1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
3 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
4 (see documentation for details.)
5 * revived support for OpenBSD (contributed by Josh Elsasser)
6 * bug fix: ROOM no longer suffers from occasional (AVER (SAP=
7 CURRENT END)) failures.
8 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
9 is now actually exported from SB-EXT as documented. (reported by
12 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
13 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
14 an executable in the search path, and does so in the child
15 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
16 been removed; it can be found in the CVS history, for somebody who
17 needs that search behavior (see the manual).
18 * minor incompatible change: compiler policy re. weakening type
19 checks has changed: now type checks are weakened on if SAFETY < 2
21 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
22 unparsing of directory pathnames as files. Analogously,
23 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
24 filename to parse into a directory pathname.
25 * enhancement: implicit generic function creation now signals a
26 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
27 users can bind handlers for and muffle around calls to LOAD.
28 * enhancement: RUN-PROGRAM allows unicode arguments and environments
29 to be used (using the default stream external format), and allows
30 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
31 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
32 strings and vectors whose element-type or simplicity is not fully
33 known at compile-time.
34 * optimization: STRING-TRIM and related functions no longer allocate
35 a new string when no trimming needs to be performed. These functions
36 are also faster than before when the input string has been declared
38 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
39 * optimization: READ-LINE is significantly faster for files containing
41 * optimization: non-open coded uses of character comparison operators
42 (e.g. char=) no longer cons when called with more than one parameter
43 on platforms supporting dynamic-extent allocation.
44 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
45 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
46 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
48 * bug fix: some sequence functions elided bounds checking when
50 * bug fix: too liberal weakening of union-type checks when SPEED >
52 * bug fix: more bogus fixnum declarations in ROOM implementation
55 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
56 * new feature: MAKE-HASH-TABLE now experimentally accepts a
57 :SYNCHRONIZED argument, which makes the hash-table safe for
58 concurrent accesses (but not iteration.) See also:
59 SB-EXT:WITH-LOCKED-HASH-TABLE, and
60 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
61 * optimization: CONCATENATE on strings is an order of magnitue faster
62 in code compiled with (> SPEED SPACE).
63 * optimization: SUBSEQ is ~50% faster on lists.
64 * bug fix: bug 417 fixed -- source location reporting is now more robust.
65 * bug fix: SUBSEQ on a list will now correctly signal an error if if
66 END is smaller then START.
67 * bug fix: SB-PROFILE will no longer report extra consing for nested
68 calls to profiled functions.
69 * bug fix: ROOM implementation had bogus fixnum declarations which
70 could cause type-errors when calling ROOM on large images.
71 * bug fix: if file compilation is aborted, the partial fasl is now
72 deleted, and COMPILE-FILE returns NIL as the primary value.
73 * bug fix: number of thread safety issues relating to SBCL's internal
74 hash-table usage have been fixed.
75 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
76 be returned from its body when the values were being returned
77 using unknown-values return convection and the W-P-O was wrapped
78 inside an UNWIND-PROTECT.
79 * bug fix: sb-posix should now compile again under Windows, enabling
82 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
83 * incompatible change: hash-table accessor functions are no longer
84 automatically protected by locks. Concurrent accesses on the same hash-table
85 from multiple threads can give inconsistent results or even corrupt the
86 hash-table completely. Multi-threaded applications should do their own
87 locking at the correct granularity. In the current implementation it is
88 still safe to have multiple readers access the same table, but it's not
89 guaranteed that this property will be maintained in future releases.
90 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
91 and will signal an error at runtime.
92 * enhancement: SB-THREAD package now exports a semaphore interface.
93 * enhancement: CONS can now stack-allocate on x86 and
94 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
96 * enhancement: nested lists can now be stack allocated on
97 platforms providing stack allocation support.
98 * enhancement: dynamic-extent support has been extended to support
99 cases where there are multiple possible sources for the stack
101 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
102 if the mutex is uncontested on Linux.
103 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
104 as specified by the CLHS. (thanks to Tobias Rittweiler)
105 * bug fix: NaN comparison now works on x86-64.
106 * bug fix: CLOSE :ABORT T on a stream with pending output now
108 * bug fix: instances of non-standard metaclasses using standard
109 instance structure protocol sometimes missed the slot type checks
111 * bug fix: known functions can be inlined in high-debug code.
113 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
114 * minor incompatible change: the MSI installer on Windows no longer
115 associates .lisp and .fasl files with the installed SBCL.
116 * minor incompatible change: :UNIX is no longer present in *FEATURES*
117 on Windows. (thanks to Luis Oliviera)
118 * new feature: SB-CLTL2 contrib module now implements
119 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
120 * optimization: scavenging weak pointers is now more efficient,
121 requiring O(1) instead of O(N) per weak pointer to identify
122 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
123 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
124 method bodies using constant slot-names when the first argument is
125 a specializer parameter for the method.
126 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
127 names, when the class of the instance is a direct instance
128 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
129 as slow as the constant slot-name case.)
130 * optimization: member type construction is now O(N) instead
132 * optimization: UNION and NUNION are now O(N+M) for large
133 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
135 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
136 now more readable in environments like Slime which display it.
137 (thanks to Tobias C. Rittweiler)
138 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
139 was able to infer the correct argument type for the object on which
140 the CAS operation was being performed.
141 * bug fix: copy propagation interfered with parallel assignment
142 semantics in local calls. (reported by Paul Khuong)
143 * bug fix: the signed modular fixnum shift compiled to wrong code on
144 x86 and x86-64. (spotted by a slight modification to some of
147 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
148 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
149 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
150 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
152 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
153 and generic function now print the object identity for anonymous
154 instances. (thanks to Attila Lendvai)
155 * enhancement: as an extension to MOP, SBCL now supports SETF of
156 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
157 (thanks to Attila Lendvai)
158 * enhancement: sb-cover has an alternate annotation mode which puts
159 the form annotations at the car of the form, instead of annotating
160 it completely. The benefit of this mode is that it shows better
161 which forms were instrumented by the compiler.
162 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
163 by printer control variables. (reported by Dan Corkill)
164 * bug fix: system leaked memory when delayed output was performed by
165 the OS in smaller chunks then expected. (thanks to David Smith)
166 * bug fix: system leaked memory when file streams were not closed
168 * bug fix: large objects written to slow streams that were modified
169 after the write could end up with the modified state written to
170 the underlying file descriptor.
171 * bug fix: multiple threads operating in parallel on the same stream
172 could cause buffer-overflows.
173 * bug fix: source location information is stored correctly
174 (broken since 1.0.6). This bug would generally show up as the
175 Slime debugger higlighting the wrong form when the sldb-show-source
177 * bug fix: Branch forms are again annotated as branches in the sb-cover
179 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
180 on x86-64. (reported by Gregory Vanuxem)
181 * bug fix: EQUALP could return wrong results for structures with raw
182 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
183 word). (reported by Vjacheslav Fyodorov)
184 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
187 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
188 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
189 atomic compare-and-swap operations on threaded platforms.
190 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
191 allows assining a global minimum value to optimization qualities
192 (overriding proclamations and declarations).
193 * enhancement: closed over variables can be stack-allocated on x86
195 * performance bug fix: GETHASH and (SETF GETHASH) are once again
197 * optimization: slot definition lookup is now O(1). This speeds up
198 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
199 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
200 * optimization: ASSOC and MEMBER can now be open-coded for all
201 combinations of keyword arguments when second argument is constant
202 and SPEED >= SPACE. In other cases a specialized version is
204 * bug fix: using obsoleted structure instances with TYPEP and
205 generic functions now signals a sensible error.
206 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
207 (reported by Kristoffer Kvello)
208 * bug fix: backtrace construction is now more careful when making
209 lisp-objects from pointers on the stack, to avoid creating bogus
210 objects that can be seen by the GC.
211 * bug fix: defaulting of values in contexts expecting more than 7
212 variables now works on x86-64. (reported by Christopher Laux)
213 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
215 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
216 as the property-list of a symbol.
217 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
218 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
219 situations requested, are once again file-compileable. (reported
222 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
223 * MOP improvement: support for user-defined subclasses of
224 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
225 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
226 the expansion of DEFMETHOD, is responsible for generating a form
227 which creates a list of specializers when evaluated. Additional
228 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
229 debugging and introspective support.
230 * minor incompatible change: the (unsupported) spinlock interface
231 has changed: free spinlock now has the value NIL, and a held spinlock
232 has the owning thread as its value.
233 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
234 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
236 * enhancement: name of a socket-stream is now "a socket" instead of
238 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
239 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
240 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
241 for generic functions. (Thanks to Tobias C. Rittweiler)
242 * optimization: bignum printing speed has been improved by 20-40%
243 (depending on the bignum size.)
244 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
246 * bug fix: the cache used by the CLOS to store precomputed effective
247 methods, slot offsets, and constant return values is now thread and
249 * bug fix: generic function dispatch function updating is now thread
250 and interrupt safe (in the sense that the known issues have been
252 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
253 * bug fix: interning EQL-specializers is now thread and interrupt safe.
254 * bug fix: asdf systems with dependencies to the SB-POSIX or
255 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
256 * bug fix: interrupt safety of applicable method computation has been
259 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
260 * new contrib: sb-cover, an experimental code coverage tool, is included
262 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
263 significantly faster.
264 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
265 produce more efficient code on x86-64 if the compiler can determine
266 that (+ INDEX <constant>) does not require a bounds check and FOO
267 has an element type at least 8 bits wide.
268 * enhancement: a new, experimental synchronous timeout facility is
269 provided. Refer to SB-SYS:WITH-DEADLINE for details.
270 * enhancement: when a symbol name conflict error arises, the
271 conflicting symbols are always printed with a package prefix.
272 (thanks to Kevin Reid)
273 * enhancement: stepping is now once again supported on the SPARC and
274 MIPS platforms. (It is also now more likely to work on CheneyGC
276 * enhancement: sb-sprof can now also track and report accurate call
278 * bug fixes: the treatment of non-standard subclasses of
279 SB-MOP:SPECIALIZER is more correct.
280 * incompatible change: PURIFY no longer copies the data from the
281 dynamic space into the static and read-only spaces on platforms
282 that use the generational garbage collector
283 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
285 * bug fix: GC race condition occasionally resulting in crashes with
286 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
288 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
290 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
291 a STYLE-WARNING to be signalled (regression from 1.0.4.)
292 * bug fix: an asynchronous interrupt could previously leave the
293 system running with GC inhibited.
294 * bug fix: a DECLARE form evaluated at top-level now causes an error
295 rather than silently (or verbosely) returning NIL.
296 * bug fix: trying to dissassemble functions compiled at high DEBUG could
297 result in a error being signalled due to source form lookup errors.
298 (reported by Peter Graves)
300 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
301 * incompatible change: removed writer methods for host-ent-name,
302 host-ent-addresses -- changing the values did not update the DNS
304 * minor incompatible change: changed experimental JOIN-THREAD interface
305 * documentation: the manual now lists reader and writer methods
306 in class slot documentation sections. (thanks to Richard M Kreuter)
307 * documentation: unwinding from asyncronous events has been
308 documented as unsafe.
309 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
310 in multithreaded application code.
311 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
312 platforms. (thanks to James Anderson for the optimization hint)
313 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
315 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
316 variants no longer cons.
317 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
318 their NOT- variants no longer cons.
319 * optimization: EQUAL hash tables no longer use SXHASH for objects
320 of all data types, but instead use an EQL hash for types for which
321 EQUAL is the same as EQL.
322 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
323 are significantly faster.
324 * optimization: new STRING-OUTPUT-STREAM implementation is much
325 faster for large outputs and conses 30% less on average.
326 * enhancement: XREF information is now collected for references made
327 to global variables using SYMBOL-VALUE and a constant argument.
328 * enhancement: SIGINT now causes a specific condition
329 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
330 * bug fix: FILE-POSITION used to signal an error in some cases where
331 ANSI requires it to return NIL.
332 * bug fix: ADJUST-ARRAY is now interrupt-safe.
333 * bug fix: adding and removing fd-handlers is now interrupt-safe.
334 * bug fix: inlined calls to C now ensure 16byte stack alignment on
336 * bug fix: bad type declaration in the CLOS implementation has
337 been fixed. (reported by James Anderson)
338 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
339 (thanks to Lutz Euler)
340 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
341 required by the CLHS. (reported by Samium Gromoff)
342 * bug fix: dead unbound variable references now signal an error.
343 (reported by Marco Monteiro)
344 * bug fix: / with an unused value was being deleted in safe code.
345 (thanks to Marco Monteiro and Kevin Reid)
346 * bug fix: number of characters that can be written onto a single
347 line in a file is unlimited.
348 * bug fix: some GC deadlocks caused by asynchronous interrupts have
349 been fixed by inhibiting interrupts for when GC is disbled.
350 * bug fix: some interrupt safety issues with GETHASH, (SETF
351 GETHASH), CLRHASH and REMHASH have been fixed.
352 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
353 specifier no longer causes infinite recursion.
354 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
355 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
356 * bug fix: an &environment argument with the correct variable information
357 is passed to macros that are expanded during byte compilation.
358 (reported by Samium Gromoff)
359 * bug fix: a logic error was causing the hash function for bignums to
360 have excessive amounts of collisions. (reported by Faré Rideau)
361 * bug fix: modifying the contents of an array could change the return
362 value of SXHASH on that array, which is only allowed for strings
363 and bit vectors. (bug introduced in 0.9.16)
364 * improvement: the x86-64/darwin port now passes all tests (except
365 for the debugger tests) but should still be considered
366 experimental until this is fixed.
367 * improvement: a style-warning is signaled for CASE (etc) clauses with
368 duplicate keys. (thanks to Kevin Reid)
369 * improvement: macos/x86 and macos/x86-64 now use mach exception
370 handlers for dealing with illegal instructions (for trapping and
371 error handling) and memory protection violations (for GC).
373 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
374 * new platform: experimental support for x86-64/darwin (MacOS).
375 * incompatible change: the thread-safe (on most platforms) getaddrinfo
376 and getnameinfo sockets functions are used instead of gethostbyaddr
377 and gethostbyname, on platforms where the newer functions are available.
378 As a result, the ALIASES field of HOST-ENT will always be NIL on these
380 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
381 don't print the startup banner, but behave as if --noinform was passed
382 as a command line argument. (thanks to Kevin Reid)
383 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
384 * optimization: code using alien values with undeclared types is much faster.
385 * optimization: the compiler is now able to open code SEARCH in more cases.
386 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
387 * bug fix: using standardized COMMON-LISP special variables as loop
388 variables no longer signals bogus package lock violations. (reported
390 * bug fix: declaring local loop variables to be of a range-limited type
391 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
392 (reported by Andras Simon)
393 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
394 bugs remain on x86-64.)
395 * bug fix: the #= and ## reader macros now interact reasonably with
396 funcallable instances.
397 * bug fix: type-checks for function arguments were compiled using the
398 compiler policy settings of the wrong lexical environment (in 1.0.2
400 * bug fix: SHADOW accepts characters as string designators, as required
401 by the spec (thanks to Eric Marsden)
402 * bug fix: fixed GC safety issues when foreign functions are called with
403 non-base strings as arguments
404 * bug fix: more consistent error output for fatal-compiler-warnings, like
406 * bug fix: functions with proper names were showing up as NIL in
407 backtraces (reported by Edi Weitz, regression in 1.0.2)
409 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
410 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
411 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
412 produce more efficient code on the x86 if the compiler can determine
413 that (+ INDEX <constant>) does not require a bounds check and FOO
414 has an element type at least 8 bits wide.
415 * bug fix: references to undefined variables are handled the same way
416 in toplevel forms as in the normal compiler
417 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
419 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
420 (thanks to Jon Buller)
421 * improvement: faster compilation times for complex functions
422 * improvement: added readlink support to SB-POSIX (thanks to Richard
425 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
426 * improvement: experimental support for mach exception handling on
427 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
429 * improvement: support for GBK external format.
430 (thanks to Chun Tian (binghe))
431 * improvement: the debugger now displays variables that have been closed
432 over, in code compiled with (DEBUG 2) or higher
433 * improvement: support for executable cores on NetBSD (thanks to
435 * new feature: added a RESTART-FRAME debugger command
436 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
437 be used to provide an implementation for FILE-POSITION on Gray streams
438 (thanks to Eric Marsden)
439 * new feature: add syslog, openlog and closelog support to SB-POSIX
440 (thanks to Richard Kreuter)
441 * optimization: the function call overhead in code compiled with
442 a high DEBUG optimization setting is significantly
443 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
444 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
445 (thanks to Tony Martinez)
446 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
447 evaluated once (reported by Kevin Reid)
448 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
449 works again (thanks to Kilian Sprotte)
450 * bug fix: an error is signaled for tagbodies with duplicate tags
451 (thanks to Stephen Wilson)
452 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
453 * bug fix: Win32 port can now handle foreign code unwinding Lisp
454 stack frames from alien callbacks.
455 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
456 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
457 65k pages (thanks to David Woodhouse)
458 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
460 changes in sbcl-1.0.1 relative to sbcl-1.0:
461 * new platform: FreeBSD/x86-64, including support for threading.
462 * new feature: the compiler stores cross-referencing information
463 abount function calls (who-calls), macroexpansion (who-macroexpands)
464 and special variables (who-binds, who-sets, who-references) for code
465 compiled with (< SPACE 3). This information is available through the
466 sb-introspect contrib.
467 * new feature: users may subclass SEQUENCE, and have instances of
468 these classes interoperate with standard Common Lisp functions if
469 a number of methods are defined. (This feature is experimental
470 and the interface subject to change based on feedback from SBCL
471 users and the general community)
472 * improvement: sb-sprof traces call stacks to an arbitrary depth on
473 x86/x86-64, rather than the previous fixed depth of 8
474 * improvement: another pthread back-end of mutex "pthread-futex".
475 * bug fix: non-ascii command-line arguments are processed correctly
476 (thanks to Yaroslav Kavenchuk)
477 * bug fix: non-required arguments were not passed correctly when a method
478 defined using DEFMETHOD was called from a mop-generated method using
479 CALL-NEXT-METHOD (reported by Pascal Costanza)
480 * bug fix: recursion is now permitted in accessors through
481 SLOT-UNBOUND. (reported by Pascal Costanza)
482 * bug fix: an error was signaled at startup if the HOME environment
483 variable was defined, but had an empty value (reported by Peter Van Eynde)
484 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
485 signal a PROGRAM-ERROR, not a BUG.
486 * bug fix: SB-POSIX:READDIR works when built with large file support.
487 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
489 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
490 for better automatic indentation support. (thanks to Matt Pillsbury)
491 * optimization: loading generic functions no longer takes O(n^2) time,
492 proportional to the amount of methods in the generic function
493 (reported by Todd Sabin and Jeremy Brown)
494 * optimization: the FIND and POSITION family of sequence functions
495 are significantly faster on arrays whose element types have been
497 * improvements to the Windows port:
498 ** Intermittent heap corruption problems have been fixed. (thanks
499 to Alastair Bridgewater)
500 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
502 ** Lisp is able to unwind foreign exception frames from alien
503 callbacks. (thanks to Alastair Bridgewater)
505 changes in sbcl-1.0 relative to sbcl-0.9.18:
506 * improvement: experimental support for threading on FreeBSD/x86.
507 (thanks to NIIMI Satoshi)
508 * improvement: runtime option --dynamic-space-size can be used
509 to set the size of the dynamic space reserved on startup.
510 * improvement: floating point modes in effect are now saved in
511 core, and restored on startup.
512 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
513 startup, not time since first call to GET-INTERNAL-REAL-TIME.
514 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
515 threads are not running after *SAVE-HOOKS* have run.
516 * improvement: writes to CLOS instance slots are type-checked in code
517 compiled with (SAFETY 3)
518 * improvement: floating-point exception handling on FreeBSD (thanks to
520 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
521 (thanks to Zach Beane)
522 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
524 * improvement: added support for the Shift-JIS external format.
525 (contributed by NIIMI Satoshi)
526 * improvement: callbacks are supported on Linux/PPC. (thanks to
528 * bug fix: compiler bug triggered by a (non-standard) VALUES
529 declaration in a LET* was fixed. (reported by Kaersten Poeck)
530 * bug fix: file compiler no longer confuses validated and already
531 dumped structurres. (reported by Kaersten Poeck)
532 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
533 fill-pointer signals a type-error as required. (thanks to
535 * bug fix: disassemly of funcallable instances works.
536 * bug fix: single stepping on PPC.
537 * bug fix: fix thread-safety problems in the type system (generally
538 manifesting as nonsensical errors like "STRING is a bad type specifier
539 for sequences" or "The value 1 is not of type FIXNUM")
540 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
541 (reported by Josip Gracin)
542 * bug fix: an error is signaled for attempts to displace arrays with
543 incompatible element types (thanks to Mario Mommer)
544 * bug fix: more correct handling of wide characters in debug info
545 (bug reported by Attila Lendvai and fixed by Juho Snellman)
546 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
547 and don't cause extra consing
548 * optimization: MAP and MAP-INTO are significantly faster on vectors
549 whose elements types have been declared.
550 * Improvements to SB-SPROF:
551 ** Support for allocation profiling
552 ** Reduced profiling overhead, especially for long profiling runs
553 * Improvements to the Windows port:
554 ** floating point exceptions are now reported correctly.
555 ** stack exhaustion detection works partially.
556 ** more accurate GET-INTERNAL-REAL-TIME.
557 ** better RUN-PROGRAM behaviour in case of errors in the child
559 ** PROBE-FILE now simplifies pathnames correctly.
560 ** DIRECTORY now works correctly with :WILD-INFERIORS.
561 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
562 1st 1970 (although time timezone data will be unreliable).
563 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
565 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
566 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
567 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
569 * bug fix: two potential GC deadlocks affecting threaded builds.
570 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
571 returns T (reported by Anton Kazennikov)
572 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
573 with non-variable places
574 * bug fix: remove a race condition in the setting of
575 funcallable-instance functions, this should make threaded CLOS
576 code more stable against memory faults.
577 * bug fix: corruption of specials when unbinding is interrupted by an
578 asynchronous unwind (reported by Hannu Koivisto)
579 * improvement: the debugger will now also display local variables that
580 are only used once, for code compiled with a DEBUG optimization quality
583 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
584 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
585 * incompatible change: External-format support for FFI calls. The
586 SB-ALIEN:C-STRING no longer implies an ASCII
587 external-format. Instead the string is subject to external-format
588 conversion. Additionally return values of type SB-ALIEN:C-STRING
589 are by default of type (SIMPLE-ARRAY CHARACTER), not
590 SIMPLE-BASE-STRING. For an alien type that behaves like the old
591 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
592 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
593 * incompatible change: SB-EXT package no longer contains the
594 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
595 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
597 * incompatible change: the single-stepper is no longer available
598 on Alpha, Mips and Sparc platforms.
599 * minor incompatible change: the direct superclasses of
600 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
601 not (STANDARD-OBJECT FUNCTION). This makes the
602 class-precedence-lists of GENERIC-FUNCTION and
603 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
605 * minor incompatible change: the default stream external format on
606 non-unicode SBCL is no longer detected from the locale, but is always
608 * new feature: Add a version of evaluator that uses an interpreter instead
609 of the compiler. EVAL still uses the compiler by default, to switch it
610 to use the interpreter, set the value of the variable
611 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
612 * minor incompatible change: the single-stepper REPL has been merged
613 with the normal debugger (see the "Stepping" heading of the debugger help
614 for more details). The debugger command STEP will no longer switch
615 to the single-stepper REPL.
616 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
617 for a type now works.
618 * bug fix: loading of default sysinit file works. (thanks to Leonid
620 * bug fix: better detection of circularities in the file-compiler.
621 (reported by Marco Monteiro)
622 * bug fix: the CL pathname functions now work with files that have
623 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
624 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
625 systems with Unix98 pty semantics.
626 * bug fix: ASDF-INSTALL will now work with bsd tar.
627 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
629 * bug fix: timers expiring in dead threads no longer cause a
630 type-error (reported by Paul "Nonny Mouse").
631 * bug fix: thanks to more lightweight single-stepper instrumentation,
632 code compiled with (DEBUG 3) will compile and execute significantly faster,
633 and will have more accurate type-inferencing than before
634 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
635 whose bindings are modified
636 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
637 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
638 ** replace dummy implementations of CL:MACHINE-INSTANCE and
639 CL:SOFTWARE-VERSION with proper version
641 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
642 * feature: implemented the READER-METHOD-CLASS and
643 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
644 as specified by AMOP.
645 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
647 * optimization: faster LOGCOUNT implementation on x86 and x86-64
648 (thanks to Lutz Euler)
649 * optimization: hashing of general arrays and vectors has been
650 improved. (reported by Any Fingerhut)
651 * enhancement: SB-INTROSPECT is now able to find definitions of
652 profiled functions. (thanks to Troels Henriksen)
653 * enhancement: compiler-macro expansion applies now to FUNCALL forms
654 as well, allowing compiler-macros for SETF-functions to expand.
655 * enhancement: step-instrumentation no longer wraps known
656 single-value functions with multiple-value context, allowing
657 better type inference.
658 * fixed bug #337: use of MAKE-METHOD in method combination now works
659 even in the presence of user-defined method classes. (reported by
660 Bruno Haible and Pascal Costanza)
661 * fixed bug #339(c): if there are applicable methods not part of any
662 long-form method-combination group, call INVALID-METHOD-ERROR.
663 (reported by Bruno Haible)
664 * fixed bug #361: the :FUNCTION initarg in the protocol for
665 initialization of methods can now be used to override
666 internally-produced optimized functions. (reported by Bruno
668 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
669 system-provided lambda expression no longer cause warnings about
670 unbound #:|pv-table| symbols.
671 * bug fix: improved the handling of type declarations and the
672 detection of violations for keyword arguments with non-constant
674 * bug fix: potentially erronous calls to PATHNAME and
675 MERGE-PATHNAMES were being flushed in some cases.
676 (reported by Richard Kreuter)
677 * bug fix: compiled calls to TYPEP were mishandling obsolete
678 instances. (reported by James Bielman and Attila Lendvai)
679 * bug fix: format strings with ~> without matching ~< no longer
680 trigger an AVER, but signal an understandable error instead.
681 (reported by Antonio Martinez)
682 * bug fix: specifying an output-file without a pathname-type for
683 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
684 of input-file instead of "fasl". (reported by Robert Dodier)
685 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
686 (reported by James Y Knight).
687 * bug fix: compiler-macros-function did not consider the environment
688 argument for shadowing by local functions.
689 * bug fix: compiler-macros expansion was inhibited by local INLINE
691 * bug fix: inline expansions of known functions were subject to
692 step-instrumentation in high debug policies, leading to problems
694 * bug fix: compiler failed to differentiate between different CONS
696 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
697 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
698 * bug fix: #368: incorrect use of expressed vs. upgraded array
700 * bug fix: #306a: more precise unions of array types.
701 * thread-safety improvements:
702 ** CONDITION-WAIT could return early on Linux, if the thread was
703 interrupted and subsequently continued with SIGCONT.
704 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
706 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
707 * added support for the ucs-2 external format. (contributed by Ivan
709 * minor incompatible change: pretty printing of objects of type
710 (cons symbol) is, in the default pprint-dispatch-table, now
711 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
713 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
714 called later in a class's lifetime, possibly as late as when the
715 first instance of the class is created. Previously,
716 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
717 class became finalizeable.
718 * fixed bug: FILE-POSITION sometimes returned inconsistent results
719 for multibyte external-format streams. (thanks to "vbzoli")
720 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
721 with :ALLOCATION :CLASS inherited from superclasses of the
723 * fixed bug: anonymous classes can now be created using the :NAME
724 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
725 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
726 * fixed bug: core-files saved with :EXECUTABLE T can again be
727 executed when SBCL_HOME isn't set. (reported by James Knight)
728 * fixed bug: toplevel LOCALLY forms with declarations could
729 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
730 * fixed bug: printing from several different threads using different
731 values of *print-case* could cause invalid output, due to
732 some internal special variables of the printer not being bound
733 thread-locally (reported by Max Mikhanosha)
734 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
735 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
736 (reported by Pascal Costanza)
737 * fixed bug: FINALIZE-INHERITANCE is called from
738 REINITIALIZE-INSTANCE on classes when the class has previously
739 been finalized, as required by AMOP.
740 * minor code generation optimizations:
741 ** better register allocation in CLOS dispatching functions
742 ** overflow detection when coercing signed bytes to fixnums on x86-64
743 ** is now implemented with one IMUL instruction instead of three shifts
744 ** more efficient bit-vector access on x86 and x86-64
745 ** more efficient access to raw structure slots on x86 and x86-64
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
750 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
751 * feature: thread support on Solaris/x86, and experimental thread support
753 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
755 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
756 default method for SLOT-UNBOUND.
757 * minor incompatible change: prevent the user from specializing the
758 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
759 somewhat counter to the intent of the protocol, I (CSR) think, and
760 additionally it just doesn't work in SBCL as currently
761 implemented, thanks to optimizations (that are always valid for
762 the other three SLOT-VALUEish functions, but not for the setter).
763 * minor incompatible change: the :SB-LDB feature is now enabled by
764 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
765 the low-level debugger.
766 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
767 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
768 down to an SB-SYS:FD-STREAM, allowing direct communication in
769 more cases. Temporary files and pipes are still used as a fallback
771 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
772 on the non-reentrant C functions localtime(3) and gmtime(3).
773 * bug fix: global optimization policy was not visible in LOCALLY and
775 * bug fix: class objects can be used as specializers in methods.
776 (reported by Pascal Costanza)
777 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
778 with a type error. (reported by blitz_ on #lisp)
779 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
780 failed with a type error. (reported by Pascal Bourguignon)
781 * bug fix: merging pathnames against defaults with :DIRECTORY
782 starting with '(:RELATIVE :BACK) should preserve the :BACK.
783 (reported by James Y Knight)
784 * bug fix: saving large (>2GB) cores on x86-64 now works
785 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
786 constant index and a (simple-array (signed-byte 32)) array
787 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
788 error (patch by Robert J. Macomber)
789 * feature: TIME now displays the amount of run-time spent in GC
790 * bug fix: The debugger now does a better job of respecting
791 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
792 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
793 workaround for bug 403.)
794 * bug fix: floating point exception handling now works on FreeBSD
795 (thanks to NIIMI Satoshi)
796 * fixed some bugs revealed by Paul Dietz' test suite:
797 ** MISC.641: LET-conversion were not supposed to work in late
799 * improvements to the Win32/x86 port:
800 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
801 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
803 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
804 * new feature: source path information is generated for macro-expansion
805 errors for use in IDE's like Slime (thanks to Helmut Eller)
806 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
807 * bug fix: compiler does not loop forever on an invalid type in
809 * improvement: compilation of most CLOS applications is significantly
811 * optimization: added a limited bytecode compiler for simple toplevel
812 forms, speeding up compilation and FASL loading
813 * bug fix: the statistical profiler now properly distinguishes anonymous
816 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
817 * minor incompatible change: in sbcl-0.9.11 (but not earlier
818 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
819 into subclasses of the finalized class. Now user calls to
820 FINALIZE-INHERITANCE finalize just the one class, and calls by the
821 system return before any subclasses are finalized.
822 * minor incompatible change: The reader no longer ignores errors
823 regarding non-existent packages in #+ and #- feature tests.
824 * new feature: command line options --no-sysinit, --no-userinit to
825 inhibit loading the corresponding init files
826 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
827 for sending data through UDP sockets (thanks to François-René Rideau)
828 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
829 error is signalled instead (thanks to François-René Rideau)
830 * minor incompatible change: Error signalling behaviour of lexical
831 operations violating package locks has changed slightly. Refer to
832 documentation on package locks for details.
833 * enhancement: EVAL can process IF-expressions without resorting to the
835 * bug fix: Some operations involving SIN, COS, or TAN weren't being
836 constant-folded properly due to missing out-of-line functions.
837 (reported by Mika Pihlajamäki)
838 * bug fix: LISTEN sometimes returned T even in cases where no data was
839 immediately available from the stream
840 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
841 were reversed. (reported by Levente Mészáros)
842 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
843 when given a vector displaced to an adjustable array. (reported
845 * bug fix: derivation of float boundaries from numbers outside the
846 appropriate float range (reported by John Wiseman)
847 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
849 * bug fix: precomputing cache entries for generic functions with
850 some subclasses of specializers as yet invalid does not attempt to
851 fill a cache line with a negative offset. (reported by Levente
853 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
854 allow condition class objects as arguments to DOCUMENTATION and
855 (SETF DOCUMENTATION); only find and set documentation for
856 structure names for the STRUCTURE doc-type. (suggested by Gary
858 * improvements to the Win32/x86 port:
859 ** (user-homedir-pathname) and default initialization file
860 locations now know about the user's "Documents and Settings"
861 directory (thanks to Yaroslav Kavenchuk)
862 ** run-program is implemented (thanks to Mike Thomas)
863 ** sockets support (thanks to Timothy Ritchey)
864 ** better backtrace support (thanks to Alastair Bridgewater)
865 ** sb-grovel supported
866 ** asdf-install and sb-posix work somewhat
867 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
868 * improvements to the Solaris/x86 port:
869 ** works on Solaris 11/Solaris Express
870 ** floating-point exception handling support
871 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
872 * fixed some bugs revealed by Paul Dietz' test suite:
873 ** REMOVE-METHOD returns its generic function argument even when
874 the method is not one of the generic functions' methods.
875 ** objects declared as MEMBER types can be admissible arguments to
877 ** printing characters should simply be printed by the FORMAT ~:C
879 ** compiler failure when compiling functions with hairy constant
880 defaults for optional parameters.
881 ** compiler produces wrong code when MAYBE-INLINE-expanding a
882 function, which is already optimized.
884 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
885 * new platform: experimental support for SBCL x86/Darwin, including
886 MacOS X 10.4.5 on Intel.
887 * new feature: Unicode character names are now known to the system
888 (through CHAR-NAME and NAME-CHAR).
889 * new feature: the filesystem location of SBCL's core file is
890 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
891 * minor incompatible change: the contrib modules SB-POSIX and
892 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
893 this change is to make it easier to distribute
894 location-independent binaries.
895 * bug fix: as implied by AMOP, standardized classes no longer have
896 slots named by external symbols of public packages. (reported by
898 * optimization: calling functions via constant symbols -- as in
899 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
900 via the function object as in (FUNCALL #'FOO).
901 * optimization: on x86 and x86-64, the calling convention for the
902 case of dealing with an unknown number of values has been altered
903 to be friendlier to the prediction heuristics implemented,
904 particularly with reference to CALL and RET pairing. (thanks to
905 Alastair Bridgewater)
906 * optimization: on x86, the code for access to thread-local symbol
907 values and binding/unbinding of thread-local symbols is smaller.
908 (thanks to Alastair Bridgewater)
909 * enhancement: CONSTANTP is now able to determine constantness of
910 more complex forms, including calls to constant-foldable standardized
911 functions and some special forms beyond QUOTE.
912 * fixed bug: occasional GC crashes on Solaris/x86
913 * optimization: x86-64 supports stack allocation of results of simple
914 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
915 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
916 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
917 (thanks to James Knight)
918 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
919 by the standard (thanks to Stephen Compall)
921 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
922 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
923 be used for bundling the runtime and the core file into one
924 executable binary. This feature is not currently supported on all SBCL
925 platforms. (thanks to James Bielman and NIIMI Satoshi)
926 * new feature: a generational or ephemeral garbage collector is now
927 the default on the PowerPC platform (both Linux and Darwin). The
928 old Cheney (stop and copy) collector is a build-time option.
929 (thanks to Cyrus Harmon, after Raymond Toy)
930 * minor incompatible change: the method by which SBCL finds its
931 contributed modules has changed; it no longer relies on symbolic
932 links from an $SBCL_HOME/systems directory, but searches directly
933 in the subdirectories of $SBCL_HOME.
934 * enhancement: the dynamic heap size on the Linux/PPC platform is
935 markedly larger, even using the older Cheney garbage collector.
936 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
937 certain complicated string types. (reported by Gary King)
938 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
939 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
940 * fixed bug: DOCUMENTATION on structures no longer signals an
941 error if no structure type of the right name exists. (reported by
943 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
944 applicable non-standard primary or :AROUND method on
945 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
946 no longer cause unbound variable errors. (reported by Kevin Reid)
947 * fixed bug: metacircle resolution in cases where methods have slots
948 added before the slots from STANDARD-METHOD. (reported by Jean
950 * fixed bug: the Power PC assembly code for calling into lisp
951 sometimes computed its return address wrongly (depending on the
952 15th bit of the address, fixed at link time). This used to
953 manifest itself in a segmentation violation while building PCL.
954 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
955 many others over the years)
956 * fixed bug: ROOM no longer reports silly numbers for stack usage if
957 the stack pointer (treated as a signed integer) is negative.
958 (thanks to Peter van Eynde)
960 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
961 * new platform: experimental support for the Windows operating
962 system has been added. (thanks to Alastair Bridgewater)
963 * fixed several bugs in and robustified the PPC FFI (including
964 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
965 * bug fix: allow non-simple string symbol names (reported by Paul
967 * bug fix: interrupt handling on NetBSD (thanks to Richard M
969 * bug fix: saving a core corrupted callbacks on x86/x86-64
970 * bug fix: closed a loophole in metacircularity detection and
971 grounding in the PCL implementation of CLOS.
972 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
973 (reported by tomppa on #lisp)
974 * bug fix: clear the direction flag on Lisp -> C transitions, as
975 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
976 (reported by Andrej Grozin and Hendrik Maryns)
977 * optimization: major improvements to GC efficiency on GENCGC platforms
978 * optimization: faster implementation of EQUAL
979 * optimization: emit more efficient opcodes for some common
980 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
981 * optimization: several other minor code-generation improvements on x86-64
982 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
984 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
985 * minor incompatible change: (SETF CLASS-NAME) and (SETF
986 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
987 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
988 by ANSI as a generic function, and for consistency (SETF
989 GENERIC-FUNCTION-NAME) is treated likewise.
990 * fixed bug #233.b: make constraint propagation notice when a variable
991 value is changed after it is referenced but before it is used
992 * fixed bug #296: no more arbitrary behaviour from filenames with
993 odd characters as --load arguments. (reported by Adam Warner)
994 * fixed bug #390: :CHARACTER-SET pathname components now work as
995 expected. (reported by Tim Daly Jr)
996 * fixed bug #391: complicated :TYPE intersections in slot
997 definitions no longer cause an error in PCL internals.
998 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
999 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1000 returning the number of octets which would be written to the
1001 file-stream. (thanks to Robert J. Macomber)
1002 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1003 arguments to the internal stream functions. (thanks to David
1005 * bug fix: stack exhaustion is now handled on the x86/SunOS
1006 platform. (thanks to Daisuke Homma)
1007 * bug fix: more accurate ROOM results on GENCGC platforms
1008 * optimization: improved type inference for arithmetic-for
1009 index variables in LOOP
1010 * optimization: faster floating-point SQRT on x86-64
1011 * fixed some bugs revealed by Paul Dietz' test suite:
1012 ** DOCUMENTATION returns NIL instead of "" for method combinations
1013 that don't have a docstring
1015 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1016 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1017 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1018 therefore are not customizeable by user code (as seems to be at
1019 least permitted and maybe required by AMOP). As a consolation,
1020 however, the SBCL implementation of these functions now calls
1021 REINITIALIZE-INSTANCE as specified by AMOP.
1022 * bug fix: slot-definition documentation is propagated as per ANSI
1023 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1024 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1025 argument, as do eslotd initialization methods. (from Pascal
1026 Costanza's "Closer" project)
1027 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1028 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1030 * bug fix: it is now possible to have more than one subclass of
1031 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1032 (reported by Bruno Haible, Pascal Costanza and others)
1033 * bug fix: the dependent update protocol now works for generic
1034 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1035 and Pascal Costanza)
1036 * bug fix: condition-class instances corresponding to
1037 DEFINE-CONDITION forms are now created eagerly. (reported by
1038 Kalle Olavi Niemitalo on comp.lang.lisp)
1039 * bug fix: floating point printing is more accurate in some
1040 circumstances. (thanks to Simon Alexander)
1041 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1042 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1043 * bug fix: callbacks on OS X now preserve stack-alignment.
1044 (thanks to Cyrus Harmon)
1045 * enhancement: the x86-64 disassembler is much better at
1046 disassembling SSE instructions. (thanks to Lutz Euler)
1047 * enhancement: CHECK-TYPE on a variable now allows type-inference
1048 (thanks to Vincent Arkesteijn)
1049 * optimization: improved performance of EUC-JP external format.
1050 (thanks to NIIMI Satoshi)
1051 * optimization: performance improvements to IO on file streams of
1052 :ELEMENT-TYPE CHARACTER
1053 * optimization: much faster memory allocation on x86-64
1054 * optimization: faster garbage collections (latency and throughput)
1056 * optimization: faster bignum subtraction and division on x86 and x86-64
1057 (thanks to Lutz Euler)
1058 * optimization: more accurate type-inference for loops that use a
1059 floating point index variable or a negative step.
1061 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1062 * bug fix: add a workaround to SBCL looping infinitely at startup on
1063 Linux kernels with apparently buggy implementations of personality().
1064 (thanks to Svein Ove Aas)
1065 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1066 (thanks to David Lichteblau)
1067 * bug fix: local bindings shadow global symbol macros for
1068 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1069 * optimization: non-open coded uses of numeric comparison operators
1070 (e.g. >) no longer cons when called with more than one parameter
1071 on platforms supporting dynamic-extent allocation.
1072 * enhancement: saving cores with foreign code loaded is now supported
1073 on MIPS/Linux in addition to the previously supported platforms.
1074 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1075 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1076 * bug fix: division by zero in sb-sprof when no samples were collected
1077 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1078 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1079 (thanks to Alexander Kjeldaas)
1080 * new feature: ignoring the return values of destructive functions
1081 like NREVERSE signals a compile-time style-warning.
1082 (thanks to Kevin Reid)
1083 * new feature: the alignment of alien structure fields can be
1084 explicitly specified. (thanks to Cyrus Harmon)
1085 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1086 and dump core on SIGQUIT
1088 ** incompatible change: threads do not inherit values of specials
1089 from their parents (see manual)
1090 ** bug fix: threads stacks belonging to dead threads are freed by the
1091 next exiting thread, no need to gc to collect thread stacks anymore
1092 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1093 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1094 ** bug fix: dynamic variable and thread start related gc lossage
1095 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1097 * fixed some bugs revealed by Paul Dietz' test suite:
1098 ** SUBTYPEP is slightly more accurate on heinously complicated
1099 CONS types where some of the members have uncertain (in the
1100 NIL, NIL sense) type relationships to each other.
1102 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1103 no longer purified unless :PURIFY T is explicitly specified.
1104 ** Non-purified cores are significantly smaller than before
1106 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1107 * new feature: timers based on Zach Beane's excellent timer package
1108 * added support for the following external formats: koi8-u,
1109 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1110 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1111 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1112 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1113 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1114 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1115 * incompatible change: a threaded SBCL will no longer revert to
1116 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1117 * bug fix: interrupts are disabled until startup is complete; no
1118 more sigsegvs when receiving a signal to soon
1119 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1121 * bug fix: add a workaround for the memory randomization features in
1122 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1123 workaround will only be in effect on systems with the proc filesystem
1125 * bug fix: printing objects of type HASH-TABLE signals a
1126 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1128 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1129 have been running for over 50 days. (reported by Gilbert Baumann)
1130 * bug fix: the logic for getting names of functions gets less
1131 confused when confronded with alternate-metaclass
1132 funcallable-instances. (reported by Cyrus Harmon)
1133 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1134 consistent, even on internal alternate-metaclass objects.
1135 * bug fix: flush closure information collected by physical
1136 environment analysis prepass before the main pass. (bug reported
1138 * bug fix: compiler pack phase does not modify a hashtable, which is
1139 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1140 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1141 may be expanded. (reported by Alexander <malishev> on c.l.l)
1142 * bug fix: no more style-warnings when compiling code using LAST
1143 under high-SPEED low-SPACE settings. (reported by David Wallin
1145 * fixed some incompatibilities between SBCL's MOP and the MOP
1147 ** the METAOBJECT class is now implemented;
1148 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1149 STANDARD-OBJECT, as required;
1150 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1151 now compatible (as required by VALIDATE-SUPERCLASS); there
1152 remains a consistency requirement over the presence or absence
1153 of the FUNCTION class in the superclass list of finalized
1154 classes; see the manual for more details;
1155 ** the :AROUND method for COMPUTE-SLOTS on
1156 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1157 requested slot ordering.
1159 ** bug fix: parent thread now can be gc'ed even with a live
1161 ** bug fix: binding a special with PROGV to no value is not
1162 the same as that symbol not having been bound (thanks to
1164 ** bug fix: binding specials is thread safe (thanks to
1166 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1167 and SB-SPROF do not die with 'no handler for signal XX in
1168 interrupt_handle_now(..)' anymore
1169 ** bug fix: WITH-TIMEOUT works with multiple running threads
1170 * fixed some bugs revealed by Paul Dietz' test suite:
1171 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1172 the :method-class keyword argument.
1174 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1175 * new port: the Solaris operating system on x86 processors is now
1176 mostly supported, though some rough edges in the environment
1177 remain. (thanks to Daisuke Homma)
1178 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1179 space for its heap. (thanks to Thiemo Seufer)
1180 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1181 closures. (thanks to Thiemo Seufer)
1182 * minor incompatible change: eof selects abort in the debugger.
1183 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1184 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1186 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1187 deprecated, and will go away in a future revision of SBCL.
1188 * minor incompatible change: GC-ON and GC-OFF are no longer
1189 implemented with a counter, it does not matter how many times gc
1190 is switched on or off
1191 * bug fix: discriminating functions for generic function classes
1192 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1193 make invalid assumptions about method precedence order. (reported
1195 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1196 platforms. (thanks to Thiemo Seufer)
1197 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1198 * bug fix: correct pathname printing with printer escaping is on.
1199 (thanks to Kevin Reid)
1200 * bug fix: complex VOP definitions in "user-space" no longer trigger
1201 package locks. (reported by Zach Beane)
1202 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1203 for funcallable-instance functions, and is no different from
1205 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1206 characters in the current readtable. (reported by Nicholas Neuss)
1207 * bug fix: SUBTYPEP on various CONS types returns more a more
1208 accurate acknowledgment of its certainty.
1209 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1210 lists in some cases. This partially fixes bug 384.
1211 * flush all standard streams before prompting in the REPL and the
1213 * bug fix: signal handling and triggering gc do not conflict
1214 directly anymore, in particular a high frequency sb-sprof does
1215 not prevent gc from running
1216 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1217 approximation for timezone and DST information between the
1218 universal time epoch and the smallest negative 32-bit time_t.
1219 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1220 year 1899 so long as the encoded time is non-negative.
1221 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1222 the correct answer for non-constant rotate amounts when performing
1223 an inline 32-bit rotation.
1225 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1226 there is only one thread in the session
1227 ** bug fix: memory leak for streams created in one thread and
1228 written to in another
1229 ** bug fix: lockup when compiled with gcc4
1230 ** bug fix: race that allows the gc to be triggered when gc is
1232 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1233 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1235 ** bug fix: finished threads are gc'ed properly
1236 * fixed some bugs revealed by Paul Dietz' test suite:
1237 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1238 has arguments with a different set of applicable methods from
1239 the orignal arguments.
1240 ** The type error thrown by MAP now has a correct expected-type
1242 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1243 name a compiled function.
1244 ** Three MISC tests where a large bignum was improperly coerced to
1245 a float that couldn't represent that bignum during type
1246 derivation were fixed.
1247 ** SUBTYPEP can now handle types involving the negation of a
1248 list-form FUNCTION type.
1249 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1250 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1251 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1253 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1254 * New feature: Experimental support for bivalent streams: streams
1255 opened with :element-type :default now allow character and binary
1256 (unsigned-byte 8) I/O
1257 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1258 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1260 * bug fix: no more highly sporadic "couldn't check whether ~S is
1261 readable" when reading a stream and an interrupt hits in the middle
1262 of a select system call
1263 * compiler better recognizes complex arrays (reported by Eduardo
1265 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1266 for variables on SBCL built with linkage-tables. (reported by Luis
1268 * various error reporting improvements.
1269 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1270 (Thanks to Thiemo Seufer)
1271 * bug fix: nested reader invokations work correctly
1272 * bug fix: it is possible to have simultaneous references to foreign
1273 code and foreign data with the same name.
1275 ** added x86-64 support
1276 ** incompatible change: the threading api now works with thread
1277 objects instead of thread ids
1278 ** bug fix: threads are protected from signals and interruption when
1279 starting up or going down
1280 ** bug fix: a race where an exiting thread could lose its stack to gc
1281 ** bug fix: don't halt on infinite error in threads if possible
1282 ** fixed numerous gc deadlocks introduced in the pthread merge
1283 ** bug fix: fixed thread safety issues in read and print
1284 ** bug fix: debugger doesn't hang on session lock if interrupted at
1285 an inappropriate moment
1286 ** bug fix: run-program is now thread safe(r)
1287 ** bug fix: inner with-recursive-lock no longer releases the mutex
1288 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1289 * fixed some bugs revealed by Paul Dietz' test suite:
1290 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1291 EXPECTED-TYPE slots filled.
1292 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1294 ** MAKE-CONDITION accepts classes as type-designators.
1295 ** COMPILE may never return NIL.
1296 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1297 range before calling Unix time functions
1299 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1300 * numerous signal handling fixes to increase stability
1301 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1302 * minor incompatible change: we now correctly canonize default
1303 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1304 as per the MOP, rather than the historical (INITARG INITFUNCTION
1305 INITFORM). (reported by Bruno Haible)
1306 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1307 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1308 for more information.
1309 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1310 pathname is a directory pathname.
1311 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1312 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1314 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1315 :METACLASS argument in addition to a class name. (reported by
1316 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1317 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1318 (reported by Bruno Haible; patch from Gerd Moellmann)
1319 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1321 * bug fix: more cleanups to the floating point exception handling on
1322 x86-64 (thanks to James Knight)
1323 * bug fix: the compiler does not try to constant fold calls of
1324 COERCE and BIT-* functions when they return freshly constructed
1325 objects. (reported by Nikodemus Siivola and Paul Dietz)
1326 * optimization: improved the MIPS versions of generic (in the
1327 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1328 * optimization: direct conversion of (unsigned-byte 32) to floats on
1329 the PowerPC platform.
1330 * optimization: structure instances with raw slots now use less
1331 memory, and probably show better memory locality. (thanks to
1333 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1334 are treated as requests for stack allocation on the x86-64,
1335 Alpha-32, PPC and SPARC platforms.
1336 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1337 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1338 (thanks to Tony Martinez)
1340 ** gcing a dead thread can no longer lead to lockups
1341 ** threads block signals until they are set up properly
1342 ** errno is no longer shared by threads
1343 ** interrupt-thread restores the eflags register on x86
1344 ** fixed some lockups due to gc/thread interaction
1345 * fixed some bugs revealed by Paul Dietz' test suite:
1346 ** invalid dotted lists no longer raise a read error when
1347 *READ-SUPPRESS* is T
1348 ** signal an error if a symbol that names a declaration is used
1349 as the name of a type, or vice versa
1350 ** allow using the (declare (typespec var*)) abbreviation for
1351 (declare (type typespec var*)) with all type specifiers
1352 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1353 ** always return NIL from PPRINT-POP when OBJECT is NIL
1354 ** don't signal errors when pretty-printing malformed LABELS,
1355 FLET or MACROLET forms
1356 ** declarations in a DOLIST body are also in scope for the
1358 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1360 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1363 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1364 * fixed cross-compiler leakages that prevented building a 32-bit
1365 target with a 64-bit host compiler.
1366 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1367 opened with :IF-EXISTS :OVERWRITE.
1368 * fixed bug 281: error for an invalid qualifier in a short-form method
1369 combination method is not signalled until the faulty method is called.
1370 * bug fix: iteration variable type inferrer failed to deal with open
1371 intervals. (reported by Alan Shields)
1372 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1373 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1374 or more runtime options were provided to the sbcl binary.
1375 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1376 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1377 * optimization: there's now a fast-path for fixnum arguments in the
1378 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1379 * optimization: the code generated on x86-64 is more compact thanks
1380 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1381 * bug fix: floating-point exception handling now partly works on
1382 x86-64. (Thanks to James Knight)
1383 * improvement to the MIPS backend from Thiemo Seufer: C-style
1384 64-bit long long arguments and return values to alien functions
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** the type-error signalled from WARN has a filled-in DATUM slot.
1388 ** the type-error required when a stream is not associated with
1389 a file has the stream as its datum.
1390 ** type-errors on single-floats on x86-64 no longer have
1391 :INVALID-OBJECT as the datum
1392 ** the type-errors signalled for invalid function names now have
1393 a correct expected type
1394 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1395 for structures defined with a :TYPE. Documentation strings for
1396 typed structures are no longer immediately discarded
1397 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1398 broadcast streams and synonym streams. FILE-LENGTH now also works
1399 on broadcast streams.
1401 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1402 * incompatible change: the --noprogrammer option, deprecated since
1403 version 0.7.5, has been removed. Please use the equivalent
1404 --disable-debugger option instead.
1405 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1407 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1408 inoperational for a while now) has been completely removed.
1409 * null lexical environments are now printed as #<NULL-LEXENV>,
1410 significantly reducing the amount of clutter in typical backtraces.
1411 * documentation on weak pointers, finalization, and after GC hooks
1412 has been added to the manual.
1413 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1414 as other specialized array types, is much faster. SUBSEQ and
1415 COPY-SEQ on such arrays have also been sped up.
1416 * optimization: EQL is now more efficient when at least other argument
1417 is known to be of type (OR FIXNUM (NOT NUMBER)).
1418 * fixed bug: the runtime is now less vulnerable to changes in the
1419 size of the SBCL object on OS X, and virtual memory is reserved for
1420 all spaces that need to be at a fixed address.
1421 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1422 * fixed bug: finalizers and after GC hooks that cause consing are now
1424 * fixed bug: compiler error messages and summaries are now printed to
1425 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1426 * fixed inference of the upper bound of an iteration variable.
1427 (reported by Rajat Datta).
1428 * fixed bug 211e: calling local functions with duplicated constant
1429 keyword argument no longer causes a bogus style warning about an
1431 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1432 ftype declaration to be disregarded. (reported by Dave Roberts)
1433 * fixed bug 373: caused by erronous compilation of references to alien
1434 variables in the runtime on ppc/darwin.
1435 * fixed bug 376: CONJUGATE type deriver.
1436 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1437 (reported by Baughn on #lisp)
1438 * a cross-compiler bug on non-x86 platforms has been identified and
1439 fixed. (thanks to Bruno Haible)
1440 * improvements to the MIPS runtime code for increased stability.
1441 (thanks to Thiemo Seufer)
1442 * increased the maximimum compact environment size to allow
1443 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1444 * partial workaround for bug 135: don't name the function we're
1445 calling for hairy cases of EVAL, so as not to accumulate one environment
1446 entry per call to EVAL. (reported by Kevin Reid)
1447 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1448 * optimization: type testing for non-vector arrays should be faster.
1449 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1451 * bug fix: setting 31st element of a bit vector to zero did not work
1453 * bug fix: redefining a class definition which failed due to a
1454 previous accessor / function clash now works (but see BUGS entry
1455 #380 for more problems in this area). (thanks to Zach Beane)
1456 * the long form of DEFINE-METHOD-COMBINATION disables method group
1457 checking when given a single method group with pattern *. (thanks
1459 * on x86 compiler supports stack allocation of results of simple
1460 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1461 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1462 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1463 * fixed some bugs related to Unicode integration:
1464 ** the restarts for recovering from input and output encoding
1465 errors only appear when there is in fact such an error to
1467 * fixed some bugs revealed by Paul Dietz' test suite:
1468 ** MISC.549 and similar: late transformation of unsafe type
1469 assertions into derived types caused unexpected code
1471 ** SCALE-FLOAT type deriver is less wrong.
1472 ** type derivers for EXP, LOG and similar functions compute result
1473 types for complex arguments better.
1474 ** (MISC.563) CONJUGATE type deriver works for very restricted
1476 ** out-of-line type testers for character strings are available.
1477 ** EQUAL compiler transform understands specialness of objects
1479 ** accessing double-floats stored on the stack now works on x86-64.
1480 ** debugger internals could sometimes create invalid lispobjs,
1481 resulting in GC crashes.
1482 ** MISC.548: type check weakening can convert required type into
1484 ** initialization forms for bindings are not in scope of free special
1487 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1488 * incompatible change: thread support for non-NPTL systems has
1489 been removed - locking is buggy and unreliable. A threaded
1490 SBCL build will now warn at startup and refuse to create
1491 new threads, unless futex support is detected in the kernel
1492 * incompatible change: the top level REPL now has only an ABORT
1493 restart associated with it, not TOPLEVEL and ABORT as it used to.
1494 TOP and TOPLEVEL are now available as debugger commands for
1495 returning to the top level.
1496 * incompatible change: forms evaluated in the REPL now use the
1497 global optimization policy.
1498 * incompatible change: user- and system-initialization files are
1499 no longer processed with LOAD, but by READ and EVAL; hence the
1500 global optimization policy, startup package, readtable, etc,
1502 * The .fasl file format number has been incremented because of
1503 various incompatible changes.
1504 * internal entry point details and argument counts no longer appear
1505 in backtraces unless explicitly requested by setting
1506 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1507 * built-in and standard functions no longer have names like "top
1508 level local call to FOO".
1509 * fixed bug 32: functions defined in non-null lexical environments
1510 now have more legible printed representation
1511 * fixed bug 33: functions defined in non-null lexical environemnts
1512 are now more amenable to inspection by INSPECT.
1513 * workaround for bug 354: XEPs no longer appear in backtraces unless
1514 explicitly requested.
1515 * fixed bug: receiving the signal which results from attempting to
1516 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1517 SIGBUS on other *BSDs) on architectures where the C stack is also
1518 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1519 to debugger confusion or wild instability. (reported by Cheuksan
1521 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1522 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1523 (reported by Lutz Euler)
1524 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1525 compiled code. (reported by Paul Dietz)
1526 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1527 specialized parameter to a method to something that is not TYPEP
1528 the specializer is now possible.
1529 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1530 face of package deletion.
1531 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1532 pathname. (reported by David Tolpin on c.l.l)
1533 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1534 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1535 STRING-STREAM and FILE-STREAM, respectively.
1536 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1537 than previously; in addition, its test suite is now run on build.
1538 (thanks to Luke Gorrie)
1539 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1540 * optimization: added a immediate representation for single-floats
1542 * fixed some bugs revealed by Paul Dietz' test suite:
1543 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1544 ** attempting to create a package with a colliding nickname causes
1545 correctable errors to be signalled.
1546 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1547 ** several x86-64 backend bugs related to sign-extension of immediate
1550 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1551 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1552 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1554 * fixed loading of multiply forward-referenced layouts.
1555 (thanks to Cheuksan Wang)
1556 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1557 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1558 (see COMPILE-FILE documentation for details of :PRINT option)
1559 * fixed bugs 19 and 317: fixed-format floating point printing is
1560 more accurate. This also fixes a bug reported by Adam Warner
1561 related to the ~@F format directive.
1562 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1564 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1565 dispatch table if the from-char is a dispatch macro character.
1566 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1567 by the wordsize no longer ignore the last word. (reported by Lutz
1569 * fixed bug in type checking of dynamic-extent variables. (reported
1571 * optimization: sequence traversal functions use their freedom to
1572 coerce function designators to functions.
1573 * optimization: code with many calls to CLOS methods specialized on
1574 CLOS classes has had redundant type checks removed.
1575 * optimization: added declarations to speed up operations that access
1576 the internal character database (for example STRING-UPCASE)
1577 * optimization: comparison operations between floats and sufficiently small
1578 fixnums no longer create extra rationals
1579 * fixed some bugs related to Unicode integration:
1580 ** portions of multibyte characters at the end of buffers for
1581 character-based file input are correctly transferred to the
1582 start of the buffer at the next read.
1583 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1584 passing it through to OPEN.
1585 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1586 argument given to internal calls to OPEN.
1587 ** the built-in comment readers (introduced by character sequences
1588 ";" and "#|") are more forgiving to encoding errors; they will
1589 STYLE-WARN and then attempt to resync the stream at a character
1590 boundary. (thanks to Teemu Kalvas)
1591 * fixed some bugs in the x86-64 port:
1592 ** Negative short int return values from c-calls are sign-extended
1594 ** The stack is aligned to 16-bytes for c-calls, as required by
1595 the ABI. (thanks to Cheuksan Wang)
1596 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1597 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1598 for enums. (thanks to Vincent Arkesteijn)
1599 ** Multiple small optimizations and bugfixes for floating point
1601 * fixed some bugs revealed by Paul Dietz' test suite:
1602 ** Space, Tab, Linefeed, Return and Page have the invalid
1603 secondary constituent character trait.
1604 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1606 ** WITH-INPUT-FROM-STRING should only update the index place on
1608 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1609 works more reliably.
1610 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1611 with one-dimensional arrays with fill pointers.
1612 ** TYPE-OF failed on a complex with an integer realpart and a
1614 ** compiler failure during type inference for the code of form
1615 (IF (EQL X (THE ...)) ...) (MISC.535).
1617 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1618 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1619 platform. The port supports 61-bit fixnums, large memory spaces
1620 and reloading shared object files.
1621 * enhancement: saving cores with foreign code loaded is now
1622 supported on ppc/Darwin in addition to the previously supported
1624 * enhancement: the statistical profiler now walks deeper into the
1625 call stack for more meaningful call-graphs and accrued time
1626 reports (x86/x86-64 only). It also now reports time spent in
1628 * enhancement: it is now possible to trace most individual methods
1629 of a generic function in addition to tracing the generic function
1631 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1633 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1634 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1635 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1636 produces an error. (thanks to Vincent Arkesteijn)
1637 * bug fix: NAMESTRING on pathnames with :WILD components in their
1638 directories works correctly. (thanks to Artem V. Andreev)
1639 * fixed bug 125: compiler preserves identity of closures. (reported
1641 * bug fixed: functions with &REST arguments sometimes failed with
1642 "Undefined function" when compiled with (DEBUG 3). (reported by
1644 * bug fix: overflow during compiling of setting element of a bit
1645 vector with constant index and value. (reported by Timmy Douglas)
1646 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1647 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1648 * on x86 compiler partially supports stack allocation of dynamic-extent
1650 * GO and RETURN-FROM do not check the extent of their exit points
1651 when compiled with SAFETY 0.
1652 * fixed some bugs related to Unicode integration:
1653 ** encoding and decoding errors are now much more robustly
1654 handled; it should now be possible to recover even from invalid
1655 input or output to the terminal. (thanks to Teemu Kalvas)
1656 ** provided a first cut at implementing STRING-TO-OCTETS and
1657 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1658 ** altered the SB-MD5 contributed module slightly, changing the
1659 interface just enough for it to be supportable for builds where
1660 lisp characters are not eight bits.
1661 * fixed some bugs revealed by Paul Dietz' test suite:
1662 ** the FORMATTER-generated functions for ~V[ conditionals require
1663 the correct number of arguments.
1664 ** READ-FROM-STRING returns the mandated second value when applied
1665 to displaced strings.
1666 ** the #\Rubout and #\Backspace characters are treated as invalid
1667 constituent characters by the tokenizer.
1669 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1670 * new feature: reloading changed shared object files with
1671 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1672 * new feature: references to foreign variables and functions
1673 can now be compiled and loaded before the corresponding shared
1674 object file is loaded, as long as the foreign definitions are
1675 available at runtime.
1676 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1677 supported. (thanks to Dan Debertin)
1678 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1679 just keywords. (thanks to Vincent Arkesteijn)
1680 * fixed bug #331: structure-class instances corresponding to
1681 DEFSTRUCT forms are now created eagerly.
1682 * fixed bug #345: backtraces from calls to undefined functions work
1683 on x86 as well. Related bug #61 is now also partially fixed on x86
1684 (backtraces from throws to unknown catch tags.)
1685 * bug fix: lambda-list parsing is now stricter vrt. order and number
1686 of lambda-list keywords.
1687 * bug fix: as specified by AMOP, an error is signalled if a
1688 class-option appears multiple times in a DEFCLASS form. (reported
1690 * bug fix: RANDOM can be compiled when the compiler derives the type
1691 of its numeric argument as a disjoint set of small integers.
1692 (reported by Paul Dietz)
1693 * bug fix: starting a core saved with shared objects loaded when
1694 those objects are not available no longer causes threaded SBCL to
1695 hang. (reported by Sean Ross)
1696 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1698 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1699 argument, which must always be NIL. (reported by Kalle Niemitalo)
1700 * bug fix: printing 1.0d+23 no longer results in an error.
1701 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1702 * bug fix: structure slot setters preserve evaluation order. (thanks
1704 * fixed some bugs related to Unicode integration:
1705 ** RUN-PROGRAM can allow its child to take input from a Lisp
1706 stream. (reported by Stefan Scholl)
1707 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1708 locales. (reported by Ken Causey)
1709 * fixed some bugs revealed by Paul Dietz' test suite:
1710 ** INCF, DECF and REMF evaluate their place form as specified in
1712 ** FORMATTER expands ~{ ~} iteration directives with V or #
1713 parameters correctly.
1714 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1715 consequent uses no arguments correctly.
1716 ** the system has a partial understanding of the (COMPLEX RATIO)
1718 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1721 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1722 * new feature: a build-time option (controlled by the :SB-UNICODE
1723 keyword feature, enabled by default) for building the system with
1724 support for the entire 21-bit character space defined by the
1726 * new feature: the system now has rudimentary external-format
1727 support; the primary user-visible change at this time is that
1728 characters with the high bit set (such as lower-case-e-acute) will
1729 print correctly to a terminal in a UTF-8 environment.
1730 * minor incompatible change: BASE-CHAR no longer names a class;
1731 however, CHARACTER continues to do so, as required by ANSI.
1732 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1733 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1734 has been moved to the SB-EXT package (temporarily re-exported from
1736 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1737 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1739 * fixed bug #351: better error-handlind and reporting for malformed
1741 * fixed bug #350: bignum-printing is now more memory-efficient,
1742 allowing printing of very large bignums, eg. (expt 2 10000000).
1743 (reported by Bruno Haible)
1744 * fixed bug #302: better primitive-type selection for intersection
1746 * fixed bug #308: non-graphic characters now all have names, as
1747 required. (reported by Bruno Haible)
1748 * bug fix: redefining a class with different superclasses now correctly
1749 removes it from the direct-subclasses of its previous superclasses.
1750 (reported by David Morse)
1751 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1752 new value now works. (reported by Bruno Haible)
1753 * bug fix: correct canonicalization of multiple non-standard slot
1754 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1755 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1756 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1757 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1758 now exists, an signals an error.
1759 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1760 during macroexpansion for non-iteration variables as well. (reported
1761 by Bruno Haible for CMUCL)
1762 * bug fix: Cyclic structures and unprintable objects in compiler
1763 messages no longer cause errors. (reported by Bruno Haible)
1764 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1765 and READ-FROM-STRING all now return a primary value of NIL if
1766 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1767 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1768 (reported by Bruno Haible for CMUCL)
1769 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1770 do not designate a condition. (reported by Bruno Haible for
1772 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1773 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1774 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1775 permitting resolution in favour of any of the conflicting symbols.
1776 (reported by Bruno Haible for CMUCL)
1777 * bug fix: EQUAL compiler optimizations is less aggressive on
1778 strings which can potentially compare true despite having distinct
1779 specialized array element types.
1780 * bug fix: unit enumerations can be defined without dividing by
1781 zero. (thanks to Vincent Arkesteijn)
1782 * FORMAT compile-time argument count checking has been enhanced.
1783 (report from Bruno Haible for CMUCL)
1784 * a partial workaround for the bug 262: the compiler does not try to
1785 inline-expand a local function doing RETURN-FROM from a deleted
1786 BLOCK. (thanks to Peter Denno for the bug report and to David
1787 Wragg for the simple test case)
1788 * fixed some bugs revealed by Paul Dietz' test suite:
1789 ** RENAME-PACKAGE allows all package designators as new package
1791 ** constraint propagation and lambda variable substitution are
1792 more cautious in dealing with partially deleted code.
1793 ** compiler sometimes lost reoptimization passes.
1794 ** CERROR, given a condition as condition designator, passes its
1795 remaining arguments to the continue format control without
1797 ** Case-altering FORMAT directives work correctly on non-ASCII
1799 ** The REAL type specifier handles bounds outside the floating
1800 point ranges without signalling FLOATING-POINT-OVERFLOW.
1801 ** Functions with IR1-transformations can create intercomponent
1802 references to global functions.
1803 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1805 ** FORMAT ~R treats a nil value for its first parameter correctly.
1807 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1808 * enhancement: saving cores with foreign code loaded is now
1809 supported on x86/NetBSD and sparc/Linux in addition to the previously
1810 supported platforms.
1811 * bug fix: on some platforms repeated installations caused multiple
1812 copies of HTML documentation to be installed -- should not happen
1813 any more. (reported by Stefan Scholl)
1814 * bug fix: parsing self-recursive alien record types multiple times
1815 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1816 original patch by Helmut Eller for CMUCL)
1817 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1818 (thanks to Richard Kreuter)
1819 * bug fix: defining classes whose accessors are methods on existing
1820 generic functions in other (locked) packages no longer signals
1821 bogus package lock violations. (reported by François-René Rideau)
1822 * bug fix: special variables as DEFMETHOD parameters no longer have
1823 associated bogus type declarations. (reported by David Wragg and
1825 * bug fix: read-write consistency on streams of element-type
1826 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1827 * bug fix: redefiniton of the only method of a generic function with
1828 no DEFGENERIC no longer emits a full WARNING. In addition,
1829 redefinition of generic functions with no DEFGENERIC to an
1830 incompatible lambda list now signals an error. (thanks to Zach
1832 * bug fix: DEFGENERIC now works even when there's a function of the
1833 same name in an enclosing lexical environment. (thanks to Zach
1835 * fixed compiler failure, caused by instrumenting code during
1836 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1837 * optimization: added loop analysis and improved register allocation
1838 * fixed some bugs revealed by Paul Dietz' test suite:
1839 ** POSITION on displaced vectors with non-zero displacement
1840 returns the right answer.
1841 ** (SIMPLE-STRING) is a valid type specifier for sequence
1843 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1845 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1846 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1848 ** The justification version of the FORMAT ~< directive treats
1849 non-zero minpad parameter correctly.
1851 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1852 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1853 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1854 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1855 the supported interface.
1856 * new feature: Single-stepping of code compiled with DEBUG 2 or
1857 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1858 * new feature: saving cores with foreign code loaded is now
1859 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1860 Timothy Moore's work for CMUCL)
1861 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1862 parameters to * instead of NIL if no initform is supplied.
1863 (reported by Johan Bockgaard)
1864 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1865 correctly when FUNCALL appears as the car of the form. Note:
1866 despite this FUNCALL forms are not currently subject to
1867 compiler-macro expansion. (port of Raymond Toy's fix for the
1868 same from CMUCL, reported by Johan Bockgaard)
1869 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1870 (thanks for Teemu Kalvas)
1871 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1872 case instead of terminating the iteration (thanks for Julian
1873 Squires, Sean Champ and Raymond Toy)
1874 * bug fix: incorrect expansion of defgeneric that caused a style
1875 warning. (thanks for Zach Beane)
1876 * bug fix: slot accessor effective method computation works properly
1877 for classes with multiple non-standard applicable methods on
1878 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1879 * on x86 compiler supports stack allocation of results of LIST and
1880 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1881 CMUCL implementation by Gerd Moellmann)
1882 * fixed some bugs revealed by Paul Dietz' test suite:
1883 ** FORMAT strings with both the ~<~:;~> form of the justification
1884 directive and pretty-printing directives cause an error.
1886 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1887 * incompatible change: the internal functions
1888 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1889 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1890 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1891 instead of the old functions.
1892 * new feature: on platforms where "dladdr" is available foreign
1893 function names now appear in backtraces. (based on Helmut Eller's
1895 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1896 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1898 * documentation: ASDF manual is now distributed with SBCL.
1899 * bug fix: SBCL can now load its contributed modules with REQUIRE
1900 even if the system-provided entries have been removed from
1901 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1903 * bug fix: backtraces involving undefined functions or assembly
1904 routines are more informative. (thanks to Brian Downing)
1905 * bug fix: mutually referent alien structures now work correctly.
1906 (reported by Rick Taube)
1907 * bug fix: structures defined by WITH-ALIEN can be referred to
1908 within other definitions in the same WITH-ALIEN.
1909 * bug fix: division operators (MOD, TRUNCATE and the like) with
1910 constant zero divisors and integer dividends no longer generate
1912 * bug fix: provide default methods for INPUT-STREAM-P and
1913 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1914 * bug fix: improve the detection and resolution of MOP metacycles
1915 (where effective method computation depends on the generic function
1916 having its effective method computed). (reported by Bruno Haible)
1917 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1918 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1919 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1920 * bug fix: pathnames with relative directory components can be
1921 represented relative to default pathnames.
1922 * optimization: in taking the GCD of bignums, reduce the two bignums
1923 to approximately the same size (using Euclid's algorithm) before
1924 applying the more sophisticated binary GCD. (thanks to Juho
1926 * optimization: COUNT on bitvectors now operates word-at-a-time.
1927 * optimization: ASH with a positive, but not necessarily constant,
1928 (leftwards) shift, when in a modular context, is compiled to a
1930 * fixed some bugs revealed by Paul Dietz' test suite:
1931 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1932 if the corresponding argument is NIL.
1933 ** FORMAT directives accepting a minpad parameter treat negative
1935 ** literal commas as character parameters to FORMAT directives are
1937 ** literal spaces directly after ~<Newline> directives within a
1938 format-logical-block (~:< ~@:>) do not induce :FILL-style
1939 conditional newlines.
1940 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1941 as specified: it no longer includes conditional newlines).
1942 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1944 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1945 :PER-LINE-PREFIX argument does not evaluate to a string.
1947 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1948 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1949 the manual for details; for now, package locks can be disabled by
1950 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1951 if no major problems are found then it is likely that they will be
1952 compiled in unconditionally.
1953 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1954 unsupported operators on all platforms. To load a shared library
1955 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1956 object file, link it into a shared library outside of SBCL and
1957 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1959 * bug workaround: ROOM T and the GC were not working together
1960 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1961 reporting facility aren't preserved. That reporting has been
1962 disabled, so now until and unless someone figures out how to make it
1963 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1964 * minor incompatible change: as threatened around sbcl-0.8.0, the
1965 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1966 The symbols are also exported from SB-PCL for backwards
1967 compatibility, but more so than before SB-PCL should be treated as
1968 an implementation-internal package.
1969 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1971 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1972 It is known as of this release not to work on the Alpha, however.
1973 * fixed bug #167: errors signalled due to illegal syntax in method
1974 bodies are now more legible.
1975 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1976 designators and can hence be used with TYPEP.
1977 * fixed bug #333: CHECK-TYPE now ensures that the type error
1978 signalled, if any, has the right object to be accessed by
1979 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1980 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1981 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1983 * fixed bug #334: programmatic addition of slots using specialized
1984 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1985 and :ALLOCATION :CLASS slots.
1986 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1987 float exponents. (rereported by Peter Seibel)
1988 * fixed a bug: #\Space (and other whitespace characters) are no
1989 longer considered to be macro characters in standard syntax by
1990 GET-MACRO-CHARACTER.
1991 * fixed bug: initialization of condition class metaobjects no longer
1992 causes an instance of the condition to be created. (reported by
1994 * fixed bug: it is now possible to have slots such that
1995 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1996 description is neither :INSTANCE nor :CLASS.
1997 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1998 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1999 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2000 non-standard SLOT-DEFINITION-ALLOCATION values.
2001 * fixed bugs: various race conditions handling exiting threads.
2002 CL-PPCRE's thread torture test now passes.
2003 * fixed arguably-a-bug: GC time proportional to number of threads in
2004 system even when most of them are idle
2005 * optimization: improved performance of BIT and SBIT on bit-vectors.
2006 * .fasl file incompatibility: The fasl file version number has
2007 been incremented because of changes associated with package locks.
2009 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2010 * minor incompatible change: the system no longer provides
2011 optimization hints (conditions which are TYPEP
2012 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2013 mode; these hints are still emitted when higher SPEED optimization
2015 * new contrib module: a sampling profiler (profiling by statistical
2016 sampling, rather than by instrumenting functions) is available as
2017 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2018 * the behaviour of the standard function ED is now customizeable by
2019 third parties through a hook variable: see ED's documentation
2020 string for information on the protocol.
2021 * the compiler no longer emits efficiency notes for (FUNCALL X)
2022 when the type of X is uncertain under default optimization
2024 * fixed bug 276: mutating a binding of a specialized parameter to a
2025 method to something that is not TYPEP the specializer is now
2027 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2028 exhibit the required behaviour on the x86 platform. (thanks to
2029 Peter van Eynde, Eric Marsden and Bruno Haible)
2030 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2031 even for difficult arguments. (reported by Peter Graves)
2032 * fixed bug 141a: the backquote printer now descends quoted
2034 * fixed another bug in backquote printing: no more destructive
2035 modification of the form's list structure. (reported by Brian
2037 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2038 move between its address being taken and the call to
2039 interrupt_thread, fixing a crashing race condition.
2040 * the SB-POSIX contrib implementation has been adjusted so that it
2041 no longer exhibits ridiculously poor performance when constructing
2042 instances corresponding to C structs.
2044 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2045 * minor incompatible change: the sb-grovel contrib now treats C
2046 structures as alien (in the sense of SB-ALIEN) objects rather than
2047 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2048 has implications for memory management of client code
2049 (sb-grovel-returned objects must now be manually managed) and for
2050 type safety (alien objects now have full types).
2051 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2052 used to control emission of compiler diagnostics, rather than the
2053 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2054 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2055 quality should be considered deprecated.
2056 * install.sh now installs the user manual as well
2057 * (not quite a new documentable feature, but worth considering in
2058 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2059 beginnings of a semantically meaningful condition hierarchy is
2060 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2062 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2063 designator as the defaults argument.
2064 * fixed bug: Displaced arrays whose displaced-to array has become
2065 too small now cause ARRAY-DIMENSION to signal an error, providing
2066 for safer bounds-checking. (reported by Bruno Haible)
2067 * fixed bug: DEFCLASS slot definitions with identical :READER and
2068 :WRITER names now signal a reasonable error. (reported by Thomas
2070 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2072 * fixed bug: Invalid slot specification errors now print correctly.
2073 (thanks to Zach Beane)
2074 * fixed bug 320: Shared to local slot value transfers in class
2075 redefinitions now happen corrently from superclasses as
2076 well. (reported by Bruno Haible)
2077 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2079 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2080 improper lists correctly. (reported by Bruno Haible)
2081 * fixed bug 313: source-transform for <fun-name> was erroneously
2082 applied to a call of a value of a variable with name <fun-name>.
2083 (reported by Antonio Menezes Leitao)
2084 * fixed bug 307: The obsolete instance protocol ensures that
2085 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2086 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2087 remove dead unknown-values globs from the middle of the stack.
2088 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2089 easier to resume long computations after using *BREAK-ON-SIGNALS*
2090 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2091 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2092 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2093 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2094 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2095 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2097 * fixed a (fairly theoretical) bug in string printing: if
2098 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2099 does not have array-element-type equal to the most general string
2101 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2102 function name. (thanks to Nikodemus Siivola)
2103 * fixed bug: compile-time format string checker failed on
2104 non-closed ~{. (reported by Thomas F Burdick)
2105 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2106 don't warn on legal loop code involving more than one aggregate
2107 boolean. (thanks to Nikodemus Siivola)
2108 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2109 in keyword position to MAKE-PACKAGE was not being reported as
2111 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2112 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2114 * fixed bugs 280 and 312: the checking for multiple definitions in a
2115 file is less likely to become confused by uses of inline
2117 * fixed bug: the #S reader macro performs the keyword coercion
2118 specified for slot names. (reported by Kalle Niemitalo)
2119 * fixed bug: lambda lists may contain symbols whose names start with
2120 & but are not lambda-list-keywords; their occurrence triggers a
2122 * fixed bug 321: define-method-combination argument lambda lists do
2123 not cause invalid code to be generated when &optional or &aux
2124 variables have default values. (reported by Bruno Haible)
2125 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2126 have CLOS classes; this appears also to have fixed reported
2127 instability in stack exhaustion detection.
2128 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2129 MUFFLE-WARNING when no associated restart is present is now
2131 * optimization: rearranged the expansion of various defining macros
2132 so that each expands into only one top-level form in a
2133 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2135 * optimization: used a previously little-used slot in symbols to
2136 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2138 * fixed some bugs revealed by Paul Dietz' test suite:
2139 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2141 ** Fixed an optimization bug related to inheritance of initforms
2142 from local to shared slots.
2143 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2144 ** CAST optimizer forgot to flush argument derived type.
2145 ** print/read consistency on floats is now orders of magnitude
2146 more likely. (thanks also to Bruno Haible for a similar report
2148 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2149 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2150 for objects of type REAL. Make it so.
2151 ** SXHASH is specified (infelicitously) to respect similarity,
2152 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2153 it so. (thanks to Markus Ziegler)
2154 ** on the Alpha, the compiler succeeds in compiling functions
2155 returning a known number of arguments greater than 63.
2156 ** fixed handling of invalid NIL arguments in keyword position
2158 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2160 ** provided a readably-printable representation for RANDOM-STATE
2162 ** ensured that pathnames, where they have a namestring, always
2163 print using #P"..." syntax.
2165 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2166 * Support for the forthcoming 2.0 version of the NetBSD kernel
2167 running on x86 hardware has been added. (thanks to Perry
2168 E. Metzger most immediately, and others for their past work)
2169 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2170 current release version; SBCL's OpenBSD support had been broken
2171 since about the time of OpenBSD's switch to ELF binary format.)
2172 * [placeholder for DX summary]
2173 ** user code with &REST lists declared dynamic-extent, under high
2174 speed or space and low safety and debug optimization policy.
2175 * The manual has been converted to Texinfo format and the debugger
2176 chapter from the cmucl manual has been added.
2177 * A facility has been added to extract documentation strings from
2178 sbcl and store them as Texinfo-formatted snippets for inclusion in
2179 the manual (via Texinfo's @include directive)
2180 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2181 the test case to Dave Roberts)
2182 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2183 pointers. (reported by Sean Ross)
2184 * bug fix: PROFILE output is printed nicely even for large numerical
2185 values. (thanks to Zach Beane)
2186 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2187 greater than 32 handle EOF correctly.
2188 * bug fix: on X86 an immediate argument of the IMUL instruction is
2189 correctly printed in disassembly. (thanks to Lutz Euler)
2190 * bug fix: class slots in redefined classes preserve their old
2191 values. (thanks to Bruno Haible and Nikodemus Siivola)
2192 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2193 succeeds. (reported by Marco Baringer)
2194 * fixed some bugs revealed by Paul Dietz' test suite:
2195 ** READ-SEQUENCE now works on ECHO-STREAMs.
2196 ** RATIONALIZE works more according to its specification. (thanks
2199 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2200 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2201 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2202 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2203 should matter to you only if you rebind the printer control
2204 variables and then find you want different bindings in the
2205 debugger than in the ordinary execution of your program.)
2206 * The runtime build system has been tweaked to support building
2207 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2208 assembler and linker. (thanks to Nikodemus Siivola)
2209 * Unbound, undefined, undeclared variables now trigger full
2210 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2211 more often programmer error than deliberate exploitation of undefined
2213 * optimization: the hash algorithm for strings has changed to one
2214 that is less vulnerable to spurious collisions. (thanks to Juho
2216 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2217 needless bounds checking. (thanks to Juho Snellman)
2218 * optimization: implemented multiplication as a modular
2219 (UNSIGNED-BYTE 32) operation on the PPC backend.
2220 * fixed some bugs revealed by Paul Dietz' test suite:
2221 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2223 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2224 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2225 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2227 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2229 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2231 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2232 their output stream on EOF from read.
2233 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2234 have been read to end-of-file.
2235 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2237 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2238 description of determination of which consecutive characters
2240 ** Printing the "Space" character with escaping on now yields "#\\ ",
2241 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2242 ** Reading floating-point numbers with *READ-BASE* set to a number
2243 less than 10 works correctly.
2244 ** Reading floating-point numbers with *READ-BASE* set to a number
2245 more than 10 works correctly.
2246 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2247 the readtable currently in effect.
2249 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2250 * minor incompatible change: parsing of namestrings on a physical
2251 (Unix) host has changed; numbers after the final #\. in a
2252 namestring are no longer interpreted as a version field. This is
2253 intented to be largely invisible to the user, except that the
2254 meaning of the namestring "*.*.*" has changed: it now refers to a
2255 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2256 should usually be replaced by
2257 (make-pathname :name :wild :type :wild :version :wild)
2258 with the added benefit that this is more likely to be portable.
2259 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2260 option to OPEN now signals an error if the file being opened
2261 exists; this may have an impact on existing code.
2262 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2263 signals received, on the PowerPC platforms (both Linux and
2264 Darwin). (thanks to Pierre Mai for pointing out the location of
2266 * several fixes on OS X: The system now builds and runs cleanly on
2267 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2268 (thanks to Brian Mastenbrook)
2269 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2270 second-resolution: integer multiples of 1/3600 between -24 and 24.
2271 (thanks to Vincent Arkesteijn)
2272 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2273 type when called with 1 argument; PEEK-CHAR checked type of
2274 PEEK-TYPE only after having read first character from a
2275 stream. (reported by Peter Graves)
2276 * bug fix: the garbage collector now has much better locality
2277 behaviour, and in particular no longer treats all memory as being
2278 exhausted when many small objects point to each other in a deeply
2280 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2281 recognized as being TYPEP their class.
2282 * bug fix: the PUSHNEW documentation string has been corrected.
2283 (thanks to Vincent Arkesteijn)
2284 * bug fix: defaulting of the value for the last of an atypically
2285 large number of multiple values being bound was not being
2286 performed correctly on the Alpha or PPC platforms
2287 * optimization: implemented multiplication as a modular
2288 (UNSIGNED-BYTE 32) operation on the x86 backend.
2289 * optimization: SEARCH on simple-base-strings can now be open-coded.
2290 (see also contrib/compiler-extras.lisp for inspiration for
2291 teaching the compiler about the Boyer-Moore algorithm).
2292 * value, returned by MAX (and MIN) called with several EQUALP, but
2293 not EQL, arguments now does not depend on compiler settings.
2294 * fixed some bugs revealed by Paul Dietz' test suite:
2295 ** in stack analysis liveness information is propagated from
2296 non-local entry points.
2297 ** pathwise CAST removing failed when the CAST node did not start
2299 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2300 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2302 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2303 host is already defined.
2304 ** RENAME-FILE works on streams instead of signalling an internal
2306 ** PEEK-CHAR uses the current readtable when determining whether
2307 or not a character is whitespace.
2308 ** MERGE-PATHNAMES handles the case when the pathname does not
2309 specify a name while the default-pathname specifies a version.
2310 ** Pathnames now stand a better chance of respecting print/read
2312 ** Attempting to use standardized file system operators with a
2313 pathname with invalid :DIRECTORY components signals a
2315 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2317 ** DIRECTORY on logical pathnames is more correct.
2318 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2319 signal a TYPE-ERROR if their argument is not a stream.
2320 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2321 designator argument does not designate a stream.
2322 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2323 examining the synonym.
2324 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2326 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2327 element-type larger than ([UN]SIGNED-BYTE 32).
2329 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2330 * When built with the :SB-FUTEX feature, threaded builds now take
2331 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2332 for faster/more reliable mutex and condition variable support.
2333 * Incompatible change (but one you probably shouldn't have been using
2334 anyway): the interface and code for arbitrating between multiple
2335 threads in the same user session has been redesigned.
2336 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2337 error when called without an explicit environment argument.
2338 (thanks to Nikodemus Siivola)
2339 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2340 confused about their position. (thanks to Adam Warner and Gerd
2342 * bug fix: Pretty printing backquoted forms with unquotations in the
2343 argument list position of various code constructs such as LAMBDA
2344 now works correctly. (reported by Paul Dietz)
2345 * bug fix: Pretty printing unquotations no longer loses all
2346 stream position information.
2347 * optimization: performance of string output streams is now less
2348 poor for multiple small sequence writes.
2349 * optimization: performance of CSUBTYPEP in the presence of complex
2350 expressions involving CONS and NOT many times has been improved.
2351 (reported by Paul Dietz)
2352 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2354 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2355 streams for the REPL, for the debugger, and so on. (thanks to
2357 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2358 * fixed some bugs revealed by Paul Dietz' test suite:
2359 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2360 is ignored if it is too small, rather than propagating through
2361 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2362 ** extremely complex negations of CONS types were not being
2363 sufficiently canonicalized, leading to inconsistencies in
2365 ** VALUES tranformer lost derived type.
2367 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2368 * fixed a bootstrapping bug: the build process no longer assumes
2369 that the various BOOLE-related constants have the same value in
2370 host and target lisps. (noted by Paul Dietz' test suite on an
2371 SBCL binary built from CLISP)
2372 * The system can now be dynamically linked on the MIPS platform,
2373 which enables dynamic loading of foreign code from Lisp. (thanks
2374 to Ralf Baechle for discussions on the MIPS ABI)
2375 * The system now records debugging information for its own source
2376 files in a filesystem-position-independent manner, relative to
2377 the "SYS" logical host.
2378 * fixed a compiler bug: MV-LET convertion did not check references
2379 to the "max args" entry point. (reported by Brian Downing)
2380 * tweaked disassembly notes to be less confident about proclaiming
2381 some instruction as an LRA. (thanks to Brian Downing)
2382 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2383 now each have their own history, command character, and other
2384 characteristics. (thanks to David Lichteblau)
2385 * fixed some bugs revealed by Paul Dietz' test suite:
2386 ** compiler failure in compiling LOGAND expressions including a
2388 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2389 shift greater than 32.
2390 ** FLUSH-DEST did not mark blocks for type check regeneration.
2391 ** HANDLER-CASE failed to accept declarations in handler clauses
2392 in some circumstances.
2394 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2395 * New code in contrib/sb-introspect (still probably not entirely
2396 stable yet) provides some support for smart Lisp development
2397 environments like SLIME.
2398 * The conditions signalled for errors occurring when loading .fasl
2399 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2400 in a way which should help ASDF recover gracefully.
2401 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2402 stuff has been cleaned up. If you code contrib/ stuff, this might
2403 affect you, and you can look at contrib/README, contrib/STANDARDS,
2404 and/or the 0.8.4.27 diff to check.
2405 * In full calls the compiler now does not generate checks for declared
2406 argument types for all arguments.
2407 * various threading fixes
2408 ** and some experimental patches which didn't make it into
2409 the main tree for this release, but which are shipped in
2410 contrib/experimental-thread.patch as a possible fix for some
2411 failures (deadlock, spinning...) in GC-intensive multithreaded
2413 * fixed PPC build problem (source code incompatibility of different
2414 library versions): added offsetof() hackery which attempts to divine
2415 where glibc maintainers put uc_mcontext today
2416 * fixed bug 282: compiler does not trust type assertions while passing
2417 arguments to a full call.
2418 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2419 &OPTIONAL VALUES type parameter.
2420 * fix bug 214: algorithm for noting rejected templates is now more
2421 similar to that of template seletion. (also reported by rydis on
2423 * fixed bug 141b: printing backquoted information readably and prettily
2424 inserts a space where necessary.
2425 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2426 (CHARACTER 10) are now reported as errors, rather than propagated
2427 as unknown types. (reported by piso on #lisp)
2428 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2429 with values NIL and :ERROR. (thanks to Milan Zamazal)
2430 * fixed bug 191c: CLOS now does proper keyword argument checking as
2431 described in CLHS 7.6.5 and 7.6.5.1.
2432 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2433 longer raise an error; further, using a list as a for-as-arithmetic
2434 counter now raises a meaningful error.
2435 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2436 now understood by sequence creation functions such as MAKE-SEQUENCE
2438 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2439 a STRING-INPUT-STREAM.
2440 * compiler enhancement: SIGNUM is now better able to derive the type
2442 * type declarations inside WITH-SLOTS are checked. (reported by
2444 * fixed some bugs revealed by Paul Dietz' test suite:
2445 ** incorrect optimization of TRUNCATE for a positive first
2446 argument and negative second.
2447 ** compiler failure in let-convertion during flushing dead code.
2448 ** compiler failure while deriving type of TRUNCATE on an
2449 interval, containing 0.
2450 ** ASH of a negative bignum by a negative bignum count now returns
2452 ** intersection of CONS types now canonicalizes properly, fixing
2453 inconsistencies in SUBTYPEP.
2455 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2456 * incompatible change: The --disable-debugger command line
2457 option now clobbers the debugger at a more fundamental
2458 level, by redefining #'INVOKE-DEBUGGER instead of by
2459 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2460 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2461 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2462 enter the debugger REPL and then suffer recursive errors
2463 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2464 while under the new system, BREAK in --disable-debugger mode
2465 terminates the system just as an unhandled error would.
2466 * fixed compiler performance when processing loops with a step >1;
2467 * bug fix: DOCUMENTATION now retrieves generic function
2468 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2469 support has been systematized, and now supports the methods
2470 specified by ANSI, along with a default method and a method for
2471 slot documentation. (reported by Nathan Froyd)
2472 * bug fix: effective methods associated with a generic function are
2473 no longer cached over a change of that generic function's method
2474 combination. (reported by Andreas Fuchs)
2475 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2476 names a non-existent directory. (thanks to Andreas Fuchs)
2477 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2478 the PPC platform, lead to stack corruption; this has been fixed.
2479 (reported by Rainer Joswig)
2480 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2481 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2482 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2483 arithmetic in general, is now much more reliable on the PPC
2485 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2486 platform now returns the right answer.
2487 * optimization: restored some effective method precomputation in
2488 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2489 precomputation is now tunable.
2490 * optimization: compiler-internal data structure use has been
2491 reviewed, and changes have been made that should improve the
2492 performance of the compiler by about 20%.
2493 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2494 simple-base-strings and simple-bit-vectors is improved.
2495 * optimization: the optimization of 32-bit logical and arithmetic
2496 functions introduced in version 0.8.3 on the x86 has been
2497 implemented on the mips, ppc and sparc platforms; an
2498 implementation of the same facility, but for 64-bit arithmetic,
2499 has been added for the alpha.
2500 * microoptimization: the compiler is better able to make use of the
2501 x86 LEA instruction for multiplication by constants.
2502 * bug fix: in some situations compiler did not report usage of
2503 generic arithmetic in (SPEED 3) policy.
2504 * bug 145b fix: compiler used wrong type specifier while converting
2505 MEMBER-types to numeric.
2506 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2508 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2509 index argument now works properly on the Alpha platform.
2510 * bug fix: floating point exception treatment on the Alpha platform
2512 * bug fix: FILE-POSITION works much better on string input and
2513 output streams. (thanks to Nikodemus Siivola)
2514 * bug fix: many threading/garbage collection symptoms sorted.
2515 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2516 might be pseudo-atomic.
2517 * internal change: Stopping for GC is now done with signals not
2518 ptrace. GC is now done in whichever thread wanted it, instead of
2520 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2522 * bug fix: result form in DO is not contained in the implicit
2524 * incompatible change: ICR structure is changed; the value part of
2525 CONTINUATION is now called LVAR; corresponding functions are
2526 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2527 * added type deriver for ISQRT (thanks to Robert E. Brown).
2528 * bug fix: better support for loading from the command line when an
2529 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2531 * fixed some bugs revealed by Paul Dietz' test suite:
2532 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2533 ** ROUND and FROUND now give the right answer when given very
2534 small float arguments.
2535 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2537 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2538 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2539 ** LCM should return a non-negative integer.
2540 ** PARSE-INTEGER returned the index of a terminator instead of the
2541 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2542 ** PARSE-INTEGER returned an incorrect index being applied to a
2544 ** LCM with two arguments of 0 returns 0 rather than signalling
2546 ** unsigned addition of a 32-bit constant with the high bit set no
2547 longer causes an internal compiler error on the x86.
2548 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2549 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2550 with negative last argument.
2551 ** byte specifiers with zero size and position no longer cause
2552 an error during type derivation.
2553 ** bignum multiplication on the Alpha platform now returns the
2555 * porting: The system now builds on SuSE AMD64, although it still
2556 generates a 32-bit binary.
2557 * .fasl file incompatibility: The fasl file version number has
2558 been incremented (because of the changes to internal compiler
2559 data structures referred to above).
2561 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2562 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2563 more accurately, on the Darwin kernel running on PowerPC hardware.
2564 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2565 * Compiler code deletion notes now signal a condition of type
2566 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2567 an associated MUFFLE-WARNING restart.
2568 * The compiler now performs limited argument count validation of
2569 constant format strings in FORMAT, and where appropriate in ERROR,
2570 CERROR and WARN. (thanks to Gerd Moellmann)
2571 * New ASDF-INSTALL contrib can be used for automatic download and
2572 installation of third-party Lisp code from CCLAN or other sites
2574 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2575 which forces another thread to execute a function supplied by the
2577 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2578 now accept and act upon their :ELEMENT-TYPE keyword argument.
2579 (reported by Martin Atzmueller, Edi Weitz)
2580 * bug fix: FILE-POSITION now accepts position designators up to
2581 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2582 is the greater. (thanks to Patrik Nordebo)
2583 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2584 Kalle Olavi Niemitalo)
2585 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2586 printable. (reported by Eric Marsden)
2587 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2588 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2589 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2590 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2591 (thanks to Stig Sandoe)
2592 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2593 UNREAD-CHAR now correctly includes the unread character in the
2594 target sequence. (thanks to Gerd Moellmann)
2595 * bug fix in threaded builds: the system can now be suspended and
2596 resumed by shell job control with minimal disruption.
2597 * bug fixes in times and timezones >2038AD
2598 * better handling of "where is GNU make?" problem in build scripts
2599 (thanks to Nikodemus Siivola)
2600 * new optimization: inside a named function any reference to a
2601 function with the same name is considered to be a self-reference;
2602 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2603 optimization quality.
2604 * new optimization on x86: logical functions and + now have
2605 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2606 used when the result is truncated to 32 bits.
2607 * VALUES declaration is partially enabled.
2608 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2609 * fixed some bugs revealed by Paul Dietz' test suite:
2610 ** The system now obeys the constraint imposed by
2611 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2612 types form a lattice under type intersection.
2613 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2614 ** ASSOC now ignores NIL elements in an alist.
2615 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2616 and (1+ MOST-POSITIVE-FIXNUM) answers.
2617 ** The addition of a method with invalid qualifiers to a generic
2618 function does not cause an error to be signalled immediately;
2619 a warning is signalled, and the error is generated only on
2620 calling the generic function.
2621 * changed .fasl file version number, in part to add type codes for
2622 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2623 obscure ANSI requirements
2625 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2626 * fixed bug 148: failure to inline-expand a local function left
2627 garbage, confusing the compiler.
2628 * fixed bugs 3cd: structure slot readers perform type check if the
2629 slot can have an invalid value (i.e. it is either not initialized
2630 or can be written with a less specific slot writer).
2631 * bug fix: the compiler now traps array references to elements off
2632 the end of an array; previously, the bounds checking in some
2633 circumstances could go off-by-one.
2634 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2636 * type declarations for array element types now obey the description
2637 on the CLHS page "Declaration TYPE", as per discussions on
2638 sbcl-help around 2003-05-08. This means that a declaration
2639 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2640 declaration, all references to BAR will be asserted or assumed
2641 (with THE, so dependent on compiler policy) to involve objects of
2642 type FOO. Note that no such declaration is implied in
2643 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2644 * declared types of functions from the "Conditions"
2645 chapter. (reported by Paul Dietz)
2646 * bug fix: CERROR accepts a function as its first argument.
2647 * bug fix: NTH an NTHCDR accept a bignum as index
2648 arguments. (reported by Adam Warner)
2649 * optimization: character compare routines now optimize comparing
2650 against a constant character. (reported by Gilbert Baumann)
2651 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2652 argument now works properly on the MIPS platform.
2653 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2654 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2655 declared to have a functional type, does not cause an error
2657 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2658 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2659 * fixed bug: if last continuation of a deleted block has a
2660 destination, this destination should be deleted too. (reported by
2662 * fixed a bug in the bootstrap process: the host compiler's values
2663 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2664 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2665 test case from Patrik Nordebo)
2666 * improved the ability of the disassembler on the PPC platform to
2667 provide helpful disassembly notes.
2668 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2669 the class in more cases than previously.
2670 * bug fix: FILE-POSITION now understands :START and :END for
2671 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2672 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2673 8) data. (thanks to David Lichteblau)
2674 * bug fix: it is possible to add a method to a generic function
2675 without lambda list.
2676 * bug fix: reader failed to signal END-OF-FILE inside an
2677 object representation. (reported by Nikodemus Siivola)
2678 * fixed some bugs revealed by Paul Dietz' test suite:
2679 ** LAST and [N]BUTLAST should accept a bignum.
2680 ** condition slot accessors are methods.
2681 ** (VECTOR NIL) is a subtype of STRING.
2683 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2684 * minor incompatible change: some nonsensical specialized lambda
2685 lists (used in DEFMETHOD) which were previously ignored now signal
2687 * minor incompatible change: the system is now aware of the types of
2688 variables in the COMMON-LISP package, and will signal errors for
2689 most violations of these type constraints (where previously they
2690 were silently accepted).
2691 * minor incompatible change: COMPILE-FILE now uses the freedom
2692 afforded (ANSI 3.2.2.3) to use derived function types for
2693 functions defined in the same file. This also permits the system
2694 to warn on static type mismatches and function
2695 redefinition. (Currently it does not work with high DEBUG level.)
2696 * minor incompatible change: VALUES declaration is disabled.
2697 * When issuing notes, the compiler now signals a condition of type
2698 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2699 restart for use in user handlers. It is expected that the
2700 COMPILER-NOTE condition will eventually become a condition
2701 supertype to a hierarchy of note types, which will then be
2702 handleable in a similar fashion. However, at the moment, no such
2703 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2704 but it's an implementation detail, not a classification for the
2706 * Changes in type checking closed the following bugs:
2707 ** type checking of unused values (192b, 194d, 203);
2708 ** template selection based on unsafe type assertions (192c, 236);
2709 ** type checking in branches (194bc).
2710 * A short form of VALUES type specifier has ANSI meaning (it has
2711 increased the number of situations when SBCL cannot perform type
2713 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2714 keywords or constants is permissible.
2715 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2716 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2717 classes. (thanks to Antonio Martinez)
2718 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2719 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2720 argument to a pretty-print stream. Instead, it leaves any such
2721 operations to DESCRIBE-OBJECT methods.
2722 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2724 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2725 lambda lists. (thanks to David Lichteblau)
2726 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2727 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2728 (thanks to Gerd Moellmann)
2729 * fixed bug 246: increased compilation speed of long
2730 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2732 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2734 * DEFCONSTANT now throws a condition of type
2735 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2736 constant to a non-EQL value; CONTINUE and ABORT restarts
2737 respectively change and preserve the value.
2738 * fixed bug 63: The code walker, part of the implementation of CLOS,
2739 is now better at handling symbol macros.
2740 * bug fix: There is no longer an internal implementation type named
2741 CL:LENGTH. (reported by Raymond Toy)
2742 * bug fix: In macro-like defining macros/special operators the
2743 implicit block does not enclose the lambda list.
2744 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2745 enclosed in parentheses) are not suitable as type specifiers, and
2746 their use properly signals an error now.
2747 * bug fix: An argument count mismatch for a type specifier in code
2748 being compiled no longer causes an unhandled error at compile
2749 time, but signals a compile-time warning.
2750 * fixed simple vector readable printing
2751 * bug fix: DESCRIBE takes more care over whether the class
2752 precedence list slot of a class is bound before accessing it.
2753 (reported by Markus Krummenacker)
2754 * bug fix: FORMATTER can successfully compile pretty-printer format
2755 strings which use variants of the ~* directive inside.
2756 * bug fix: SEARCH now applies its TEST predicate to the elements of
2757 the arguments in the correct order. (thanks to Wolfhard Buss)
2758 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2759 in type inference. (thanks to Robert E. Brown)
2760 * bug fix: Reading in symbols with an explicit package name of ""
2761 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2762 * fixed some bugs revealed by Paul Dietz' test suite:
2763 ** NIL is now allowed as a structure slot name.
2764 ** Arbitrary numbers, not just REALs, are allowed in certain
2765 circumstances in LOOP for-as-arithmetic clauses.
2766 ** Multiple class redefinitions before slot access no longer
2767 causes a type error.
2768 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2769 association between the name and a class.
2770 ** Generic functions with non-standard method-combination and over
2771 five methods all of which return constants no longer return NIL
2772 after the first few invocations. (thanks to Gerd Moellmann)
2773 ** CALL-NEXT-METHOD with no arguments now passes the original
2774 values of the arguments, even in the presence of assignment.
2775 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2777 ** DISASSEMBLE works with closures and funcallable instances.
2778 ** ADD-METHOD now returns the generic function, not the new method.
2779 ** FIND-METHOD signals an error if the lengths of the specializers
2780 is incompatible with the generic function, even if the ERRORP
2782 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2783 which its argument is a member.
2784 ** DEFCLASS only redefines the class named by its class-name
2785 argument if that name is the proper name of the class;
2786 otherwise, it creates a new class.
2787 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2788 of the UNBOUND-SLOT condition to the name of the slot.
2789 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2790 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2791 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2792 SLOT-MAKUNBOUND in the specified fashion.
2794 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2795 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2796 cross-compilation host. As a consequence, we can now bootstrap our
2797 way up to SBCL starting with a bare gcc toolchain and human-readable
2798 source code (first the source to CLISP, then the source to SBCL).
2799 * A contributed module containing a partial implementation of the
2800 simple-streams interface has been included. (thanks to Rudi
2802 * A contributed module implementing the RFC1321 Message Digest
2803 Algorithm, known as MD5, has been included.
2804 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2805 no longer has any effect, as the code controlled by this feature
2806 has been deleted. (As far as we know, no-one has ever built using
2807 this feature, and its semantics were confused in any case).
2808 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2809 work on conditions (as required by the ANSI specification),
2810 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2811 expected behaviour on conditions. Users should note, however,
2812 that such behaviour is not required by the ANSI specification,
2813 and so use of this behaviour may render their code unportable.
2814 * fixed some bugs revealed by Paul Dietz' test suite:
2815 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2817 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2818 ** SXHASH on condition objects no longer returns NIL.
2819 ** :ALLOCATION :CLASS slots are better treated; their values are
2820 updated on class redefinition, and initforms inherited from
2821 superclasses are applied.
2822 ** REMOVE-METHOD returns its generic function argument even when
2823 no method was removed.
2824 ** SHARED-INITIALIZE now initializes the values of the requested
2825 slots, including those with :ALLOCATION :CLASS.
2826 ** ALLOCATE-INSTANCE now works on structure classes defined via
2827 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2829 ** SLOT-EXISTS-P now works on conditions, as well as structures
2831 ** MAKE-LOAD-FORM now has the required methods on
2832 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2833 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2834 keyword, and now implements the SLOT-NAMES argument.
2835 ** methods with &OPTIONAL arguments no longer allow too many
2836 arguments to be passed in the call without error.
2837 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2838 option is consistent with the required arguments of the generic
2839 function lambda list.
2840 * bug fix: REQUIRE accepts a string designator. (Thanks to
2842 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2843 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2844 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2845 * bug fix: The compiler now checks for duplicated variables in macro
2847 * bug fix: SETQ on globals returns the correct value.
2848 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2849 not a COMPILER-ERROR (followed by some other strange error on
2850 choosing the CONTINUE restart).
2851 * bug fix: make.sh and friends are now more consistent in the way that
2852 they look for GNU "make".
2854 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2855 * experimental native threads support (on x86 Linux >=2.4 only).
2856 This is not compiled in by default: you need to add :SB-THREAD to
2857 the target features. See the "Beyond ANSI" chapter of the manual
2859 * fix for longstanding nonANSIism: The old distinction between
2860 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2861 The return value from CL:FIND-CLASS is now a CLOS class, and
2862 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2863 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2864 * An interface to the MetaObject Protocol, as described in Kiczales,
2865 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2866 MIT Press, 1991, is available from the SB-MOP package.
2867 * incompatible change: the SB-PCL package should now be considered
2868 a private implementation detail, and no longer a semi-private MOP
2870 * minor incompatible change: due to rearrangement for threads, the
2871 control stack and binding stack are now allocated at arbitrary
2872 addresses instead of being hardcoded per-port. Users affected by
2873 this probably have to be doing advanced things with shared
2874 libraries, and will know who they are.
2875 * minor incompatible change: Previously, all --eval forms used were
2876 processed with READ before any of them were processed with EVAL.
2877 Now each --eval form is processed with both READ and EVAL before
2878 the next --eval form is processed. (Thus package operations like
2879 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2880 work as the user might reasonably expect.)
2881 * minor incompatible change: *STANDARD-INPUT* is now only an
2882 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2884 * minor incompatible change: Y-OR-N-P is now character-oriented,
2885 not line oriented. Also, YES-OR-NO-P now works without errors.
2886 (thanks to Antonio Martinez)
2887 * sb-aclrepl module improvements: an integrated inspector, added
2888 repl features, and a bug fix to :trace command.
2889 * Known functions, which cannot be open coded by the backend, are
2890 considered to be able to check types of their arguments. (fixing
2891 a bug report by Nathan J. Froyd)
2892 * fixed a bug in computing method discriminating functions: It is
2893 now possible to define methods specialized on classes which have
2894 forward-referenced superclasses. (thanks to Gerd Moellmann)
2895 * fixed evaluation order in optional entries (reported by Gilbert
2897 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2898 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2899 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2900 required-by-AMOP NAME argument, as well as CLASS and
2901 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2902 * fixed bug 20: DEFMETHOD can define methods using names that are
2903 not the proper names of classes to designate class specializers.
2904 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2905 Unix file descriptors, instead of blowing up. (thanks to Antonio
2907 * Garbage collection refactoring: user-visible change is that a
2908 call to the GC function during WITHOUT-GCING will not do garbage
2909 collection until the end of the WITHOUT-GCING. If you were doing
2910 this you were probably losing anyway.
2911 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2912 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2913 TYPEP the latter but not the former.
2914 * The compiler issues a full WARNING for calls to undefined functions
2915 with names from the CL package.
2916 * MAP-INTO for a vector destination is open coded. (reported by
2917 Brian Downing on c.l.l)
2918 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2919 documentation string.
2920 * fixed some bugs revealed by Paul Dietz' test suite:
2921 ** COPY-ALIST now signals an error if its argument is a dotted
2923 ** Condition slots are now accessed more correctly in the presence
2924 of multiple initargs for a given slot.
2925 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2926 exclude restarts of the same name associated with a different
2928 ** DEFCLASS of forward-referenced classes with another
2929 forward-referenced class in the superclasses list no longer
2931 ** Condition slots are now initialized once each, not multiple
2932 times. (thanks to Gerd Moellmann)
2933 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2935 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2936 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2937 arguments contain duplicated elements.
2938 ** RESTART-CASE understands local macros.
2939 ** RESTART-CASE associates exactly its own restarts with a condition.
2940 ** ENDP in safe mode checks its argument to be of type LIST.
2941 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2942 in question is unbound.
2943 ** Optimized MAKE-INSTANCE functions no longer cause internal
2944 assertion failures in the presence of duplicate initargs.
2945 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2946 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2947 propagated even in the ctor optimized implementation of
2949 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2951 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2952 * a better implementation of SXHASH on (simple) bit vectors,
2953 measured both in execution speed and in distribution of results
2954 over the positive fixnums, has been installed. Likewise, a better
2955 implementation of EQUAL for simple bit vectors is now available.
2956 * fixed CEILING optimization for a divisor of form 2^k.
2957 * fixed bug 240 (emitting extra style warnings "using the lexical
2958 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2959 by Antonio Martinez)
2960 * fixed SXHASH, giving different results for NIL depending on type
2961 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2962 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2963 variable. (found by Rolf Wester)
2964 * SBCL does not ignore type declarations for special
2965 variables. (reported by rif on c.l.l 2003-03-05)
2966 * some bug fixes in contrib/sb-aclrepl/
2967 * fixed some bugs revealed by Paul Dietz' test suite:
2968 ** a bug in the CONS type specifier, whereby the CAR and CDR
2969 types got intertwined, has been fixed;
2970 ** the type system is now able to reason about the interaction
2971 between INTEGER and RATIO types more completely;
2972 ** APPEND, [N]REVERSE and NRECONC check that those their
2973 arguments, which must be proper lists, are really so;
2974 ** An array specialized to be unable to hold elements has been
2975 implemented, as required -- yes, really -- by ANSI;
2976 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2977 on malformed property lists;
2979 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2980 * incompatible packaging change: in line with Unix convention,
2981 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2982 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2983 * REQUIRE and PROVIDE are now optionally capable of doing something
2984 useful. See the documentation string for REQUIRE.
2985 * infrastructure for a managed SBCL contrib system: contributed
2986 modules in this release include:
2987 ** the ASDF system definition facility;
2988 ** an interface to the BSD Sockets API;
2989 ** an ACL-like convenience interface to the repl;
2990 (thanks to Kevin Rosenberg)
2991 ** an implementation of ROTATE-BYTE, with efficient implementation
2993 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2994 gives a closer approximation to the right answer than previously.
2995 (thanks to Raymond Toy)
2996 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2997 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2998 optional environment arguments, as required by ANSI.
2999 * fixed bugs in other functions taking environment objects, allowing
3000 calls with an explicit NIL environment argument to be compiled
3002 * fixed bug 228: primary return values from
3003 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3004 COMPILE or FUNCTION.
3005 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3006 :TYPE will no longer signal errors on innocuous objects.
3007 * fixed bug 231b: SETQ is better at respecting type declarations in
3008 the lexical environment.
3009 * fixed a bug in DEFCLASS: classes named by symbols with no or
3010 unprintable packages can now be defined.
3011 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3012 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3013 * fixed compiler failure related to checking types of functions.
3014 (reported by Robert E. Brown)
3015 * the compiler is now much more consistent in its error-checking
3016 treatment of bounding index arguments to sequence functions: in
3017 (SAFETY 3) code, errors will be signalled in almost all cases if
3018 invalid sequence bounding indices are passed to functions defined
3019 by ANSI to operate on sequences.
3020 * fixed a bug in the build procedure: documentation of SBCL-specific
3021 packages is now preserved and available in the final Lisp image.
3022 * lifted FDEFINITION lookup out of loops in the implementation of
3023 many list operations. (thanks to Robert E. Brown)
3024 * fixed a bug in the reader: the #n# reader macro now works for
3025 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3026 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3027 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3028 E. Brown and Tony Martinez respectively)
3029 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3030 SLOT-BOUNDP: in particular, it is now able to optimize them much
3031 better, and is now not vulnerable to having packages renamed.
3032 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3033 always called when a slot is not present in an instance. (thanks
3035 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3036 longer signal an error on unreading a character following EOF on
3037 the previous constituent stream. (thanks to Tony Martinez)
3038 * fixed some bugs revealed by Paul Dietz' test suite:
3039 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3040 not just nonnegative fixnums;
3041 ** the logical bit-array operators such as BIT-AND now accept an
3042 explicit NIL for their "opt-arg" argument (to indicate a
3043 freshly-consed result bit-array);
3044 ** ELT now signals an error on an invalid sequence index in safe
3046 ** the type system is now cleverer about negations of numeric
3047 types, and consequently understands the BIGNUM and RATIO types
3049 ** the type system is now cleverer about the interaction between
3050 INTEGER and RATIO types: while bugs still remain, many more
3051 cases are accurately computed;
3052 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3053 if it is in the last clause;
3054 ** CONSTANTLY now correctly returns a side-effect-free function in
3056 ** DECLARE is no longer treated as a special-operator; in
3057 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3058 * incremented fasl file version number due to the change in the
3059 DEFSTRUCT-SLOT-DESCRIPTION structure.
3061 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3062 * minor incompatible change: code processed by the "interpreter" or
3063 EVAL now has a compilation optimization policy of (DEBUG 2)
3064 (changed from (DEBUG 1)) to improve debuggability of interactive
3065 development, and to allow the use of the debug RETURN command in
3067 * an experimental implementation of the RETURN command for the
3068 debugger has been included. (thanks to Frederik Kuivinen)
3069 * fixed bug 62: constraints were not propagated into a loop.
3070 * fixed bug in embedded calls of SORT (reported and investigated by
3072 * fixed some bugs revealed by Paul Dietz' test suite:
3073 ** printing and reading of arrays with some dimensions having
3074 length 0 (thanks to Gerd Moellmann);
3075 ** BOA constructor with &AUX argument without a default value does
3076 not cause a type error;
3077 ** CONSTANTP now returns true for all self-evaluating objects.
3079 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3080 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3081 accessors that are related by inheritance, as specified in the
3082 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3083 Valtteri Vuorikoski)
3084 * The compiler is now able to inline functions that were defined in
3085 a complex lexical environment (e.g. inside a MACROLET).
3086 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3088 * Support for the upcoming FreeBSD-5.0 release has been included.
3089 (thanks to Dag-Erling Smorgrav)
3090 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3091 effect when it is not in a toplevel context.
3092 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3093 stand a better chance of being correct. (thanks to Gerd
3095 * fixed bug in COERCE, which now signals an error on coercing a
3096 rational to a bounded real type which excludes the expected
3098 * The compiler is now able to derive types more accurately from the
3099 COERCE and COMPILE functions.
3100 * fixed bug 223: functional binding is considered to be constant
3101 only for symbols in the CL package.
3102 * fixed bug 231: SETQ did not check the type of a variable being set
3103 (reported by Robert E. Brown)
3104 * A new optimization for MAKE-INSTANCE has been included, fixing
3105 various bugs (including relating to :ALLOCATION :CLASS slots and
3106 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3108 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3109 ** As required by ANSI, LOOP now disallows anonymous collection
3110 clauses such as COLLECT I in conjunction with aggregate boolean
3111 clauses such as THEREIS (= I 1);
3112 ** LOOP now signals an error when any variable is reused in the
3113 same loop (including the potentially useful construct analogous
3114 to WITH A = 1 WITH A = (1+ A);
3115 ** IT is only a special loop symbol within the first clause of a
3116 conditional loop clause;
3117 ** LOOP with a typed iteration variable over a hashtable now
3118 signals a type error iff it should.
3119 * fixed some other bugs revealed by Paul Dietz' test suite:
3120 ** FILE-STREAM now names the class previously known as FD-STREAM;
3121 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3122 argument) no longer signals an error;
3123 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3124 of the slot symbol, rather than using the current package
3125 ((:CONC-NAME "") continues to intern the slot's name in the
3127 * incremented fasl file version number, because of the incompatible
3128 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3129 of the new implementation of DEFINE-COMPILER-MACRO.
3131 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3132 * Support for building SBCL for MIPS platforms running in
3133 little-endian mode has now been checked in, and basic
3134 functionality on said platforms verified.
3135 * minor incompatible change: PCL now records the pathname of a file
3136 in which methods and the like are defined, rather than its
3138 * minor incompatible change: TRUENAME now considers the truename of
3139 a file naming a directory to be the pathname with :DIRECTORY
3140 component indicating that directory.
3141 * minor incompatible change: a NAMED clause in the extended form of
3142 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3143 reason for the previous behaviour is unclear.
3144 * more systematization and improvement of CLOS and MOP conformance
3145 in PCL (thanks to Gerd Moellman and Pierre Mai):
3146 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3148 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3149 primary methods with no specializers;
3150 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3152 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3153 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3154 FORWARD-REFERENCED-CLASSes; error reporting on
3155 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3157 ** SXHASH on CLOS instances now uses a slot internal to the
3158 instance to return different numbers on distinct instances,
3159 while preserving the same return value through invocations of
3161 ** DEFMETHOD signals errors when methods with longer incongruent
3162 lambda lists are added to generic functions;
3163 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3164 CLASS, as specified in AMOP;
3165 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3166 on the order returned by the primary method for classes of
3167 class STANDARD-CLASS;
3168 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3169 * fixed some bugs shown by Paul Dietz' test suite:
3170 ** DOLIST puts its body in TAGBODY;
3171 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3173 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3174 value producing form;
3175 ** if more variables are given to PROGV than values, extra
3176 variables are bound and made to have no value;
3177 ** NSUBSTITUTE on list arguments gets the right answer with
3179 ** ELT signals an error of type TYPE-ERROR when the index argument
3180 is not a valid sequence index;
3181 ** LOOP signals (at macroexpansion time) an error of type
3182 PROGRAM-ERROR when duplicate variable names are found;
3183 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3184 ** FUNCALL of special-operators now cause an error of type
3186 ** PSETQ now works as required in the presence of side-effecting
3187 symbol-macro places;
3188 ** NCONC accepts any object as its last argument;
3189 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3191 ** loop-for-as-package does not require a package to be explicitely
3193 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3195 * fixed bug 166: compiler preserves "there is a way to go"
3196 invariant when deleting code.
3197 * fixed bug 172: macro lambda lists with required arguments after
3198 &REST arguments now cause an error to be signalled. (thanks to
3200 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3201 bug. (thanks to Matthew Danish)
3202 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3204 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3205 arguments are lexically rebound. (thanks to Gerd Moellmann and
3207 * fixed bug 194: error messages are now more informative when there
3208 is no primary method applicable in a call to a generic
3209 function. (thanks to Gerd Moellmann)
3210 * fixed bug in command line argument checking (thanks to Julian
3212 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3213 (thanks to Matthew Danish)
3214 * incremented fasl file version number, because of the
3215 SXHASH-related changes in the layout of CLOS data structures
3217 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3218 * minor incompatible change: The runtime (the Unix executable named
3219 "sbcl") is now much pickier about the .core files it will load.
3220 Essentially it now requires .core files to descend from the same
3221 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3222 as the runtime does. (The intent is to prevent the crashes which
3223 can occur, and which can even be reported as mysterious failures,
3224 when people patch the sources or change the build parameters
3225 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3226 sbcl and .core files.)
3227 * fixed bug: VALUES-LIST is no longer optimized away.
3228 * fixed bug 142: The FFI conversion of C string values to Lisp
3229 string values no longer conses excessively. (thanks to Nathan
3230 Froyd porting Raymond Toy's fix to CMU CL)
3231 * began to systematize and improve MOP conformance in PCL (thanks to
3232 Nathan Froyd, Gerd Moellman and Pierre Mai):
3233 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3235 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3236 ** FINALIZE-INHERITANCE is now called on class finalization;
3237 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3238 argument precedence order.
3239 * fixed bug 202: The compiler no longer fails on functions whose
3240 derived types contradict their declared type.
3241 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3242 so it can be non-toplevel.
3243 * The fasl file version number has changed (because of the new
3244 implementation of DEFMACRO).
3245 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3246 safe code, that any length requirement by their type-specifier
3247 argument is valid. The exceptions to this are described in bug
3249 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3250 functions now check the types of their inputs as required by ANSI.
3251 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3252 introduced symbol is DECLAREd to be SPECIAL.
3253 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3254 * fixed bug: :COUNT argument to sequence functions may be negative
3255 * fixed bug: body of DO-SYMBOLS may contain declarations
3256 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3257 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3258 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3259 symbol macro only once
3260 * fixed printing of call frame when argument list is unavailable
3261 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3262 * compiler no longer signals WARNING on unknown keyword
3265 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3266 * A beta-quality port to the mips architecture running Linux,
3267 based on the old CMUCL backend, has been made. It has been tested
3268 on a big-endian kernel, and works sufficiently well to be able to
3269 rebuild itself; it has not been tested in little-endian mode.
3270 * fixed an inconsistency between gencgc.c and purify.c which made
3271 dumping/loading .core files unreliable
3272 * fixed bug 120a: The compiler now deals correctly with IFs where
3273 the consequent is the same as the alternative, instead of
3274 misderiving the return type. (thanks to Alexey Dejneka)
3275 * fixed bug 113: Logical pathnames are now dumpable (the logical
3276 host is resolved at load-time, throwing an error if it is not
3278 * fixed bug 174: FORMAT's error message is slightly clearer when a
3279 non-printing character is used in a format directive.
3280 * fixed several bugs in compiler checking of type declarations, i.e.
3281 violations of the Python "declarations are assertions" principle
3282 (thanks to Alexey Dejneka)
3283 * fixed several bugs in PCL's error checking (thanks to Gerd
3285 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3287 * fixed bug in compilation of functions as first class values
3288 (thanks to Antonio Martinez-Shotton)
3289 * The compiler's handling TYPE-ERRORs which it can prove will
3290 inevitably happen at runtime has been cleaned up and corrected
3291 in several ways. (thanks to Alexey Dejneka)
3292 * improved argument type checking for various basic arithmetic
3293 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3294 lavished on them in the past that they can be compiled in many
3295 ways in different special cases
3296 * fixed bug 181: compiler checks validity of user supplied type
3298 * cleaned up code flushing in optimization: Function calls which
3299 should signal errors for safety purposes (e.g. which ANSI says
3300 should signal errors when their arguments are of incorrect type)
3301 are no longer optimized away.
3302 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3303 * incremented fasl file version number, because changes in the
3304 implementation of sequence functions like COERCE caused
3305 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3306 in old inline expansions) to become undefined. (Actually these
3307 changes were later undone, so we might very well be binary
3308 compatible with 0.7.7 after all, but leaving the version number
3309 incremented seemed like the simplest and most conservative
3312 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3313 * An alpha-quality port to the parisc architecture running Linux,
3314 based on the old CMUCL backend, has been made. This, even more so
3315 than the other backends, should be considered still a work in
3316 progress; known problems include that the Linux kernel in 64-bit
3317 mode does not propagate the correct sigcontext structure to
3318 userspace, and consequently SBCL on a parisc64 kernel will not
3320 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3321 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3322 suppressed.) Also now that the compiler is looking at declarations
3323 in the environment, it checks optimization declarations as well,
3324 and suppresses inlining when (> DEBUG SPEED).
3325 * More fixes have been made to treatment of floating point exception
3326 treatment and other Unix signals. In particular, floating point
3327 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3328 * The detection and handling of control stack exhaustion (infinite
3329 or very deeply nested recursion) has changed. Stack exhaustion
3330 detection is now done by write-protecting pages at the OS level
3331 and applies at all optimization settings; when found, a
3332 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3333 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3334 be caught using IGNORE-ERRORS.
3335 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3336 correctly when transforming recursive LABELS functions to LETs.
3337 (thanks to Alexey Dejneka)
3338 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3339 that are names of constants or global variables.
3340 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3341 alien routines with docstrings.
3342 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3343 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3345 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3346 (thanks to Eric Marsden)
3347 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3348 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3349 * Bug fix: The SPARC backend can now compile functions involving
3350 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3351 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3352 to a functions expecting a PATHNAME-DESIGNATOR.
3353 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3354 lambda lists. (thanks to Alexey Dejneka)
3355 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3356 functions so that EXPORTing the name of the class doesn't cause
3357 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3358 to Antonio Martinez for reporting this)
3359 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3360 constant-folded without failing an assertion. (thanks to Einar
3361 Floystad Dorum for reporting this)
3362 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3363 types are now dealt with more correctly by the compiler.
3364 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3365 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3366 behaviour analogous to RENAME-FILE. This puts its behaviour more
3367 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3369 * The fasl file version number has changed again. (because of the
3370 bug fix involving the names of PCL MAKE-INSTANCE functions)
3372 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3373 * bug fix: Floating point exceptions are treated much more
3374 consistently on the x86/Linux and PPC/Linux platforms.
3375 * Array initialization with :INITIAL-ELEMENT is now much faster for
3376 cases when the compiler cannot open code the array creation, but
3377 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3378 array accesses have also seen a speed increase.
3379 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3380 specified. (This was at the root of some bad interactions between
3381 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3383 * bug fix: Internal error arguments for undefined functions are now
3384 computed correctly on the PPC/Linux platform.
3385 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3386 Raymond Toy's patch for CMU CL)
3387 * Support for the Solaris 9 operating environment has been included
3388 (thanks to Daniel Merritt)
3389 * A very ugly but hopefully complete draft of the missing FFI chapter
3390 of the manual has been created by reformatting the corresponding
3391 CMU CL manual chapter into (currently very ugly and incoherent)
3392 DocBook and bringing it up to date for SBCL behavior. Thus, the
3393 manual is now essentially complete, at least by my extreme
3394 once-and-only-once standards, whereby it's acceptable to refer to
3395 the doc strings of SB-EXT functions as the primary documentation.
3396 * The fasl file version number has changed again, due to cleanup of
3397 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3398 is no longer a static symbol.)
3400 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3401 * SBCL now builds with OpenMCL (version 0.12) as the
3402 cross-compilation host; also, more progress has been made toward
3403 bootstrapping under CLISP.
3404 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3406 * bug 158 fixed: The compiler can now deal with integer loop
3407 increments different from 1; fixing this turned out also to fix
3409 * bug 169 fixed: no more bogus warnings about using lexical bindings
3410 despite the presence of perfectly good SPECIAL declarations (thanks
3411 to David Lichteblau)
3412 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3413 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3414 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3416 * bug 180 fixed: Method combination specifications no longer ignore
3417 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3418 * bug fix: Structure type predicate functions now check their argument
3419 count as they should.
3420 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3421 correctly. (thanks to Pierre Mai)
3422 * minor incompatible change: The --noprogrammer option is deprecated
3423 in favor of the new --disable-debugger option, which is very similar.
3424 (The major difference is that it takes effect at a slightly different
3425 time at startup, causing handling of errors in --sysinit and
3426 --userinit files will be affected differently.) The
3427 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3428 been added to allow this functionality to be controlled from ordinary
3429 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3430 maintainers, who might want to run non-interactive scripts to
3431 build SBCL cores which will later be used interactively.)
3432 * minor incompatible change: The LOAD function no longer, when given
3433 a wild pathname to load, loads all files matching that pathname.
3434 Instead, an error of type FILE-ERROR is signalled.
3436 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3437 * bug 147 fixed: The compiler preserves its block link/count
3438 invariants more correctly now so that it doesn't crash. (thanks
3440 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3442 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3443 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3444 no-value-for-this-character as NIL (as specified by ANSI).
3445 * HOST-NAMESTRING on physical pathnames now returns a string that is
3446 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3447 (thanks to Christophe Rhodes)
3448 * The Alpha port handles icache flushing more correctly. (thanks to
3450 * More progress has been made toward bootstrapping under CLISP. (thanks
3451 to Christophe Rhodes)
3452 * The fasl file format has changed again, because dynamic loading
3453 on OpenBSD (which has non-ELF object files) motivated some cleanups
3454 in the way that foreign symbols are transformed and passed around.
3455 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3456 is no longer treated as whitespace by the reader, but instead as
3457 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3458 |A
\x7fB|, instead of A as it used to.
3460 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3461 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3462 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3463 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3464 this as of 0.7.1.45, but was left out of the previous news section
3465 (thanks to Dan Barlow)
3466 * SBCL now runs on the Solaris operating system on SPARC architectures
3467 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3468 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3469 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3470 * SPARC backend cleanups, allowing builds of cores optimized for V8
3471 and V9 SPARCS, and also emission of code targeted to a particular
3472 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3474 * SBCL is closer to bootstrapping under CLISP, thanks to various
3475 fixes by Christophe Rhodes.
3476 * The fasl file format has changed again, to allow the compiler's
3477 INFO database to support symbol macros.
3478 * The user manual (in doc/) is formatted into HTML more nicely.
3479 (thanks to coreythomas)
3480 * The system is smarter about SUBTYPEP relationships, especially
3481 those involving NOT types (including types such as ATOM which are
3482 represented internally using NOT types). Thus SUBTYPEP is less
3483 likely to return (VALUES NIL NIL) in general, and in particular
3484 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3486 * The fasl file format has changed again, because the internal
3487 representation of types now includes a new slot to support the new
3488 SUBTYPEP-of-NOT-types logic.
3489 * (not a change in the main branch of SBCL, but a related prototype
3490 which can hopefully be merged into the main branch of SBCL in the
3491 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3492 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3493 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3494 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3495 dispatch macro character combinations. (thanks to Alexey Dejneka)
3496 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3497 that superficially look like logical namestrings correctly.
3498 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3500 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3501 default supplied-p) for &optional and &key arguments. (thanks to
3504 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3505 * incompatible change: The compiler is now less aggressive about
3506 tail call optimization, doing it only when (> SPACE DEBUG) or
3507 (> SPEED DEBUG). (This is an incompatible change because there are
3508 programs which relied on the old CMU-CL-style behavior to optimize
3509 away their unbounded recursion which will now die of stack overflow.)
3510 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3511 for non-GENCGC systems has been increased to 20M (since that
3512 seems much closer to the likely performance optimum for modern
3513 systems than the old 4M value was)
3514 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3515 and *DEBUG-PRINT-LEVEL*
3516 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3517 of CMU CL's support for SPARC, and various endianness and other
3518 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3519 * new syntactic sugar for the Unix command line: --load foo.bar is now
3520 an alternate notation for --eval '(load "foo.bar")'.
3522 ** The system now detects stack overflow and handles it gracefully,
3523 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3524 optimization settings. (This is a good thing in general, and
3525 its introduction in this version should be particularly timely
3526 for anyone whose code fails because of suppression of tail
3528 ** The system now hunts for the C variable "environ" in a more
3529 devious way, to avoid segfaults when the C library version differs
3530 between compile time and run time. (thanks to Christophe Rhodes)
3531 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3532 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3533 ** The compiler no longer issues bogus style warnings for undefined
3534 classes in the same source file as the DEFCLASSes which defined
3535 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3537 ** fixes in CONDITION class precedence list for undefined function
3538 errors (thanks to Alexei Dejneka)
3539 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3540 correctly. (thanks to Dan Barlow)
3541 ** portability fixes aiming at bootstrapping under CLISP (thanks
3542 to Dave McDonald and Christophe Rhodes)
3543 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3544 ** fixes in type translation and and type inference (thanks to
3546 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3547 ** various fixes in the new ports (thanks to Dan Barlow)
3548 * several changes related to debugging:
3549 ** suppression of tail recursion, as noted above
3550 ** stack overflow detection, as noted above
3551 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3552 is now the default. (For some time encapsulation has been more
3553 reliable than the breakpoint-based :ENCAPSULATE NIL
3554 implementation, at least on X86 systems; and I just noticed that
3555 encapsulation also seems closer to the spirit of the ANSI
3558 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3560 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3561 up properly again. (There was a packaging bug in 0.7.0 which
3562 left their definitions in SB-SYS::LOAD-FOREIGN and
3563 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3564 vital for most things which interface to C-level interfaces,
3565 like extensions working with sockets or databases or
3566 Perl-compatible regexes or whatever, and the need to fix
3567 this bug is the main reason that 0.7.1 was released so
3569 ** DEFGENERIC is now choosier about the methods it redefines, so that
3570 reLOADing a previously-LOADed file containing DEFGENERICs does
3571 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3572 cycle now works as it should. (thanks to Alexey Dejneka)
3573 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3574 fixed by Christophe Rhodes. (He actually submitted this patch
3575 months ago, and I delayed until after 0.7.0.)
3576 ** Bug 111 (internal compiler confusion about runtime checks on
3577 FUNCTION types) was fixed by Alexey Dejneka.
3578 * Some internal cleanups (getting rid of variables which aren't
3579 needed now that the byte interpreter is gone) caused the fasl
3580 file format number to change again.
3582 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3583 * major incompatible change: The default fasl file extension, i.e. the
3584 default extension for files produced by COMPILE-FILE, has changed
3585 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3587 ** There are many changes in the implementation of the compiler.
3588 SBCL is now essentially a compiler-only implementation of ANSI
3589 Common Lisp. EVAL still "interprets" a few special cases, but
3590 almost all the interesting cases are handled by creating
3591 a LAMBDA expression, calling COMPILE on it, then calling
3592 FUNCALL on the result.
3593 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3594 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3595 Since the code is newer, there might still be some new bugs
3596 (though not as many as before Martin Atzmueller's fixes:-). But
3597 the new code is substantially simpler and clearer, and hopefully
3598 any remaining bugs will be simpler, less fundamental, and more
3599 fixable then the bugs in the old code.
3600 ** The revised compiler is still a little unsteady on its feet.
3602 *** The debugging information it produces (particularly the names
3603 of FUNCTION objects) is sometimes much less useful than what
3604 the old compiler produced.
3605 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3606 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3607 within a MACROLET) has been temporarily weakened.
3608 ** There are new compiler optimizations for various functions:
3609 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3610 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3611 *** the math functions TRUNCATE, FLOOR, and CEILING
3612 *** the function-of-all-trades COERCE
3613 Mostly these should be transparent, but there's one
3614 potentially-annoying problem (bug 117): when the compiler
3615 inline-expands a function and does type analysis on the result,
3616 it can create control paths which have type mismatches, and
3617 when it can't prove that those control paths aren't taken,
3618 it will issue WARNINGs about the type mismatches. This is
3619 a particular problem in practice for the new sequence functions.
3620 It's not clear how this should be fixed, and for now, a
3621 workaround is given in the entry for 117 in the BUGS file.
3622 ** (Because of the interaction between the two previous items --
3623 occasional inlining problems and new inline expansions -- some
3624 of the new sequence function optimizations won't really kick in
3625 completely until debugging information, and then inlining, are
3626 straightened out in some future version.)
3627 * minor incompatible changes:
3628 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3629 DIRECTORY no longer implicitly promotes NIL slots of its
3630 pathname argument to :WILD. In particular, when you ask for the
3631 contents of a directory (which you used to be able to do without
3632 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3633 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3634 ** changes in behavior that ANSI explicitly defines to be
3635 implementation dependent:
3636 *** The new compiler-only implementation still conforms with ANSI,
3637 but acts a little different than before. Besides the obvious
3638 changes in performance tradeoffs (that the cost per form passed
3639 to EVAL has gone up, and the cost per form executed by EVAL
3640 has gone down), the behavior of the system changes a little
3641 because there are no longer any interpreted function objects.
3642 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3643 e.g. doing COMPILE on the output of interactive DEFUN is
3645 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3647 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3648 has changed from (:CL) to NIL.
3649 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3650 e.g. #\Newline and #\Tab, don't have names specified in the
3651 ANSI Common Lisp standard, is now based on their ASCII symbolic
3652 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3653 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3654 are no longer used for output.
3655 ** changes in internal implementation constants:
3656 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3657 4 million. (If your application spends a lot of time GCing and
3658 you have a lot of RAM, you might want to experiment with
3659 increasing it even more.)
3660 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3661 However, almost all old code should still continue to work without
3662 immediate update, as SB-C-CALL is now a (deprecated) nickname
3664 ** Old operator names in the style DEF-FOO are now deprecated in
3665 favor of new corresponding names DEFINE-FOO, for consistency with
3666 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3667 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3668 internal symbols, but a few supported extensions like
3669 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3670 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3671 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3672 etc. as you get deeper into recursive calls to the debugger
3673 command loop, instead of the old "5]", "5]]", "5]]]"
3674 sequence. (I was motivated to do this when squabbles between
3675 ILISP and SBCL left me very deeply nested in the debugger. In the
3676 short term, this change will probably provoke more ILISP/SBCL
3677 squabbles, but hopefully it will be an improvement in the long run.)
3678 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3679 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3680 function, since declaiming return types involving aliens is
3681 (1) annoyingly messy to do by hand and (2) vital to efficient
3682 compilation of code which calls such functions.
3683 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3684 longer reexported by the SB-EXT package. They're solely useful
3685 for alien code, so it seems more logical that you should get
3686 them from the SB-ALIEN package, not in SB-EXT.
3687 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3688 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3689 features. Instead, the code that they used to control is always
3690 built into the system.
3691 * many other bug fixes
3692 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3693 advantage of the new EVAL-WHEN stuff and to clean them up in
3694 general, and they are now more ANSI-compliant in a number of
3695 ways. Martin Atzmueller is responsible for a lot of this.
3696 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3698 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3699 *** correct ERROR type for various file operations
3700 *** some fixes for Lisp streams
3701 *** DEFMETHOD syntax checking
3702 *** changing old weird representation of debug information as
3703 strings (which, among their other deficiencies, don't transform
3704 correctly when you rename packages, and don't change their
3705 print representation when you change things like *PACKAGE*
3706 and *PRINT-LENGTH*) to symbols and lists of symbols
3707 He also made several improvements and fixed several bugs in DESCRIBE.
3708 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3710 *** misbehavior of WRITE-STRING/WRITE-LINE
3711 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3712 and several other LOOP problems as well
3713 *** DIRECTORY when similar filenames are present
3714 *** DEFGENERIC with :METHOD options
3715 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3716 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3717 *** argument ordering in FIND with :TEST option
3718 *** mishandled package designator argument in APROPOS-LIST
3719 *** various problems in the backquote readmacro
3720 *** a bug in APROPOS
3721 *** probably some others that I'm not describing very well here,
3722 since the CVS log documents them by reference to sbcl-devel
3723 messages, and the SourceForge archives aren't working well.:-(
3724 ** Dan Barlow improved the Alpha port (and is making progress on the
3725 PPC port, for those of you who think different).
3726 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3727 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3728 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3729 continues to work on the SPARC port (for those of you in a position
3730 to look down upon our little PC-compatible boxes from a great height).
3731 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3732 and uses that copy, rather than the current dynamic value, when
3733 it's trying to decide whether to truncate output. Thus e.g.
3734 (let ((*print-lines* 50))
3735 (pprint-logical-block (stream nil)
3737 (let ((*print-lines* 8))
3738 (print (aref possiblybigthings i) stream)))))
3739 should now truncate the logical block only at 50 lines, instead of
3740 often truncating it at 8 lines, as it did before.
3741 * The doc/cmucl/ directory, containing old CMU CL documentation
3742 from the time of the fork, is no longer part of the base system.
3743 SourceForge has shut down its anonymous FTP service, and with it
3744 my original plan for distributing the old CMU CL documentation
3745 there. For now, if you need these files you can download an old
3746 SBCL source release and extract them from it.
3747 * The fasl file version number changed again, for dozens of reasons,
3748 some of which are apparent above.
3750 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3751 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3752 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3753 patch, so that DISASSEMBLE output is much nicer.
3754 * The code in the SB-PROFILE package now seems reasonably stable.
3755 I still haven't decided what the final interface should look like
3756 (I'd like PROFILE to interact cleanly with TRACE, since both
3757 facilities use function encapsulation) but if you have a need
3758 for profiling now, you can probably use it successfully with
3759 the current CMU-CL-style interface.
3760 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3761 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3762 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3763 link now returns the dangling link itself, and for similar
3764 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3765 link itself. (In these cases the old code signalled an error and
3766 looped endlessly, respectively.) Thus, DIRECTORY now works even
3767 in the presence of dangling and cyclic symbolic links.
3768 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3769 is now a supported extension again, since the consensus on
3770 sbcl-devel was that it can be useful for ordinary development
3771 work, not just for debugging SBCL itself.
3772 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3773 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3774 that currently-defined functions might be redefined later with
3775 different return types.
3776 * Hash tables can be printed readably, as inspired by CMU CL code
3777 of Eric Marsden and SBCL code of Martin Atzmueller.
3778 * better error handling in CLOS method combination, thanks to
3779 Martin Atzmueller porting Pierre Mai's CMU CL patches
3780 * more overflow fixes for >16Mbyte I/O buffers
3781 * A bug in READ has been fixed, so that now a single Ctrl-D
3782 character suffices to cause end-of-file on character streams.
3783 In particular, now you only need one Ctrl-D at the command
3784 line (not two) to exit SBCL.
3785 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3787 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3788 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3789 does the right thing.
3790 * The implementation of some type tests, especially for CONDITION
3791 types, is now tidier and maybe faster, due to CMU CL code
3792 originally by Douglas Crosher, ported by Martin Atzmueller.
3793 * Some math functions have been fixed, and there are new
3794 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3795 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3796 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3797 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3798 likely to be built into sbcl-0.7.0.)
3799 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3800 CL, ported to SBCL by Martin Atzmueller)
3801 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3802 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3803 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3804 * There's a new slam.sh hack to shorten the edit/compile/debug
3805 cycle for low-level changes to SBCL itself, and a new
3806 :SB-AFTER-XC-CORE target feature to control the generation of
3807 the after-xc.core file needed by slam.sh.
3808 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3809 COMPILE-FILE is no longer supported, so that now every function
3810 gets an entry point, so that block compilation looks a little
3811 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3812 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3813 returns the number of bytes consed since the system started,
3814 rather than the number consed since the first time the function
3815 was called. (The new definition parallels ANSI functions like
3816 CL:GET-INTERNAL-RUN-TIME.)
3817 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3818 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3819 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3820 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3821 * Fasl file version numbers are now independent of the target CPU,
3822 since historically most system changes which required version
3823 number changes have affected all CPUs equally. Similarly,
3824 the byte fasl file version is now equal to the ordinary
3827 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3828 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3829 is no longer recognized. I apologize for this, because it was
3830 listed in SB-EXT as a supported extension, but I found that
3831 its existing behavior was poorly specified, as well as incorrectly
3832 specified, and it looked like too much of a mess to straighten it
3833 out. I have enough on my hands trying to get ANSI stuff to work..
3834 * many patches ported from CMU CL by Martin Atzmueller, with
3835 half a dozen bug fixes in pretty-printing and the debugger, and
3836 half a dozen others elsewhere
3837 * fixed bug 13: Floating point infinities are now supported again.
3838 They might still be a little bit flaky, but thanks to bug reports
3839 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3840 as flaky as they were.
3841 * The --noprogrammer command line option is now supported. (Its
3842 behavior is slightly different in detail from what the old man
3843 page claimed it would do, but it's still appropriate under the
3844 same circumstances that the man page talks about.)
3845 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3846 are now supported, and enabled by default. Thus, the compiler can
3847 handle many floating point and complex operations much less
3848 inefficiently. (Thus e.g. you can implement a complex FFT
3850 * The compiler now detects type mismatches between DECLAIM FTYPE
3851 and DEFUN better, and implements CHECK-TYPE more correctly, and
3852 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3853 thanks to patches from Martin Atzmueller.
3854 * various fixes to make the cross-compiler more portable to
3855 ANSI-conforming-but-different cross-compilation hosts (notably
3856 Lispworks for Windows, following bug reports from Arthur Lemmens)
3857 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3858 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3859 fixed thanks to Pierre Mai's CMU CL patches.
3860 * fixes to keep the system from overflowing internal counters when
3861 it tries to use i/o buffers larger than 16M bytes
3862 * fixed bug 45a: Various internal functions required to support
3863 complex special functions have been merged from CMU CL sources.
3864 (When I was first setting up SBCL, I misunderstood a compile-time
3865 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3866 * improved support for type intersection and union, fixing bug 12
3867 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3868 more obscure bugs as well
3869 * some steps toward byte-compiling non-performance-critical
3870 parts of the system, courtesy of patches from Martin Atzmueller
3871 * Christophe Rhodes has made some debian packages of sbcl at
3872 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3873 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3874 stable, but are nonetheless usable. When he's ready, I'd be happy
3875 to add them to the SourceForge "File Releases" section. (And if
3876 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3877 * new fasl file format version number (because of changes in
3878 internal representation of (OR ..) types to accommodate the new
3879 support for (AND ..) types, among other things)
3881 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3882 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3884 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3886 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3887 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3888 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3889 * DESCRIBE now works on structure objects again.
3890 * Most function call argument type mismatches are now handled as
3891 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3892 know whether the function will be redefined before the call is
3893 executed. (The compiler could flag local calls with full WARNINGs,
3894 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3895 it doesn't keep track of enough information to know whether calls
3896 are local in this sense.)
3897 * Compiler output is now more verbose, with messages truncated
3898 later than before. (There should be some supported way for users
3899 to override the default verbosity, but I haven't decided how to
3900 provide it yet, so this behavior is still controlled by the internal
3901 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3902 src/compiler/ir1util.lisp.)
3903 * Fasl file format version numbers have increased again, because
3904 support for the Gray streams extension changes the layout of the
3905 system's STREAM objects.
3906 * The Gray subclassable streams extension now works, thanks to a
3907 patch from Martin Atzmueller.
3908 * The full LOAD-FOREIGN extension (not just the primitive
3909 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3910 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3911 but like most other programs, it defaults to copying the Unix
3912 environment from the original process instead of starting the
3913 new process in an empty environment.
3914 * Extensions which manipulate the Unix environment now support
3915 an :ENVIRONMENT keyword option which doesn't smash case or
3916 do other bad things. The CMU-CL-style :ENV option is retained
3917 for porting convenience.
3918 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3919 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3921 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3923 * A patch from Martin Atzmueller seems to have solved the SIGINT
3924 problem, and as far as we know, signal-handling now works cleanly.
3925 (If you find any new bugs, please report them!)
3926 * The system no longer defaults Lisp source file names to types
3927 ".l", ".cl", or ".lsp", but only to ".lisp".
3928 * The compiler no longer uses special default file extensions for
3929 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3930 seems to expect a single default extension for all compiled code,
3931 and there's no compelling reason to try to stretch the standard
3932 to allow two different extensions.) Instead, byte-compiled files
3933 default to the same extension as native-compiled files.
3934 * Fasl file format version numbers have increased again, because
3935 a rearrangement of internal implementation packages made some
3936 dumped symbols in old fasl files unreadable in new cores.
3937 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3938 many fewer weird special cases.
3939 * Bug #17 (differing COMPILE-FILE behavior between logical and
3940 physical pathnames) has been fixed, and some related misbehavior too,
3941 thanks to a patch from Martin Atzmueller.
3942 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3943 by Tim Moore, ported to SBCL by Martin Atzmueller.
3944 * Martin Atzmueller fixed several filesystem-related problems,
3945 including bug #36, in part by porting CMU CL patches, which were
3946 written in part by Paul Werkowski.
3947 * More compiler warnings in src/runtime/ are gone, thanks to
3948 more patches from Martin Atzmueller.
3949 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3952 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3954 * DESCRIBE now works on CONDITION objects.
3955 * The debugger now handles errors which arise when trying to print
3956 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3958 * The build system now uses an additional file, customize-target-features.lisp,
3959 to allow local modifications to the target *FEATURES* list. (The point of
3960 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3961 debugging features enabled, without having to worry about propagating it
3962 into everyone's system when I do a "cvs update".) When no
3963 customize-target-features.lisp file exists, the target *FEATURES* list
3964 should be constructed the same way as before.
3965 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3966 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3967 warn about behavior which is undefined under the ANSI spec. Note: This
3968 is specified by ANSI, but it's not very popular with programmers.
3969 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3970 macro in the SBCL sources for an example of a workaround which you
3971 might use to make portable ANSI-standard code which does what you want.
3972 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3973 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3974 when it's used as a non-toplevel form. (This is required in order
3975 to implement the DEFCONSTANT-EQX macro.)
3976 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3977 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3978 funny things, due to bugs in the implementation of EVAL-WHEN
3979 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3980 non-toplevel DEFCONSTANTs are uncommon.)
3981 * The core file version number and fasl file version number have been
3982 incremented, because the old noncompliant DEFCONSTANT behavior involved
3983 calling functions which no longer exist, and because I also took the
3984 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3985 * fixed bug 1 (error handling before read-eval-print loop starts), and
3986 redid debugger restarts and related debugger commands somewhat while
3988 ** The QUIT debugger command is gone, since it did something
3989 rather different than the SB-EXT:QUIT command, and since it never
3990 worked properly outside the main toplevel read/eval/print loop.
3991 Invoking the new TOPLEVEL restart provides the same functionality.
3992 ** The GO debugger command is also gone, since you can just invoke
3993 the CONTINUE restart directly instead.
3994 ** The TOP debugger command is also gone, since it's redundant with the
3995 FRAME 0 command, and since it interfered with abbreviations for the
3997 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3999 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4000 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4001 some porting work by Martin Atzmueller)
4002 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4003 with a boatload of patches which he ported to SBCL. Now that those
4005 ** The system tries to make sure that its low-priority messages
4006 are prefixed by semicolons, to help people who like to use
4007 syntax highlighting in their ILISP buffer. (This patch
4008 was originally due to Raymond Toy.)
4009 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4010 patches originally written by Raymond Toy.
4011 ** The compiler understands coercion between single-value and
4012 multiple-VALUES type expressions better, getting rid of some very
4013 weird behavior, thanks to patches originally by Robert MacLachlan
4014 and Douglas Crosher.
4015 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4016 lambda lists, thanks to a patch originally by Pierre Mai.
4017 ** The system no longer bogusly warns about "abbreviated type
4019 ** The compiler gets less confused by inlining and RETURN-FROM,
4020 thanks to some patches originally by Tim Moore.
4021 ** The system no longer hangs when dumping circular lists to fasl
4022 files, thanks to a patch originally from Douglas Crosher.
4023 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4024 undefined function error.
4025 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4026 in structure slot initforms) for now, documented workaround instead:-|
4027 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4028 * fixed bug 5: added stubs for various Gray stream functions called
4029 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4030 installed, at least appropriate type errors are generated
4031 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4032 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4033 consistently in DEFMETHOD forms.
4034 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4035 it doesn't seem to affect SBCL after all
4036 * The C runtime system now builds with better optimization and many
4037 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4039 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4041 * The system is now under CVS at SourceForge (instead of the
4042 CVS repository on my home machine).
4043 * The new signal handling code has been tweaked to treat register
4044 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4045 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4046 downstream, e.g. in debug-int.lisp, has implicit dependencies
4047 on the unsignedness of integer representation of machine words,
4048 and that caused the system to bomb out with infinite regress
4049 when trying to recover from type errors involving signed values,
4050 e.g. (BUTLAST '(1 2 3) -1).)
4051 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4052 (This was one of the bugs Peter Van Eynde reported back in July.)
4053 * The system now uses code inspired by Colin Walters' O(N)
4054 implementation of MAP (from the cmucl-imp@cons.org mailing
4055 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4056 inline the MAP operation, and there is more than one
4057 sequence argument to the MAP call (so that it can't just
4058 do ETYPECASE once and for all based on the type of the
4059 single sequence argument). (The old non-inline implementation
4060 of the general M-argument sequence-of-length-N case required
4061 O(M*N*N) time when any of the sequence arguments were LISTs.)
4062 * The QUIT :UNIX-CODE keyword argument has been renamed to
4063 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4064 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4066 * Raymond Wiker's patches to port dynamic loading from Linux to
4067 FreeBSD have been added.
4068 * The BUGS file is now more nearly up to date, thanks in large part
4069 to Martin Atzmueller's review of it.
4070 * The debugger now flushes standard output streams before it begins
4071 its output ("debugger invoked" and so forth).
4072 * The core version number and fasl file version number have both
4073 been incremented, because of incompatible changes in the layout
4075 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4076 used to not be called for a saved Lisp image.)
4077 * Martin Atzmueller's version of a patch to fix a compiler crash,
4078 as posted on sbcl-devel 13 September 2000, has been installed.
4079 * Instead of installing Martin Atzmueller's patch for the
4080 compiler transform for SUBSEQ, I deleted the compiler transform,
4081 and transforms for some similar consing operations.
4082 * A bug in signal handling which kept TRACE from working on OpenBSD
4084 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4085 away by constant folding
4086 * The system now defines its address space constants in one place
4087 (in the Lisp sources), and propagates them automatically elsewhere
4088 (through GENESIS and the sbcl.h file). Therefore, patching the
4089 address map is less unnecessarily tedious and error-prone. The
4090 Lisp names of address space constants have also been systematized.
4091 * CVS tags like dollar-Header-dollar have been removed from
4092 the sources, because they have never saved me trouble and
4093 they've been source of trouble working with patches and other
4094 diff-related operations.
4095 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4096 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4098 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4100 * The system has been ported to OpenBSD.
4101 * The system now compiles with a simple "sh make.sh" on the systems
4102 that it's supported on. I.e., now you no longer need to tweak
4103 text in the target-features.lisp-expr and symlinks in src/runtime/
4104 by hand, the make.sh takes care of it for you.
4105 * The system is no longer so grossly inefficient when compiling code
4106 involving vectors implemented as general (not simple) vectors (VECTOR T),
4107 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4108 which dares to use the various sequence functions on non-simple
4109 vectors, takes less of a performance hit.
4110 * There is now a primitive type predicate VECTOR-T-P
4111 to test for the (VECTOR T) type, so that e.g.
4112 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4113 can now be compiled with some semblance of efficiency. (The old code
4114 turned the type declaration into a full call to %TYPEP at runtime!)
4115 * AREF on (VECTOR T) is still not fast, since it's still compiled
4116 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4117 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4118 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4119 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4120 without making full calls to SUBTYPEP at runtime.
4121 (Some analogous efficiency issues for non-simple vectors specialized to
4122 element types other than T, or for non-simple multidimensional arrays,
4123 have not been addressed. They could almost certainly be handled the
4124 same way if anyone is motivated to do so.)
4125 * The changes in array handling break binary compatibility, so
4126 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4127 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4128 instead of (VALUES T T).
4129 * By following the instructions that Dan Barlow posted to sbcl-devel
4130 on 2 July 2000, I was able to enable primitive dynamic object
4131 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4132 functionality is not implemented (since it calls ld to resolve
4133 library references automatically, requiring RUN-PROGRAM for its
4134 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4135 try to resolve library references) is now supported.
4136 * The system now flushes the standard output streams when it terminates,
4137 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4138 them at several other probably-convenient times, e.g. in each pass of
4139 the toplevel read-eval-print loop, and after evaluating a form given
4140 as an "--eval" command-line option. (These changes were motivated by a
4141 discussion of stream flushing issues on cmucl-imp in August 2000.)
4142 * The source transform for TYPEP of array types no longer assumes
4143 that an array whose element type is a not-yet-defined type
4144 is implemented as an array of T, but instead punts, so that the
4145 type will be interpreted at runtime.
4146 * There is now some support for cross-compiling in make.sh: each of
4147 the phases of make.sh has its own script. (This should be transparent
4148 to people doing ordinary, non-cross-compile builds.)
4149 * Since my laptop doesn't have hundreds of megabytes of memory like
4150 my desktop machine, I became more motivated to do some items on
4151 my to-do list in order to reduce the size of the system a little:
4152 ** Arrange for various needed-only-at-cold-init things to be
4153 uninterned after cold init. To support this, those things have
4154 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4155 symbols with such names are now uninterned after cold init).
4156 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4157 fasl files for cold load.
4158 ** Remove the old compiler structure pooling code (which used to
4159 be conditional on the target feature :SB-ALLOC) completely.
4160 ** Redo the representation of some data in cold init to be more compact.
4161 (I also looked into supporting byte compiled code at bootstrap time,
4162 which would probably reduce the size of the system a lot, but that
4163 looked too complicated, so I punted for now.)
4164 * The maximum signal nesting depth in the src/runtime/ support code has
4165 been reduced from 4096 to 256. (I don't know any reason for the very
4166 large old value. If the new smaller value turns out to break something,
4167 I'll probably just bump it back up.)
4168 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4170 * Many, many bugs reported by Peter Van Eynde have been added to
4171 the BUGS list; some have even been fixed.
4172 * While enabling dynamic object file loading, I tried to make the
4173 code easier to understand, renaming various functions and variables
4174 with less ambiguous names, and changing some function calling
4175 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4176 * While trying to figure out how to do the OpenBSD port, I tried to
4177 clean up some of the code in src/runtime/. In particular, I dropped
4178 support for non-POSIX signal handling, added various comments,
4179 tweaked the code to reduce the number of compilation warnings, and
4180 renamed some files to increase consistency.
4181 * To support the new automatic configuration functionality in make.sh,
4182 the source file target-features.lisp-expr has been replaced with the
4183 source file base-target-features.lisp-expr and the machine-generated
4184 file local-target-features.lisp-expr.
4185 * fixed a stupid quoting error in make.sh so that using CMU CL
4186 "lisp -batch" as cross-compilation host works again
4188 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4190 * DESCRIBE no longer tries to call itself recursively to describe
4191 bound/fbound values, so that it no longer fails on symbols which are
4192 bound to themselves (like keywords, T, and NIL).
4193 * DESCRIBE now works on generic functions.
4194 * The printer now prints less-screwed-up representations of closures
4195 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4196 * A private symbol is used instead of the :EMPTY keyword previously
4197 used to mark empty slots in hash tables. Thus
4198 (DEFVAR *HT* (MAKE-HASH-TABLE))
4199 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4200 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4201 now does what ANSI says that it should. (You can still get
4202 similar noncompliant behavior if bang on the hash table
4203 implementation with all the symbols you get back from
4204 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4205 This breaks binary compatibility, since tests for equality to
4206 :EMPTY are wired into things like the macroexpansion of
4207 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4209 * There's now a minimal placeholder implementation for CL:STEP,
4210 as required by ANSI.
4211 * An obscure bug in the interaction of the normal compiler, the byte
4212 compiler, inlining, and structure predicates has been patched
4213 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4214 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4216 * Missing ordinary arguments in a macro call are now detected even
4217 when the macro lambda list contains &KEY or &REST.
4218 * The debugger no longer complains about encountering the top of the
4219 stack when you type "FRAME 0" to explicitly instruct it to go to
4220 the top of the stack. And it now prints the frame you request even
4221 if it's the current frame (instead of saying "You are here.").
4222 * As specified by ANSI, the system now always prints keywords
4223 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4225 * The default initial SIZE of HASH-TABLEs is now smaller.
4226 * Type information from CLOS class dispatch is now propagated
4227 into DEFMETHOD bodies, so that e.g.
4228 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4230 is now basically equivalent to
4231 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4232 (DECLARE (TYPE SINGLE-FLOAT X))
4234 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4235 operation, without having to do run-time type dispatch.
4236 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4237 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4238 * Since I don't seem to be making big file reorganizations very often
4239 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4240 is overflowing my ability to conveniently back them up), I've finally
4241 checked the system into CVS. (The CVS repository is on my home system,
4242 not at SourceForge -- putting it on SourceForge might come later.)
4243 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4244 high-level GC-NOTIFY-FOO functions send their output. (There's
4245 still very little control of where low-level verbose GC functions
4246 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4247 less than it used to -- the GC-NOTIFY-FOO functions are now under
4248 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4249 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4250 in only one place in the source code, and propagates it automatically
4251 everywhere that it's needed. Thus e.g. when I bump the version from
4252 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4253 * The C source files now include boilerplate legalese and documentation
4254 at the head of each file (just as the Lisp source files already did).
4255 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4256 to his page will be replaced with a link to his new CLiki service.
4258 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4260 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4261 * The build process now looks for GNU make under the default name "gmake",
4262 instead of "make" as it used to. If GNU make is not available as "gmake"
4263 on your system, you can change this default behavior by setting the
4264 GNUMAKE environment variable.
4265 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4266 can build without error under CMU CL.
4268 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4270 * There is now a partial SBCL user manual (with some new text and some
4271 text cribbed from the CMU CL manual).
4272 * The beginnings of a profiler have been added (starting with the
4273 CMU CL profiler and simplifying and cleaning up). Eventually the
4274 main interface should be through the TRACE macro, but for now,
4275 it's still accessed through vaguely CMU-CL-style functions and macros
4276 exported from the package SB-PROFILE.
4277 * Some problems left over from porting CMU CL to the new
4278 cross-compilation bootstrap process have been cleaned up:
4279 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4280 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4281 part when getting rid of the compiler *BACKEND* stuff.)
4282 ** The value of *NULL-TYPE* was screwed up, because it was
4283 being initialized before the type system knew the final
4284 definition of the 'NULL type. This screwed up several key
4285 optimizations in the compiler, causing inefficiency in all sorts
4286 of places. (I found it because I wanted to understand why
4287 GET-INTERNAL-RUN-TIME was consing.)
4288 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4289 PROCLAIM FTYPE information. Unfortunately this broke binary
4290 compatibility again, since now the forms output by DEFGENERIC
4291 to refer to functions which didn't exist in 0.6.3.
4292 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4293 can use the new (as of 0.6.3) transform for SOME into MAP into
4295 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4296 values to (INTEGER 0 1000000), so that the time code will no longer
4297 occasionally get blown up by Linux returning 1000000 microseconds
4298 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4299 its output conform to the ANSI spec. (Alas, this makes its output
4300 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4301 * A full call to MAP NIL with a single sequence argument no longer conses.
4302 * fixes to problems pointed out by Martin Atzmueller:
4303 * The manual page no longer talks about multiprocessing as though
4304 it were currently supported.
4305 * The ILISP support patches have been removed from the distribution,
4306 because as of version 5.10.1, ILISP now supports SBCL without us
4307 having to maintain patches.
4308 * added a modified version of Raymond Toy's recent CMU CL patch for
4309 EQUALP comparison of HASH-TABLE
4311 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4313 * The system still can't cross-compile itself with
4314 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4315 entails), but at least it can compile more of itself that way
4316 than it used to be able to, and various buglets which were uncovered
4317 by trying to cross-compile itself that way have now been fixed.
4318 * This release breaks binary compatibility again. This time
4319 at least I've incremented the FASL file format version to 2, so that the
4320 problem can be detected reliably instead of just causing weird errors.
4321 * various new style warnings:
4322 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4323 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4324 ** using the lexical binding of a variable named in the *FOO* style
4325 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4326 as specified by ANSI.
4327 * *RANDOM-STATE* is no longer automatically initialized from
4328 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4329 default behavior of the system is to repeat its behavior every time
4330 it's run. If you'd like to change this behavior, you can always
4331 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4332 old convention there was no comparably easy way to get the system to
4333 repeat its behavior every time it was run.
4334 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4335 FTYPE declarations has been removed, in favor of their ANSI
4336 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4337 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4338 the types of their sequence arguments can be determined at compile time.
4339 This is done through a new open code expansion for MAP which eliminates
4340 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4341 argument types can be determined at compile time.
4342 * The optimizer now transforms COERCE into an identity operation when it
4343 can prove that the coerced object is already of the correct type. (This
4344 can be a win for machine generated code, including the output of other
4345 optimization transforms, such as the MAP transform above.)
4346 * Credit information has been moved from source file headers into CREDITS.
4347 * Source file headers have been made more standard.
4348 * The CASE macro now compiles without complaining even when it has
4351 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4353 * (Note that the way that the PCL macroexpansions were rewritten
4354 to accommodate the change in DEFGENERIC below breaks binary
4355 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4356 not run under sbcl-0.6.2. Once we get out of alpha releases,
4357 i.e. hit release 1.0.0, we'll probably try to maintain binary
4358 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4359 and sbcl-1.4.4. Until then, however, it might be fairly common
4360 for maintenance releases to break binary compatibility.)
4361 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4363 * The reporting of unhandled signals has been changed to print some
4364 explanatory text as well as the report form. (Previously only
4365 the report form was printed.)
4366 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4367 it defines, so that the compiler no longer issues undefined function
4368 warnings for compiled-but-not-yet-loaded generic functions.
4369 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4370 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4371 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4372 section of the ANSI spec are supported.
4373 * The "" nickname for the "KEYWORD" package has been removed.
4374 The reader still handles symbol tokens which begin with a package marker
4375 as keywords, but it doesn't expose its mechanism for doing so in the
4376 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4377 * The system now issues STYLE-WARNINGs for contradictory TYPE
4378 proclamations. (Warnings for contradictory FTYPE proclamations would
4379 be nice too, but those can't be done usefully unless the type system
4380 is made smarter about FUNCTION types.)
4381 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4382 systematized, so that "*target-*.lisp is supposed to exist only on the
4383 target and imply that there's a related file which exists on the
4384 host, and *host-*.lisp is supposed to exist only on the host and imply
4385 that there's a related file which exists on the target. This involves a
4386 lot of renaming. Hopefully the acute confusion caused by the renaming
4387 will be justified by the reduction in chronic confusion..
4388 ** runtime-type.lisp -> early-target-type.lisp
4389 ** target-type.lisp -> late-target-type.lisp
4390 ** early-host-format.lisp -> early-format.lisp
4391 ** late-host-format.lisp -> late-format.lisp
4392 ** host-error.lisp -> misc-error.lisp
4393 ** early-error.lisp -> early-target-error.lisp
4394 ** late-error.lisp -> late-target-error.lisp
4395 ** host-defboot.lisp -> early-defboot.lisp
4396 ** code/misc.lisp -> code/target-misc.lisp
4397 ** code/host-misc.lisp -> code/misc.lisp
4398 ** code/numbers.lisp -> code/target-numbers.lisp
4399 ** code/early-numbers.lisp -> numbers.lisp
4400 ** early-host-type.lisp -> early-type.lisp
4401 ** late-host-type.lisp -> late-type.lisp
4402 ** host-typep.lisp -> typep.lisp
4403 ** load.lisp -> target-load.lisp
4404 ** host-load.lisp -> load.lisp
4405 ** host-disassem.lisp -> disassem.lisp
4406 ** host-insts.lisp -> insts.lisp
4407 ** byte-comp.lisp -> target-byte-comp.lisp
4408 ** host-byte-comp.lisp -> byte-comp.lisp
4409 ** host-signal.lisp -> signal.lisp
4410 ** host-defstruct.lisp -> defstruct.lisp
4411 ** late-target-type.lisp -> deftypes-for-target.lisp
4412 Furthermore, several other previously target-only files foo.lisp (e.g.
4413 hash-table.lisp and random.lisp) have been split into a target-and-host
4414 foo.lisp file and a target-only target-foo.lisp file, with their key type
4415 definitions in the target-and-host part, so that the cross-compiler will
4416 know more about target types.
4417 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4418 gone away. In their place are various *BACKEND-FOO* variables
4419 corresponding to the slots of the old structure.
4420 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4421 into the target SBCL has been fixed.
4422 * The chill.lisp system for loading cold code into a running SBCL
4424 * Support for the CMU CL "scavenger hook" extension has been removed.
4425 (It was undocumented and unused in the CMU CL sources that SBCL was
4426 derived from, and stale in sbcl-0.6.1.)
4427 * Various errors in the cross-compiler type system were detected
4428 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4429 (enabling various consistency checks). Many of them were fixed,
4430 but some hard problems remain, so the compiler is back to
4431 running without *TYPE-SYSTEM-INITIALIZED* for now.
4432 * As part of the cross-compiler type system cleanup, I implemented
4433 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4434 * I have started adding UNCROSS calls throughout the type system
4435 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4436 on cross-compiler input files will be able to go away, and various
4438 * CONSTANTP now returns true for quoted forms (as explicitly required
4441 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4443 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4444 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4445 (MAKE-LIST -1). In the longer term, it ought to become true
4446 that declarations are assertions even at SAFETY 1. For now, it's not
4447 quite true even at SAFETY 3, but it's at least more nearly true..
4448 (Note that this change seems to increases the size of the system by
4449 O(5%) and to decrease the speed of the compiler by 20% or more.)
4450 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4451 for the problem of printing dozens of lines of distracting information
4452 about low-level system machinery as part of the top stack frame
4453 on entry to the debugger when an undefined function was called.
4454 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4455 is not reset to COMMON-LISP-USER.
4456 * Compilation of stuff related to dyncount.lisp has been made conditional
4457 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4458 smaller. The various dyncount-related symbols have been moved into
4459 a new "SB-DYNCOUNT" package.
4460 * tty-inspect.lisp has been renamed to inspect.lisp.
4461 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4462 feature has gone away. (When we eventually port to other flavors of
4463 libc and/or Unix, we'll try to make the differences between flavors
4464 invisible at the user level.)
4465 * Various other *FEATURES* tags, and/or their associated conditionals,
4466 have been removed if obsolescent, or given better documentation, or
4467 sometimes given more-mnemonic names.
4469 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4471 * tidied up "make.sh" script
4472 * tidied up system directory structure
4473 * better "clean.sh" behavior
4474 * added doc/FOR-CMUCL-DEVELOPERS
4475 * many many small tweaks to output format, e.g. removing possibly-confusing
4476 trailing #\. character in DESCRIBE-INSTANCE
4477 * (EQUALP #\A 'A) no longer signals an error.
4478 * new hashing code, including EQUALP hashing
4479 * tidied up Lisp initialization and toplevel
4480 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4481 * command line argument processing
4482 * added POSIX-GETENV function to deal with Unix-ish environment variables
4483 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4484 terminating SBCL on EOF
4485 * non-verbose GC by default
4486 * There is no more "sbcl" shell script; the sbcl file is now the C
4487 runtime executable (just like CMU CL).
4488 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4490 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4491 symbol and package information as Lisp native symbol and package objects
4492 instead of strings naming symbols and strings naming packages. This way,
4493 whenever packages are renamed (as in warm init), debug information is
4494 transformed along with everything else.
4495 * tweaked the optimization policy declarations which control the building
4496 of SBCL itself. Now, among other things, the system no longer saves
4497 source location debugging information. (This helps two problems at once
4498 by reducing SBCL size and by keeping SBCL from trying to look for its
4499 sources -- which may not exist -- when reporting errors.)
4500 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4501 debugging and testing purposes
4502 * cleaned up printing, making the printer call PRINT-OBJECT for
4503 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4504 methods, giving nearly-ANSI behavior
4505 * converted almost all special variables to use *FOO* naming convention
4506 * deleted PARSE-TIME functionality, since it can be done portably
4507 * moved some files out of cold init into warm init
4508 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4510 * regularized formatting of source files
4511 * added an install.sh script
4512 * fixed ridiculous memory usage of cross-compiler by making
4513 compiler/alloc.lisp not try to do pooling unless it can hook
4514 itself into the GC of the cross-compilation host. Now the system
4515 builds nicely on my old laptop.
4516 * added :SB-ALLOC in target-features.lisp-expr
4517 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4518 was not implemented)
4519 * re-did condition handling and note reporting in the compiler. Notes
4520 are no longer handled by signalling conditions. Style warnings
4521 and warnings are handled more correctly and reported in such a way
4522 that it's easy to find one or the other in your output (so that you
4523 can e.g. figure out which of many problems caused COMPILE-FILE to
4525 * changed the severity of several compiler warnings from full WARNING
4526 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4527 compiler note reporting so that it doesn't use the condition system
4528 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4529 COMPILE-FILE command)
4530 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4531 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4532 no longer supported, since their implementation was deeply intertwingled
4533 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4534 * removed bogus "support" for compiler macros named (SETF FOO), and
4535 removed the compiler macro for SETF INFO (but only after making a fool
4536 of myself on the cmucl-imp mailing list by posting a bogus patch for
4537 DEFINE-COMPILER-MACRO..)
4538 * Compiled files containing forms which have side effects on the Lisp
4539 reader (such as DEFPACKAGE forms) are now handled more correctly.
4540 (Compiler queuing of top level lambdas has been suppressed by setting
4541 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4542 * deleted various currently-unused source files, e.g. gengc.lisp. They
4543 may be added back at some point e.g. when porting to other architectures,
4544 but until they are it's distracting to distribute them and to try to
4546 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4547 there were so many of them they're just distractions, and UNCROSS is
4548 known to be able to handle the current sources
4549 * moved PROFILE functionality into TRACE, so that it will be clear
4550 how the wrapping and unwrapping of functions when you profile them
4551 interacts with the wrapping and unwrapping of functions when you
4552 trace them. (Actually, the functionality isn't there yet, but at least
4553 the interface specification is there. Hopefully, the functionality will
4554 arrive with some future maintenance release.)
4555 * removed host-oops.lisp
4556 * changed signature of QUIT function to allow UNIX-CODE argument
4557 * fixed READ-SEQUENCE bug
4558 * tweaked verbose GC output so that it looks more like the progress
4559 output that ANSI specifies for functions like LOAD
4560 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4561 * added <http://sbcl.sourceforge.com> to the banner information printed by