for users
3886 to override the default verbosity, but I haven't decided how to
3887 provide it yet, so this behavior is still controlled by the internal
3888 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3889 src/compiler/ir1util.lisp.)
3890 * Fasl file format version numbers have increased again, because
3891 support for the Gray streams extension changes the layout of the
3892 system's STREAM objects.
3893 * The Gray subclassable streams extension now works, thanks to a
3894 patch from Martin Atzmueller.
3895 * The full LOAD-FOREIGN extension (not just the primitive
3896 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3897 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3898 but like most other programs, it defaults to copying the Unix
3899 environment from the original process instead of starting the
3900 new process in an empty environment.
3901 * Extensions which manipulate the Unix environment now support
3902 an :ENVIRONMENT keyword option which doesn't smash case or
3903 do other bad things. The CMU-CL-style :ENV option is retained
3904 for porting convenience.
3905 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3906 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3908 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3910 * A patch from Martin Atzmueller seems to have solved the SIGINT
3911 problem, and as far as we know, signal-handling now works cleanly.
3912 (If you find any new bugs, please report them!)
3913 * The system no longer defaults Lisp source file names to types
3914 ".l", ".cl", or ".lsp", but only to ".lisp".
3915 * The compiler no longer uses special default file extensions for
3916 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3917 seems to expect a single default extension for all compiled code,
3918 and there's no compelling reason to try to stretch the standard
3919 to allow two different extensions.) Instead, byte-compiled files
3920 default to the same extension as native-compiled files.
3921 * Fasl file format version numbers have increased again, because
3922 a rearrangement of internal implementation packages made some
3923 dumped symbols in old fasl files unreadable in new cores.
3924 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3925 many fewer weird special cases.
3926 * Bug #17 (differing COMPILE-FILE behavior between logical and
3927 physical pathnames) has been fixed, and some related misbehavior too,
3928 thanks to a patch from Martin Atzmueller.
3929 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3930 by Tim Moore, ported to SBCL by Martin Atzmueller.
3931 * Martin Atzmueller fixed several filesystem-related problems,
3932 including bug #36, in part by porting CMU CL patches, which were
3933 written in part by Paul Werkowski.
3934 * More compiler warnings in src/runtime/ are gone, thanks to
3935 more patches from Martin Atzmueller.
3936 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3939 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3941 * DESCRIBE now works on CONDITION objects.
3942 * The debugger now handles errors which arise when trying to print
3943 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3945 * The build system now uses an additional file, customize-target-features.lisp,
3946 to allow local modifications to the target *FEATURES* list. (The point of
3947 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3948 debugging features enabled, without having to worry about propagating it
3949 into everyone's system when I do a "cvs update".) When no
3950 customize-target-features.lisp file exists, the target *FEATURES* list
3951 should be constructed the same way as before.
3952 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3953 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3954 warn about behavior which is undefined under the ANSI spec. Note: This
3955 is specified by ANSI, but it's not very popular with programmers.
3956 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3957 macro in the SBCL sources for an example of a workaround which you
3958 might use to make portable ANSI-standard code which does what you want.
3959 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3960 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3961 when it's used as a non-toplevel form. (This is required in order
3962 to implement the DEFCONSTANT-EQX macro.)
3963 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3964 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3965 funny things, due to bugs in the implementation of EVAL-WHEN
3966 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3967 non-toplevel DEFCONSTANTs are uncommon.)
3968 * The core file version number and fasl file version number have been
3969 incremented, because the old noncompliant DEFCONSTANT behavior involved
3970 calling functions which no longer exist, and because I also took the
3971 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3972 * fixed bug 1 (error handling before read-eval-print loop starts), and
3973 redid debugger restarts and related debugger commands somewhat while
3975 ** The QUIT debugger command is gone, since it did something
3976 rather different than the SB-EXT:QUIT command, and since it never
3977 worked properly outside the main toplevel read/eval/print loop.
3978 Invoking the new TOPLEVEL restart provides the same functionality.
3979 ** The GO debugger command is also gone, since you can just invoke
3980 the CONTINUE restart directly instead.
3981 ** The TOP debugger command is also gone, since it's redundant with the
3982 FRAME 0 command, and since it interfered with abbreviations for the
3984 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3986 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3987 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3988 some porting work by Martin Atzmueller)
3989 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3990 with a boatload of patches which he ported to SBCL. Now that those
3992 ** The system tries to make sure that its low-priority messages
3993 are prefixed by semicolons, to help people who like to use
3994 syntax highlighting in their ILISP buffer. (This patch
3995 was originally due to Raymond Toy.)
3996 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3997 patches originally written by Raymond Toy.
3998 ** The compiler understands coercion between single-value and
3999 multiple-VALUES type expressions better, getting rid of some very
4000 weird behavior, thanks to patches originally by Robert MacLachlan
4001 and Douglas Crosher.
4002 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4003 lambda lists, thanks to a patch originally by Pierre Mai.
4004 ** The system no longer bogusly warns about "abbreviated type
4006 ** The compiler gets less confused by inlining and RETURN-FROM,
4007 thanks to some patches originally by Tim Moore.
4008 ** The system no longer hangs when dumping circular lists to fasl
4009 files, thanks to a patch originally from Douglas Crosher.
4010 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4011 undefined function error.
4012 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4013 in structure slot initforms) for now, documented workaround instead:-|
4014 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4015 * fixed bug 5: added stubs for various Gray stream functions called
4016 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4017 installed, at least appropriate type errors are generated
4018 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4019 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4020 consistently in DEFMETHOD forms.
4021 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4022 it doesn't seem to affect SBCL after all
4023 * The C runtime system now builds with better optimization and many
4024 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4026 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4028 * The system is now under CVS at SourceForge (instead of the
4029 CVS repository on my home machine).
4030 * The new signal handling code has been tweaked to treat register
4031 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4032 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4033 downstream, e.g. in debug-int.lisp, has implicit dependencies
4034 on the unsignedness of integer representation of machine words,
4035 and that caused the system to bomb out with infinite regress
4036 when trying to recover from type errors involving signed values,
4037 e.g. (BUTLAST '(1 2 3) -1).)
4038 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4039 (This was one of the bugs Peter Van Eynde reported back in July.)
4040 * The system now uses code inspired by Colin Walters' O(N)
4041 implementation of MAP (from the cmucl-imp@cons.org mailing
4042 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4043 inline the MAP operation, and there is more than one
4044 sequence argument to the MAP call (so that it can't just
4045 do ETYPECASE once and for all based on the type of the
4046 single sequence argument). (The old non-inline implementation
4047 of the general M-argument sequence-of-length-N case required
4048 O(M*N*N) time when any of the sequence arguments were LISTs.)
4049 * The QUIT :UNIX-CODE keyword argument has been renamed to
4050 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4051 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4053 * Raymond Wiker's patches to port dynamic loading from Linux to
4054 FreeBSD have been added.
4055 * The BUGS file is now more nearly up to date, thanks in large part
4056 to Martin Atzmueller's review of it.
4057 * The debugger now flushes standard output streams before it begins
4058 its output ("debugger invoked" and so forth).
4059 * The core version number and fasl file version number have both
4060 been incremented, because of incompatible changes in the layout
4062 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4063 used to not be called for a saved Lisp image.)
4064 * Martin Atzmueller's version of a patch to fix a compiler crash,
4065 as posted on sbcl-devel 13 September 2000, has been installed.
4066 * Instead of installing Martin Atzmueller's patch for the
4067 compiler transform for SUBSEQ, I deleted the compiler transform,
4068 and transforms for some similar consing operations.
4069 * A bug in signal handling which kept TRACE from working on OpenBSD
4071 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4072 away by constant folding
4073 * The system now defines its address space constants in one place
4074 (in the Lisp sources), and propagates them automatically elsewhere
4075 (through GENESIS and the sbcl.h file). Therefore, patching the
4076 address map is less unnecessarily tedious and error-prone. The
4077 Lisp names of address space constants have also been systematized.
4078 * CVS tags like dollar-Header-dollar have been removed from
4079 the sources, because they have never saved me trouble and
4080 they've been source of trouble working with patches and other
4081 diff-related operations.
4082 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4083 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4085 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4087 * The system has been ported to OpenBSD.
4088 * The system now compiles with a simple "sh make.sh" on the systems
4089 that it's supported on. I.e., now you no longer need to tweak
4090 text in the target-features.lisp-expr and symlinks in src/runtime/
4091 by hand, the make.sh takes care of it for you.
4092 * The system is no longer so grossly inefficient when compiling code
4093 involving vectors implemented as general (not simple) vectors (VECTOR T),
4094 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4095 which dares to use the various sequence functions on non-simple
4096 vectors, takes less of a performance hit.
4097 * There is now a primitive type predicate VECTOR-T-P
4098 to test for the (VECTOR T) type, so that e.g.
4099 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4100 can now be compiled with some semblance of efficiency. (The old code
4101 turned the type declaration into a full call to %TYPEP at runtime!)
4102 * AREF on (VECTOR T) is still not fast, since it's still compiled
4103 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4104 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4105 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4106 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4107 without making full calls to SUBTYPEP at runtime.
4108 (Some analogous efficiency issues for non-simple vectors specialized to
4109 element types other than T, or for non-simple multidimensional arrays,
4110 have not been addressed. They could almost certainly be handled the
4111 same way if anyone is motivated to do so.)
4112 * The changes in array handling break binary compatibility, so
4113 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4114 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4115 instead of (VALUES T T).
4116 * By following the instructions that Dan Barlow posted to sbcl-devel
4117 on 2 July 2000, I was able to enable primitive dynamic object
4118 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4119 functionality is not implemented (since it calls ld to resolve
4120 library references automatically, requiring RUN-PROGRAM for its
4121 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4122 try to resolve library references) is now supported.
4123 * The system now flushes the standard output streams when it terminates,
4124 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4125 them at several other probably-convenient times, e.g. in each pass of
4126 the toplevel read-eval-print loop, and after evaluating a form given
4127 as an "--eval" command-line option. (These changes were motivated by a
4128 discussion of stream flushing issues on cmucl-imp in August 2000.)
4129 * The source transform for TYPEP of array types no longer assumes
4130 that an array whose element type is a not-yet-defined type
4131 is implemented as an array of T, but instead punts, so that the
4132 type will be interpreted at runtime.
4133 * There is now some support for cross-compiling in make.sh: each of
4134 the phases of make.sh has its own script. (This should be transparent
4135 to people doing ordinary, non-cross-compile builds.)
4136 * Since my laptop doesn't have hundreds of megabytes of memory like
4137 my desktop machine, I became more motivated to do some items on
4138 my to-do list in order to reduce the size of the system a little:
4139 ** Arrange for various needed-only-at-cold-init things to be
4140 uninterned after cold init. To support this, those things have
4141 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4142 symbols with such names are now uninterned after cold init).
4143 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4144 fasl files for cold load.
4145 ** Remove the old compiler structure pooling code (which used to
4146 be conditional on the target feature :SB-ALLOC) completely.
4147 ** Redo the representation of some data in cold init to be more compact.
4148 (I also looked into supporting byte compiled code at bootstrap time,
4149 which would probably reduce the size of the system a lot, but that
4150 looked too complicated, so I punted for now.)
4151 * The maximum signal nesting depth in the src/runtime/ support code has
4152 been reduced from 4096 to 256. (I don't know any reason for the very
4153 large old value. If the new smaller value turns out to break something,
4154 I'll probably just bump it back up.)
4155 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4157 * Many, many bugs reported by Peter Van Eynde have been added to
4158 the BUGS list; some have even been fixed.
4159 * While enabling dynamic object file loading, I tried to make the
4160 code easier to understand, renaming various functions and variables
4161 with less ambiguous names, and changing some function calling
4162 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4163 * While trying to figure out how to do the OpenBSD port, I tried to
4164 clean up some of the code in src/runtime/. In particular, I dropped
4165 support for non-POSIX signal handling, added various comments,
4166 tweaked the code to reduce the number of compilation warnings, and
4167 renamed some files to increase consistency.
4168 * To support the new automatic configuration functionality in make.sh,
4169 the source file target-features.lisp-expr has been replaced with the
4170 source file base-target-features.lisp-expr and the machine-generated
4171 file local-target-features.lisp-expr.
4172 * fixed a stupid quoting error in make.sh so that using CMU CL
4173 "lisp -batch" as cross-compilation host works again
4175 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4177 * DESCRIBE no longer tries to call itself recursively to describe
4178 bound/fbound values, so that it no longer fails on symbols which are
4179 bound to themselves (like keywords, T, and NIL).
4180 * DESCRIBE now works on generic functions.
4181 * The printer now prints less-screwed-up representations of closures
4182 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4183 * A private symbol is used instead of the :EMPTY keyword previously
4184 used to mark empty slots in hash tables. Thus
4185 (DEFVAR *HT* (MAKE-HASH-TABLE))
4186 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4187 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4188 now does what ANSI says that it should. (You can still get
4189 similar noncompliant behavior if bang on the hash table
4190 implementation with all the symbols you get back from
4191 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4192 This breaks binary compatibility, since tests for equality to
4193 :EMPTY are wired into things like the macroexpansion of
4194 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4196 * There's now a minimal placeholder implementation for CL:STEP,
4197 as required by ANSI.
4198 * An obscure bug in the interaction of the normal compiler, the byte
4199 compiler, inlining, and structure predicates has been patched
4200 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4201 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4203 * Missing ordinary arguments in a macro call are now detected even
4204 when the macro lambda list contains &KEY or &REST.
4205 * The debugger no longer complains about encountering the top of the
4206 stack when you type "FRAME 0" to explicitly instruct it to go to
4207 the top of the stack. And it now prints the frame you request even
4208 if it's the current frame (instead of saying "You are here.").
4209 * As specified by ANSI, the system now always prints keywords
4210 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4212 * The default initial SIZE of HASH-TABLEs is now smaller.
4213 * Type information from CLOS class dispatch is now propagated
4214 into DEFMETHOD bodies, so that e.g.
4215 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4217 is now basically equivalent to
4218 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4219 (DECLARE (TYPE SINGLE-FLOAT X))
4221 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4222 operation, without having to do run-time type dispatch.
4223 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4224 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4225 * Since I don't seem to be making big file reorganizations very often
4226 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4227 is overflowing my ability to conveniently back them up), I've finally
4228 checked the system into CVS. (The CVS repository is on my home system,
4229 not at SourceForge -- putting it on SourceForge might come later.)
4230 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4231 high-level GC-NOTIFY-FOO functions send their output. (There's
4232 still very little control of where low-level verbose GC functions
4233 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4234 less than it used to -- the GC-NOTIFY-FOO functions are now under
4235 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4236 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4237 in only one place in the source code, and propagates it automatically
4238 everywhere that it's needed. Thus e.g. when I bump the version from
4239 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4240 * The C source files now include boilerplate legalese and documentation
4241 at the head of each file (just as the Lisp source files already did).
4242 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4243 to his page will be replaced with a link to his new CLiki service.
4245 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4247 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4248 * The build process now looks for GNU make under the default name "gmake",
4249 instead of "make" as it used to. If GNU make is not available as "gmake"
4250 on your system, you can change this default behavior by setting the
4251 GNUMAKE environment variable.
4252 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4253 can build without error under CMU CL.
4255 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4257 * There is now a partial SBCL user manual (with some new text and some
4258 text cribbed from the CMU CL manual).
4259 * The beginnings of a profiler have been added (starting with the
4260 CMU CL profiler and simplifying and cleaning up). Eventually the
4261 main interface should be through the TRACE macro, but for now,
4262 it's still accessed through vaguely CMU-CL-style functions and macros
4263 exported from the package SB-PROFILE.
4264 * Some problems left over from porting CMU CL to the new
4265 cross-compilation bootstrap process have been cleaned up:
4266 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4267 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4268 part when getting rid of the compiler *BACKEND* stuff.)
4269 ** The value of *NULL-TYPE* was screwed up, because it was
4270 being initialized before the type system knew the final
4271 definition of the 'NULL type. This screwed up several key
4272 optimizations in the compiler, causing inefficiency in all sorts
4273 of places. (I found it because I wanted to understand why
4274 GET-INTERNAL-RUN-TIME was consing.)
4275 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4276 PROCLAIM FTYPE information. Unfortunately this broke binary
4277 compatibility again, since now the forms output by DEFGENERIC
4278 to refer to functions which didn't exist in 0.6.3.
4279 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4280 can use the new (as of 0.6.3) transform for SOME into MAP into
4282 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4283 values to (INTEGER 0 1000000), so that the time code will no longer
4284 occasionally get blown up by Linux returning 1000000 microseconds
4285 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4286 its output conform to the ANSI spec. (Alas, this makes its output
4287 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4288 * A full call to MAP NIL with a single sequence argument no longer conses.
4289 * fixes to problems pointed out by Martin Atzmueller:
4290 * The manual page no longer talks about multiprocessing as though
4291 it were currently supported.
4292 * The ILISP support patches have been removed from the distribution,
4293 because as of version 5.10.1, ILISP now supports SBCL without us
4294 having to maintain patches.
4295 * added a modified version of Raymond Toy's recent CMU CL patch for
4296 EQUALP comparison of HASH-TABLE
4298 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4300 * The system still can't cross-compile itself with
4301 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4302 entails), but at least it can compile more of itself that way
4303 than it used to be able to, and various buglets which were uncovered
4304 by trying to cross-compile itself that way have now been fixed.
4305 * This release breaks binary compatibility again. This time
4306 at least I've incremented the FASL file format version to 2, so that the
4307 problem can be detected reliably instead of just causing weird errors.
4308 * various new style warnings:
4309 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4310 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4311 ** using the lexical binding of a variable named in the *FOO* style
4312 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4313 as specified by ANSI.
4314 * *RANDOM-STATE* is no longer automatically initialized from
4315 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4316 default behavior of the system is to repeat its behavior every time
4317 it's run. If you'd like to change this behavior, you can always
4318 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4319 old convention there was no comparably easy way to get the system to
4320 repeat its behavior every time it was run.
4321 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4322 FTYPE declarations has been removed, in favor of their ANSI
4323 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4324 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4325 the types of their sequence arguments can be determined at compile time.
4326 This is done through a new open code expansion for MAP which eliminates
4327 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4328 argument types can be determined at compile time.
4329 * The optimizer now transforms COERCE into an identity operation when it
4330 can prove that the coerced object is already of the correct type. (This
4331 can be a win for machine generated code, including the output of other
4332 optimization transforms, such as the MAP transform above.)
4333 * Credit information has been moved from source file headers into CREDITS.
4334 * Source file headers have been made more standard.
4335 * The CASE macro now compiles without complaining even when it has
4338 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4340 * (Note that the way that the PCL macroexpansions were rewritten
4341 to accommodate the change in DEFGENERIC below breaks binary
4342 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4343 not run under sbcl-0.6.2. Once we get out of alpha releases,
4344 i.e. hit release 1.0.0, we'll probably try to maintain binary
4345 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4346 and sbcl-1.4.4. Until then, however, it might be fairly common
4347 for maintenance releases to break binary compatibility.)
4348 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4350 * The reporting of unhandled signals has been changed to print some
4351 explanatory text as well as the report form. (Previously only
4352 the report form was printed.)
4353 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4354 it defines, so that the compiler no longer issues undefined function
4355 warnings for compiled-but-not-yet-loaded generic functions.
4356 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4357 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4358 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4359 section of the ANSI spec are supported.
4360 * The "" nickname for the "KEYWORD" package has been removed.
4361 The reader still handles symbol tokens which begin with a package marker
4362 as keywords, but it doesn't expose its mechanism for doing so in the
4363 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4364 * The system now issues STYLE-WARNINGs for contradictory TYPE
4365 proclamations. (Warnings for contradictory FTYPE proclamations would
4366 be nice too, but those can't be done usefully unless the type system
4367 is made smarter about FUNCTION types.)
4368 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4369 systematized, so that "*target-*.lisp is supposed to exist only on the
4370 target and imply that there's a related file which exists on the
4371 host, and *host-*.lisp is supposed to exist only on the host and imply
4372 that there's a related file which exists on the target. This involves a
4373 lot of renaming. Hopefully the acute confusion caused by the renaming
4374 will be justified by the reduction in chronic confusion..
4375 ** runtime-type.lisp -> early-target-type.lisp
4376 ** target-type.lisp -> late-target-type.lisp
4377 ** early-host-format.lisp -> early-format.lisp
4378 ** late-host-format.lisp -> late-format.lisp
4379 ** host-error.lisp -> misc-error.lisp
4380 ** early-error.lisp -> early-target-error.lisp
4381 ** late-error.lisp -> late-target-error.lisp
4382 ** host-defboot.lisp -> early-defboot.lisp
4383 ** code/misc.lisp -> code/target-misc.lisp
4384 ** code/host-misc.lisp -> code/misc.lisp
4385 ** code/numbers.lisp -> code/target-numbers.lisp
4386 ** code/early-numbers.lisp -> numbers.lisp
4387 ** early-host-type.lisp -> early-type.lisp
4388 ** late-host-type.lisp -> late-type.lisp
4389 ** host-typep.lisp -> typep.lisp
4390 ** load.lisp -> target-load.lisp
4391 ** host-load.lisp -> load.lisp
4392 ** host-disassem.lisp -> disassem.lisp
4393 ** host-insts.lisp -> insts.lisp
4394 ** byte-comp.lisp -> target-byte-comp.lisp
4395 ** host-byte-comp.lisp -> byte-comp.lisp
4396 ** host-signal.lisp -> signal.lisp
4397 ** host-defstruct.lisp -> defstruct.lisp
4398 ** late-target-type.lisp -> deftypes-for-target.lisp
4399 Furthermore, several other previously target-only files foo.lisp (e.g.
4400 hash-table.lisp and random.lisp) have been split into a target-and-host
4401 foo.lisp file and a target-only target-foo.lisp file, with their key type
4402 definitions in the target-and-host part, so that the cross-compiler will
4403 know more about target types.
4404 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4405 gone away. In their place are various *BACKEND-FOO* variables
4406 corresponding to the slots of the old structure.
4407 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4408 into the target SBCL has been fixed.
4409 * The chill.lisp system for loading cold code into a running SBCL
4411 * Support for the CMU CL "scavenger hook" extension has been removed.
4412 (It was undocumented and unused in the CMU CL sources that SBCL was
4413 derived from, and stale in sbcl-0.6.1.)
4414 * Various errors in the cross-compiler type system were detected
4415 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4416 (enabling various consistency checks). Many of them were fixed,
4417 but some hard problems remain, so the compiler is back to
4418 running without *TYPE-SYSTEM-INITIALIZED* for now.
4419 * As part of the cross-compiler type system cleanup, I implemented
4420 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4421 * I have started adding UNCROSS calls throughout the type system
4422 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4423 on cross-compiler input files will be able to go away, and various
4425 * CONSTANTP now returns true for quoted forms (as explicitly required
4428 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4430 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4431 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4432 (MAKE-LIST -1). In the longer term, it ought to become true
4433 that declarations are assertions even at SAFETY 1. For now, it's not
4434 quite true even at SAFETY 3, but it's at least more nearly true..
4435 (Note that this change seems to increases the size of the system by
4436 O(5%) and to decrease the speed of the compiler by 20% or more.)
4437 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4438 for the problem of printing dozens of lines of distracting information
4439 about low-level system machinery as part of the top stack frame
4440 on entry to the debugger when an undefined function was called.
4441 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4442 is not reset to COMMON-LISP-USER.
4443 * Compilation of stuff related to dyncount.lisp has been made conditional
4444 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4445 smaller. The various dyncount-related symbols have been moved into
4446 a new "SB-DYNCOUNT" package.
4447 * tty-inspect.lisp has been renamed to inspect.lisp.
4448 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4449 feature has gone away. (When we eventually port to other flavors of
4450 libc and/or Unix, we'll try to make the differences between flavors
4451 invisible at the user level.)
4452 * Various other *FEATURES* tags, and/or their associated conditionals,
4453 have been removed if obsolescent, or given better documentation, or
4454 sometimes given more-mnemonic names.
4456 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4458 * tidied up "make.sh" script
4459 * tidied up system directory structure
4460 * better "clean.sh" behavior
4461 * added doc/FOR-CMUCL-DEVELOPERS
4462 * many many small tweaks to output format, e.g. removing possibly-confusing
4463 trailing #\. character in DESCRIBE-INSTANCE
4464 * (EQUALP #\A 'A) no longer signals an error.
4465 * new hashing code, including EQUALP hashing
4466 * tidied up Lisp initialization and toplevel
4467 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4468 * command line argument processing
4469 * added POSIX-GETENV function to deal with Unix-ish environment variables
4470 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4471 terminating SBCL on EOF
4472 * non-verbose GC by default
4473 * There is no more "sbcl" shell script; the sbcl file is now the C
4474 runtime executable (just like CMU CL).
4475 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4477 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4478 symbol and package information as Lisp native symbol and package objects
4479 instead of strings naming symbols and strings naming packages. This way,
4480 whenever packages are renamed (as in warm init), debug information is
4481 transformed along with everything else.
4482 * tweaked the optimization policy declarations which control the building
4483 of SBCL itself. Now, among other things, the system no longer saves
4484 source location debugging information. (This helps two problems at once
4485 by reducing SBCL size and by keeping SBCL from trying to look for its
4486 sources -- which may not exist -- when reporting errors.)
4487 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4488 debugging and testing purposes
4489 * cleaned up printing, making the printer call PRINT-OBJECT for
4490 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4491 methods, giving nearly-ANSI behavior
4492 * converted almost all special variables to use *FOO* naming convention
4493 * deleted PARSE-TIME functionality, since it can be done portably
4494 * moved some files out of cold init into warm init
4495 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4497 * regularized formatting of source files
4498 * added an install.sh script
4499 * fixed ridiculous memory usage of cross-compiler by making
4500 compiler/alloc.lisp not try to do pooling unless it can hook
4501 itself into the GC of the cross-compilation host. Now the system
4502 builds nicely on my old laptop.
4503 * added :SB-ALLOC in target-features.lisp-expr
4504 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4505 was not implemented)
4506 * re-did condition handling and note reporting in the compiler. Notes
4507 are no longer handled by signalling conditions. Style warnings
4508 and warnings are handled more correctly and reported in such a way
4509 that it's easy to find one or the other in your output (so that you
4510 can e.g. figure out which of many problems caused COMPILE-FILE to
4512 * changed the severity of several compiler warnings from full WARNING
4513 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4514 compiler note reporting so that it doesn't use the condition system
4515 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4516 COMPILE-FILE command)
4517 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4518 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4519 no longer supported, since their implementation was deeply intertwingled
4520 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4521 * removed bogus "support" for compiler macros named (SETF FOO), and
4522 removed the compiler macro for SETF INFO (but only after making a fool
4523 of myself on the cmucl-imp mailing list by posting a bogus patch for
4524 DEFINE-COMPILER-MACRO..)
4525 * Compiled files containing forms which have side effects on the Lisp
4526 reader (such as DEFPACKAGE forms) are now handled more correctly.
4527 (Compiler queuing of top level lambdas has been suppressed by setting
4528 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4529 * deleted various currently-unused source files, e.g. gengc.lisp. They
4530 may be added back at some point e.g. when porting to other architectures,
4531 but until they are it's distracting to distribute them and to try to
4533 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4534 there were so many of them they're just distractions, and UNCROSS is
4535 known to be able to handle the current sources
4536 * moved PROFILE functionality into TRACE, so that it will be clear
4537 how the wrapping and unwrapping of functions when you profile them
4538 interacts with the wrapping and unwrapping of functions when you
4539 trace them. (Actually, the functionality isn't there yet, but at least
4540 the interface specification is there. Hopefully, the functionality will
4541 arrive with some future maintenance release.)
4542 * removed host-oops.lisp
4543 * changed signature of QUIT function to allow UNIX-CODE argument
4544 * fixed READ-SEQUENCE bug
4545 * tweaked verbose GC output so that it looks more like the progress
4546 output that ANSI specifies for functions like LOAD
4547 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4548 * added <http://sbcl.sourceforge.com> to the banner information printed by