1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
3 * new feature: MAKE-HASH-TABLE now experimentally accepts a
4 :SYNCHRONIZED argument, which makes the hash-table safe for
5 concurrent accesses (but not iteration.) See also:
6 SB-EXT:WITH-LOCKED-HASH-TABLE, and
7 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
8 * optimization: CONCATENATE on strings is an order of magnitue faster
9 in code compiled with (> SPEED SPACE).
10 * optimization: SUBSEQ is ~50% faster on lists.
11 * bug fix: SUBSEQ on a list will now correctly signal an error if if
12 END is smaller then START.
13 * bug fix: SB-PROFILE will no longer report extra consing for nested
14 calls to profiled functions.
15 * bug fix: ROOM implementation had bogus fixnum declarations which
16 could cause type-errors when calling ROOM on large images.
17 * bug fix: if file compilation is aborted, the partial fasl is now
18 deleted, and COMPILE-FILE returns NIL as the primary value.
19 * bug fix: number of thread safety issues relating to SBCL's internal
20 hash-table usage have been fixed.
21 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
22 be returned from its body when the values were being returned
23 using unknown-values return convection and the W-P-O was wrapped
24 inside an UNWIND-PROTECT.
25 * bug fix: sb-posix should now compile again under Windows, enabling
28 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
29 * incompatible change: hash-table accessor functions are no longer
30 automatically protected by locks. Concurrent accesses on the same hash-table
31 from multiple threads can give inconsistent results or even corrupt the
32 hash-table completely. Multi-threaded applications should do their own
33 locking at the correct granularity. In the current implementation it is
34 still safe to have multiple readers access the same table, but it's not
35 guaranteed that this property will be maintained in future releases.
36 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
37 and will signal an error at runtime.
38 * enhancement: SB-THREAD package now exports a semaphore interface.
39 * enhancement: CONS can now stack-allocate on x86 and
40 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
42 * enhancement: nested lists can now be stack allocated on
43 platforms providing stack allocation support.
44 * enhancement: dynamic-extent support has been extended to support
45 cases where there are multiple possible sources for the stack
47 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
48 if the mutex is uncontested on Linux.
49 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
50 as specified by the CLHS. (thanks to Tobias Rittweiler)
51 * bug fix: NaN comparison now works on x86-64.
52 * bug fix: CLOSE :ABORT T on a stream with pending output now
54 * bug fix: instances of non-standard metaclasses using standard
55 instance structure protocol sometimes missed the slot type checks
57 * bug fix: known functions can be inlined in high-debug code.
59 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
60 * minor incompatible change: the MSI installer on Windows no longer
61 associates .lisp and .fasl files with the installed SBCL.
62 * minor incompatible change: :UNIX is no longer present in *FEATURES*
63 on Windows. (thanks to Luis Oliviera)
64 * new feature: SB-CLTL2 contrib module now implements
65 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
66 * optimization: scavenging weak pointers is now more efficient,
67 requiring O(1) instead of O(N) per weak pointer to identify
68 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
69 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
70 method bodies using constant slot-names when the first argument is
71 a specializer parameter for the method.
72 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
73 names, when the class of the instance is a direct instance
74 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
75 as slow as the constant slot-name case.)
76 * optimization: member type construction is now O(N) instead
78 * optimization: UNION and NUNION are now O(N+M) for large
79 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
81 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
82 now more readable in environments like Slime which display it.
83 (thanks to Tobias C. Rittweiler)
84 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
85 was able to infer the correct argument type for the object on which
86 the CAS operation was being performed.
87 * bug fix: copy propagation interfered with parallel assignment
88 semantics in local calls. (reported by Paul Khuong)
89 * bug fix: the signed modular fixnum shift compiled to wrong code on
90 x86 and x86-64. (spotted by a slight modification to some of
93 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
94 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
95 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
96 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
98 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
99 and generic function now print the object identity for anonymous
100 instances. (thanks to Attila Lendvai)
101 * enhancement: as an extension to MOP, SBCL now supports SETF of
102 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
103 (thanks to Attila Lendvai)
104 * enhancement: sb-cover has an alternate annotation mode which puts
105 the form annotations at the car of the form, instead of annotating
106 it completely. The benefit of this mode is that it shows better
107 which forms were instrumented by the compiler.
108 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
109 by printer control variables. (reported by Dan Corkill)
110 * bug fix: system leaked memory when delayed output was performed by
111 the OS in smaller chunks then expected. (thanks to David Smith)
112 * bug fix: system leaked memory when file streams were not closed
114 * bug fix: large objects written to slow streams that were modified
115 after the write could end up with the modified state written to
116 the underlying file descriptor.
117 * bug fix: multiple threads operating in parallel on the same stream
118 could cause buffer-overflows.
119 * bug fix: source location information is stored correctly
120 (broken since 1.0.6). This bug would generally show up as the
121 Slime debugger higlighting the wrong form when the sldb-show-source
123 * bug fix: Branch forms are again annotated as branches in the sb-cover
125 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
126 on x86-64. (reported by Gregory Vanuxem)
127 * bug fix: EQUALP could return wrong results for structures with raw
128 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
129 word). (reported by Vjacheslav Fyodorov)
130 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
133 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
134 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
135 atomic compare-and-swap operations on threaded platforms.
136 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
137 allows assining a global minimum value to optimization qualities
138 (overriding proclamations and declarations).
139 * enhancement: closed over variables can be stack-allocated on x86
141 * performance bug fix: GETHASH and (SETF GETHASH) are once again
143 * optimization: slot definition lookup is now O(1). This speeds up
144 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
145 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
146 * optimization: ASSOC and MEMBER can now be open-coded for all
147 combinations of keyword arguments when second argument is constant
148 and SPEED >= SPACE. In other cases a specialized version is
150 * bug fix: using obsoleted structure instances with TYPEP and
151 generic functions now signals a sensible error.
152 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
153 (reported by Kristoffer Kvello)
154 * bug fix: backtrace construction is now more careful when making
155 lisp-objects from pointers on the stack, to avoid creating bogus
156 objects that can be seen by the GC.
157 * bug fix: defaulting of values in contexts expecting more than 7
158 variables now works on x86-64. (reported by Christopher Laux)
159 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
161 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
162 as the property-list of a symbol.
163 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
164 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
165 situations requested, are once again file-compileable. (reported
168 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
169 * MOP improvement: support for user-defined subclasses of
170 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
171 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
172 the expansion of DEFMETHOD, is responsible for generating a form
173 which creates a list of specializers when evaluated. Additional
174 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
175 debugging and introspective support.
176 * minor incompatible change: the (unsupported) spinlock interface
177 has changed: free spinlock now has the value NIL, and a held spinlock
178 has the owning thread as its value.
179 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
180 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
182 * enhancement: name of a socket-stream is now "a socket" instead of
184 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
185 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
186 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
187 for generic functions. (Thanks to Tobias C. Rittweiler)
188 * optimization: bignum printing speed has been improved by 20-40%
189 (depending on the bignum size.)
190 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
192 * bug fix: the cache used by the CLOS to store precomputed effective
193 methods, slot offsets, and constant return values is now thread and
195 * bug fix: generic function dispatch function updating is now thread
196 and interrupt safe (in the sense that the known issues have been
198 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
199 * bug fix: interning EQL-specializers is now thread and interrupt safe.
200 * bug fix: asdf systems with dependencies to the SB-POSIX or
201 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
202 * bug fix: interrupt safety of applicable method computation has been
205 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
206 * new contrib: sb-cover, an experimental code coverage tool, is included
208 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
209 significantly faster.
210 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
211 produce more efficient code on x86-64 if the compiler can determine
212 that (+ INDEX <constant>) does not require a bounds check and FOO
213 has an element type at least 8 bits wide.
214 * enhancement: a new, experimental synchronous timeout facility is
215 provided. Refer to SB-SYS:WITH-DEADLINE for details.
216 * enhancement: when a symbol name conflict error arises, the
217 conflicting symbols are always printed with a package prefix.
218 (thanks to Kevin Reid)
219 * enhancement: stepping is now once again supported on the SPARC and
220 MIPS platforms. (It is also now more likely to work on CheneyGC
222 * enhancement: sb-sprof can now also track and report accurate call
224 * bug fixes: the treatment of non-standard subclasses of
225 SB-MOP:SPECIALIZER is more correct.
226 * incompatible change: PURIFY no longer copies the data from the
227 dynamic space into the static and read-only spaces on platforms
228 that use the generational garbage collector
229 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
231 * bug fix: GC race condition occasionally resulting in crashes with
232 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
234 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
236 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
237 a STYLE-WARNING to be signalled (regression from 1.0.4.)
238 * bug fix: an asynchronous interrupt could previously leave the
239 system running with GC inhibited.
240 * bug fix: a DECLARE form evaluated at top-level now causes an error
241 rather than silently (or verbosely) returning NIL.
242 * bug fix: trying to dissassemble functions compiled at high DEBUG could
243 result in a error being signalled due to source form lookup errors.
244 (reported by Peter Graves)
246 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
247 * incompatible change: removed writer methods for host-ent-name,
248 host-ent-addresses -- changing the values did not update the DNS
250 * minor incompatible change: changed experimental JOIN-THREAD interface
251 * documentation: the manual now lists reader and writer methods
252 in class slot documentation sections. (thanks to Richard M Kreuter)
253 * documentation: unwinding from asyncronous events has been
254 documented as unsafe.
255 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
256 in multithreaded application code.
257 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
258 platforms. (thanks to James Anderson for the optimization hint)
259 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
261 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
262 variants no longer cons.
263 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
264 their NOT- variants no longer cons.
265 * optimization: EQUAL hash tables no longer use SXHASH for objects
266 of all data types, but instead use an EQL hash for types for which
267 EQUAL is the same as EQL.
268 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
269 are significantly faster.
270 * optimization: new STRING-OUTPUT-STREAM implementation is much
271 faster for large outputs and conses 30% less on average.
272 * enhancement: XREF information is now collected for references made
273 to global variables using SYMBOL-VALUE and a constant argument.
274 * enhancement: SIGINT now causes a specific condition
275 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
276 * bug fix: FILE-POSITION used to signal an error in some cases where
277 ANSI requires it to return NIL.
278 * bug fix: ADJUST-ARRAY is now interrupt-safe.
279 * bug fix: adding and removing fd-handlers is now interrupt-safe.
280 * bug fix: inlined calls to C now ensure 16byte stack alignment on
282 * bug fix: bad type declaration in the CLOS implementation has
283 been fixed. (reported by James Anderson)
284 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
285 (thanks to Lutz Euler)
286 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
287 required by the CLHS. (reported by Samium Gromoff)
288 * bug fix: dead unbound variable references now signal an error.
289 (reported by Marco Monteiro)
290 * bug fix: / with an unused value was being deleted in safe code.
291 (thanks to Marco Monteiro and Kevin Reid)
292 * bug fix: number of characters that can be written onto a single
293 line in a file is unlimited.
294 * bug fix: some GC deadlocks caused by asynchronous interrupts have
295 been fixed by inhibiting interrupts for when GC is disbled.
296 * bug fix: some interrupt safety issues with GETHASH, (SETF
297 GETHASH), CLRHASH and REMHASH have been fixed.
298 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
299 specifier no longer causes infinite recursion.
300 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
301 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
302 * bug fix: an &environment argument with the correct variable information
303 is passed to macros that are expanded during byte compilation.
304 (reported by Samium Gromoff)
305 * bug fix: a logic error was causing the hash function for bignums to
306 have excessive amounts of collisions. (reported by Faré Rideau)
307 * bug fix: modifying the contents of an array could change the return
308 value of SXHASH on that array, which is only allowed for strings
309 and bit vectors. (bug introduced in 0.9.16)
310 * improvement: the x86-64/darwin port now passes all tests (except
311 for the debugger tests) but should still be considered
312 experimental until this is fixed.
313 * improvement: a style-warning is signaled for CASE (etc) clauses with
314 duplicate keys. (thanks to Kevin Reid)
315 * improvement: macos/x86 and macos/x86-64 now use mach exception
316 handlers for dealing with illegal instructions (for trapping and
317 error handling) and memory protection violations (for GC).
319 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
320 * new platform: experimental support for x86-64/darwin (MacOS).
321 * incompatible change: the thread-safe (on most platforms) getaddrinfo
322 and getnameinfo sockets functions are used instead of gethostbyaddr
323 and gethostbyname, on platforms where the newer functions are available.
324 As a result, the ALIASES field of HOST-ENT will always be NIL on these
326 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
327 don't print the startup banner, but behave as if --noinform was passed
328 as a command line argument. (thanks to Kevin Reid)
329 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
330 * optimization: code using alien values with undeclared types is much faster.
331 * optimization: the compiler is now able to open code SEARCH in more cases.
332 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
333 * bug fix: using standardized COMMON-LISP special variables as loop
334 variables no longer signals bogus package lock violations. (reported
336 * bug fix: declaring local loop variables to be of a range-limited type
337 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
338 (reported by Andras Simon)
339 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
340 bugs remain on x86-64.)
341 * bug fix: the #= and ## reader macros now interact reasonably with
342 funcallable instances.
343 * bug fix: type-checks for function arguments were compiled using the
344 compiler policy settings of the wrong lexical environment (in 1.0.2
346 * bug fix: SHADOW accepts characters as string designators, as required
347 by the spec (thanks to Eric Marsden)
348 * bug fix: fixed GC safety issues when foreign functions are called with
349 non-base strings as arguments
350 * bug fix: more consistent error output for fatal-compiler-warnings, like
352 * bug fix: functions with proper names were showing up as NIL in
353 backtraces (reported by Edi Weitz, regression in 1.0.2)
355 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
356 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
357 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
358 produce more efficient code on the x86 if the compiler can determine
359 that (+ INDEX <constant>) does not require a bounds check and FOO
360 has an element type at least 8 bits wide.
361 * bug fix: references to undefined variables are handled the same way
362 in toplevel forms as in the normal compiler
363 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
365 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
366 (thanks to Jon Buller)
367 * improvement: faster compilation times for complex functions
368 * improvement: added readlink support to SB-POSIX (thanks to Richard
371 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
372 * improvement: experimental support for mach exception handling on
373 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
375 * improvement: support for GBK external format.
376 (thanks to Chun Tian (binghe))
377 * improvement: the debugger now displays variables that have been closed
378 over, in code compiled with (DEBUG 2) or higher
379 * improvement: support for executable cores on NetBSD (thanks to
381 * new feature: added a RESTART-FRAME debugger command
382 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
383 be used to provide an implementation for FILE-POSITION on Gray streams
384 (thanks to Eric Marsden)
385 * new feature: add syslog, openlog and closelog support to SB-POSIX
386 (thanks to Richard Kreuter)
387 * optimization: the function call overhead in code compiled with
388 a high DEBUG optimization setting is significantly
389 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
390 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
391 (thanks to Tony Martinez)
392 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
393 evaluated once (reported by Kevin Reid)
394 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
395 works again (thanks to Kilian Sprotte)
396 * bug fix: an error is signaled for tagbodies with duplicate tags
397 (thanks to Stephen Wilson)
398 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
399 * bug fix: Win32 port can now handle foreign code unwinding Lisp
400 stack frames from alien callbacks.
401 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
402 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
403 65k pages (thanks to David Woodhouse)
404 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
406 changes in sbcl-1.0.1 relative to sbcl-1.0:
407 * new platform: FreeBSD/x86-64, including support for threading.
408 * new feature: the compiler stores cross-referencing information
409 abount function calls (who-calls), macroexpansion (who-macroexpands)
410 and special variables (who-binds, who-sets, who-references) for code
411 compiled with (< SPACE 3). This information is available through the
412 sb-introspect contrib.
413 * new feature: users may subclass SEQUENCE, and have instances of
414 these classes interoperate with standard Common Lisp functions if
415 a number of methods are defined. (This feature is experimental
416 and the interface subject to change based on feedback from SBCL
417 users and the general community)
418 * improvement: sb-sprof traces call stacks to an arbitrary depth on
419 x86/x86-64, rather than the previous fixed depth of 8
420 * improvement: another pthread back-end of mutex "pthread-futex".
421 * bug fix: non-ascii command-line arguments are processed correctly
422 (thanks to Yaroslav Kavenchuk)
423 * bug fix: non-required arguments were not passed correctly when a method
424 defined using DEFMETHOD was called from a mop-generated method using
425 CALL-NEXT-METHOD (reported by Pascal Costanza)
426 * bug fix: recursion is now permitted in accessors through
427 SLOT-UNBOUND. (reported by Pascal Costanza)
428 * bug fix: an error was signaled at startup if the HOME environment
429 variable was defined, but had an empty value (reported by Peter Van Eynde)
430 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
431 signal a PROGRAM-ERROR, not a BUG.
432 * bug fix: SB-POSIX:READDIR works when built with large file support.
433 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
435 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
436 for better automatic indentation support. (thanks to Matt Pillsbury)
437 * optimization: loading generic functions no longer takes O(n^2) time,
438 proportional to the amount of methods in the generic function
439 (reported by Todd Sabin and Jeremy Brown)
440 * optimization: the FIND and POSITION family of sequence functions
441 are significantly faster on arrays whose element types have been
443 * improvements to the Windows port:
444 ** Intermittent heap corruption problems have been fixed. (thanks
445 to Alastair Bridgewater)
446 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
448 ** Lisp is able to unwind foreign exception frames from alien
449 callbacks. (thanks to Alastair Bridgewater)
451 changes in sbcl-1.0 relative to sbcl-0.9.18:
452 * improvement: experimental support for threading on FreeBSD/x86.
453 (thanks to NIIMI Satoshi)
454 * improvement: runtime option --dynamic-space-size can be used
455 to set the size of the dynamic space reserved on startup.
456 * improvement: floating point modes in effect are now saved in
457 core, and restored on startup.
458 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
459 startup, not time since first call to GET-INTERNAL-REAL-TIME.
460 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
461 threads are not running after *SAVE-HOOKS* have run.
462 * improvement: writes to CLOS instance slots are type-checked in code
463 compiled with (SAFETY 3)
464 * improvement: floating-point exception handling on FreeBSD (thanks to
466 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
467 (thanks to Zach Beane)
468 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
470 * improvement: added support for the Shift-JIS external format.
471 (contributed by NIIMI Satoshi)
472 * improvement: callbacks are supported on Linux/PPC. (thanks to
474 * bug fix: compiler bug triggered by a (non-standard) VALUES
475 declaration in a LET* was fixed. (reported by Kaersten Poeck)
476 * bug fix: file compiler no longer confuses validated and already
477 dumped structurres. (reported by Kaersten Poeck)
478 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
479 fill-pointer signals a type-error as required. (thanks to
481 * bug fix: disassemly of funcallable instances works.
482 * bug fix: single stepping on PPC.
483 * bug fix: fix thread-safety problems in the type system (generally
484 manifesting as nonsensical errors like "STRING is a bad type specifier
485 for sequences" or "The value 1 is not of type FIXNUM")
486 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
487 (reported by Josip Gracin)
488 * bug fix: an error is signaled for attempts to displace arrays with
489 incompatible element types (thanks to Mario Mommer)
490 * bug fix: more correct handling of wide characters in debug info
491 (bug reported by Attila Lendvai and fixed by Juho Snellman)
492 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
493 and don't cause extra consing
494 * optimization: MAP and MAP-INTO are significantly faster on vectors
495 whose elements types have been declared.
496 * Improvements to SB-SPROF:
497 ** Support for allocation profiling
498 ** Reduced profiling overhead, especially for long profiling runs
499 * Improvements to the Windows port:
500 ** floating point exceptions are now reported correctly.
501 ** stack exhaustion detection works partially.
502 ** more accurate GET-INTERNAL-REAL-TIME.
503 ** better RUN-PROGRAM behaviour in case of errors in the child
505 ** PROBE-FILE now simplifies pathnames correctly.
506 ** DIRECTORY now works correctly with :WILD-INFERIORS.
507 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
508 1st 1970 (although time timezone data will be unreliable).
509 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
511 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
512 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
513 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
515 * bug fix: two potential GC deadlocks affecting threaded builds.
516 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
517 returns T (reported by Anton Kazennikov)
518 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
519 with non-variable places
520 * bug fix: remove a race condition in the setting of
521 funcallable-instance functions, this should make threaded CLOS
522 code more stable against memory faults.
523 * bug fix: corruption of specials when unbinding is interrupted by an
524 asynchronous unwind (reported by Hannu Koivisto)
525 * improvement: the debugger will now also display local variables that
526 are only used once, for code compiled with a DEBUG optimization quality
529 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
530 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
531 * incompatible change: External-format support for FFI calls. The
532 SB-ALIEN:C-STRING no longer implies an ASCII
533 external-format. Instead the string is subject to external-format
534 conversion. Additionally return values of type SB-ALIEN:C-STRING
535 are by default of type (SIMPLE-ARRAY CHARACTER), not
536 SIMPLE-BASE-STRING. For an alien type that behaves like the old
537 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
538 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
539 * incompatible change: SB-EXT package no longer contains the
540 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
541 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
543 * incompatible change: the single-stepper is no longer available
544 on Alpha, Mips and Sparc platforms.
545 * minor incompatible change: the direct superclasses of
546 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
547 not (STANDARD-OBJECT FUNCTION). This makes the
548 class-precedence-lists of GENERIC-FUNCTION and
549 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
551 * minor incompatible change: the default stream external format on
552 non-unicode SBCL is no longer detected from the locale, but is always
554 * new feature: Add a version of evaluator that uses an interpreter instead
555 of the compiler. EVAL still uses the compiler by default, to switch it
556 to use the interpreter, set the value of the variable
557 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
558 * minor incompatible change: the single-stepper REPL has been merged
559 with the normal debugger (see the "Stepping" heading of the debugger help
560 for more details). The debugger command STEP will no longer switch
561 to the single-stepper REPL.
562 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
563 for a type now works.
564 * bug fix: loading of default sysinit file works. (thanks to Leonid
566 * bug fix: better detection of circularities in the file-compiler.
567 (reported by Marco Monteiro)
568 * bug fix: the CL pathname functions now work with files that have
569 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
570 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
571 systems with Unix98 pty semantics.
572 * bug fix: ASDF-INSTALL will now work with bsd tar.
573 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
575 * bug fix: timers expiring in dead threads no longer cause a
576 type-error (reported by Paul "Nonny Mouse").
577 * bug fix: thanks to more lightweight single-stepper instrumentation,
578 code compiled with (DEBUG 3) will compile and execute significantly faster,
579 and will have more accurate type-inferencing than before
580 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
581 whose bindings are modified
582 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
583 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
584 ** replace dummy implementations of CL:MACHINE-INSTANCE and
585 CL:SOFTWARE-VERSION with proper version
587 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
588 * feature: implemented the READER-METHOD-CLASS and
589 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
590 as specified by AMOP.
591 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
593 * optimization: faster LOGCOUNT implementation on x86 and x86-64
594 (thanks to Lutz Euler)
595 * optimization: hashing of general arrays and vectors has been
596 improved. (reported by Any Fingerhut)
597 * enhancement: SB-INTROSPECT is now able to find definitions of
598 profiled functions. (thanks to Troels Henriksen)
599 * enhancement: compiler-macro expansion applies now to FUNCALL forms
600 as well, allowing compiler-macros for SETF-functions to expand.
601 * enhancement: step-instrumentation no longer wraps known
602 single-value functions with multiple-value context, allowing
603 better type inference.
604 * fixed bug #337: use of MAKE-METHOD in method combination now works
605 even in the presence of user-defined method classes. (reported by
606 Bruno Haible and Pascal Costanza)
607 * fixed bug #339(c): if there are applicable methods not part of any
608 long-form method-combination group, call INVALID-METHOD-ERROR.
609 (reported by Bruno Haible)
610 * fixed bug #361: the :FUNCTION initarg in the protocol for
611 initialization of methods can now be used to override
612 internally-produced optimized functions. (reported by Bruno
614 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
615 system-provided lambda expression no longer cause warnings about
616 unbound #:|pv-table| symbols.
617 * bug fix: improved the handling of type declarations and the
618 detection of violations for keyword arguments with non-constant
620 * bug fix: potentially erronous calls to PATHNAME and
621 MERGE-PATHNAMES were being flushed in some cases.
622 (reported by Richard Kreuter)
623 * bug fix: compiled calls to TYPEP were mishandling obsolete
624 instances. (reported by James Bielman and Attila Lendvai)
625 * bug fix: format strings with ~> without matching ~< no longer
626 trigger an AVER, but signal an understandable error instead.
627 (reported by Antonio Martinez)
628 * bug fix: specifying an output-file without a pathname-type for
629 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
630 of input-file instead of "fasl". (reported by Robert Dodier)
631 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
632 (reported by James Y Knight).
633 * bug fix: compiler-macros-function did not consider the environment
634 argument for shadowing by local functions.
635 * bug fix: compiler-macros expansion was inhibited by local INLINE
637 * bug fix: inline expansions of known functions were subject to
638 step-instrumentation in high debug policies, leading to problems
640 * bug fix: compiler failed to differentiate between different CONS
642 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
643 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
644 * bug fix: #368: incorrect use of expressed vs. upgraded array
646 * bug fix: #306a: more precise unions of array types.
647 * thread-safety improvements:
648 ** CONDITION-WAIT could return early on Linux, if the thread was
649 interrupted and subsequently continued with SIGCONT.
650 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
652 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
653 * added support for the ucs-2 external format. (contributed by Ivan
655 * minor incompatible change: pretty printing of objects of type
656 (cons symbol) is, in the default pprint-dispatch-table, now
657 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
659 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
660 called later in a class's lifetime, possibly as late as when the
661 first instance of the class is created. Previously,
662 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
663 class became finalizeable.
664 * fixed bug: FILE-POSITION sometimes returned inconsistent results
665 for multibyte external-format streams. (thanks to "vbzoli")
666 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
667 with :ALLOCATION :CLASS inherited from superclasses of the
669 * fixed bug: anonymous classes can now be created using the :NAME
670 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
671 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
672 * fixed bug: core-files saved with :EXECUTABLE T can again be
673 executed when SBCL_HOME isn't set. (reported by James Knight)
674 * fixed bug: toplevel LOCALLY forms with declarations could
675 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
676 * fixed bug: printing from several different threads using different
677 values of *print-case* could cause invalid output, due to
678 some internal special variables of the printer not being bound
679 thread-locally (reported by Max Mikhanosha)
680 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
681 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
682 (reported by Pascal Costanza)
683 * fixed bug: FINALIZE-INHERITANCE is called from
684 REINITIALIZE-INSTANCE on classes when the class has previously
685 been finalized, as required by AMOP.
686 * minor code generation optimizations:
687 ** better register allocation in CLOS dispatching functions
688 ** overflow detection when coercing signed bytes to fixnums on x86-64
689 ** is now implemented with one IMUL instruction instead of three shifts
690 ** more efficient bit-vector access on x86 and x86-64
691 ** more efficient access to raw structure slots on x86 and x86-64
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
696 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
697 * feature: thread support on Solaris/x86, and experimental thread support
699 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
701 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
702 default method for SLOT-UNBOUND.
703 * minor incompatible change: prevent the user from specializing the
704 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
705 somewhat counter to the intent of the protocol, I (CSR) think, and
706 additionally it just doesn't work in SBCL as currently
707 implemented, thanks to optimizations (that are always valid for
708 the other three SLOT-VALUEish functions, but not for the setter).
709 * minor incompatible change: the :SB-LDB feature is now enabled by
710 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
711 the low-level debugger.
712 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
713 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
714 down to an SB-SYS:FD-STREAM, allowing direct communication in
715 more cases. Temporary files and pipes are still used as a fallback
717 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
718 on the non-reentrant C functions localtime(3) and gmtime(3).
719 * bug fix: global optimization policy was not visible in LOCALLY and
721 * bug fix: class objects can be used as specializers in methods.
722 (reported by Pascal Costanza)
723 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
724 with a type error. (reported by blitz_ on #lisp)
725 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
726 failed with a type error. (reported by Pascal Bourguignon)
727 * bug fix: merging pathnames against defaults with :DIRECTORY
728 starting with '(:RELATIVE :BACK) should preserve the :BACK.
729 (reported by James Y Knight)
730 * bug fix: saving large (>2GB) cores on x86-64 now works
731 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
732 constant index and a (simple-array (signed-byte 32)) array
733 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
734 error (patch by Robert J. Macomber)
735 * feature: TIME now displays the amount of run-time spent in GC
736 * bug fix: The debugger now does a better job of respecting
737 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
738 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
739 workaround for bug 403.)
740 * bug fix: floating point exception handling now works on FreeBSD
741 (thanks to NIIMI Satoshi)
742 * fixed some bugs revealed by Paul Dietz' test suite:
743 ** MISC.641: LET-conversion were not supposed to work in late
745 * improvements to the Win32/x86 port:
746 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
747 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
749 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
750 * new feature: source path information is generated for macro-expansion
751 errors for use in IDE's like Slime (thanks to Helmut Eller)
752 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
753 * bug fix: compiler does not loop forever on an invalid type in
755 * improvement: compilation of most CLOS applications is significantly
757 * optimization: added a limited bytecode compiler for simple toplevel
758 forms, speeding up compilation and FASL loading
759 * bug fix: the statistical profiler now properly distinguishes anonymous
762 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
763 * minor incompatible change: in sbcl-0.9.11 (but not earlier
764 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
765 into subclasses of the finalized class. Now user calls to
766 FINALIZE-INHERITANCE finalize just the one class, and calls by the
767 system return before any subclasses are finalized.
768 * minor incompatible change: The reader no longer ignores errors
769 regarding non-existent packages in #+ and #- feature tests.
770 * new feature: command line options --no-sysinit, --no-userinit to
771 inhibit loading the corresponding init files
772 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
773 for sending data through UDP sockets (thanks to François-René Rideau)
774 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
775 error is signalled instead (thanks to François-René Rideau)
776 * minor incompatible change: Error signalling behaviour of lexical
777 operations violating package locks has changed slightly. Refer to
778 documentation on package locks for details.
779 * enhancement: EVAL can process IF-expressions without resorting to the
781 * bug fix: Some operations involving SIN, COS, or TAN weren't being
782 constant-folded properly due to missing out-of-line functions.
783 (reported by Mika Pihlajamäki)
784 * bug fix: LISTEN sometimes returned T even in cases where no data was
785 immediately available from the stream
786 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
787 were reversed. (reported by Levente Mészáros)
788 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
789 when given a vector displaced to an adjustable array. (reported
791 * bug fix: derivation of float boundaries from numbers outside the
792 appropriate float range (reported by John Wiseman)
793 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
795 * bug fix: precomputing cache entries for generic functions with
796 some subclasses of specializers as yet invalid does not attempt to
797 fill a cache line with a negative offset. (reported by Levente
799 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
800 allow condition class objects as arguments to DOCUMENTATION and
801 (SETF DOCUMENTATION); only find and set documentation for
802 structure names for the STRUCTURE doc-type. (suggested by Gary
804 * improvements to the Win32/x86 port:
805 ** (user-homedir-pathname) and default initialization file
806 locations now know about the user's "Documents and Settings"
807 directory (thanks to Yaroslav Kavenchuk)
808 ** run-program is implemented (thanks to Mike Thomas)
809 ** sockets support (thanks to Timothy Ritchey)
810 ** better backtrace support (thanks to Alastair Bridgewater)
811 ** sb-grovel supported
812 ** asdf-install and sb-posix work somewhat
813 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
814 * improvements to the Solaris/x86 port:
815 ** works on Solaris 11/Solaris Express
816 ** floating-point exception handling support
817 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
818 * fixed some bugs revealed by Paul Dietz' test suite:
819 ** REMOVE-METHOD returns its generic function argument even when
820 the method is not one of the generic functions' methods.
821 ** objects declared as MEMBER types can be admissible arguments to
823 ** printing characters should simply be printed by the FORMAT ~:C
825 ** compiler failure when compiling functions with hairy constant
826 defaults for optional parameters.
827 ** compiler produces wrong code when MAYBE-INLINE-expanding a
828 function, which is already optimized.
830 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
831 * new platform: experimental support for SBCL x86/Darwin, including
832 MacOS X 10.4.5 on Intel.
833 * new feature: Unicode character names are now known to the system
834 (through CHAR-NAME and NAME-CHAR).
835 * new feature: the filesystem location of SBCL's core file is
836 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
837 * minor incompatible change: the contrib modules SB-POSIX and
838 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
839 this change is to make it easier to distribute
840 location-independent binaries.
841 * bug fix: as implied by AMOP, standardized classes no longer have
842 slots named by external symbols of public packages. (reported by
844 * optimization: calling functions via constant symbols -- as in
845 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
846 via the function object as in (FUNCALL #'FOO).
847 * optimization: on x86 and x86-64, the calling convention for the
848 case of dealing with an unknown number of values has been altered
849 to be friendlier to the prediction heuristics implemented,
850 particularly with reference to CALL and RET pairing. (thanks to
851 Alastair Bridgewater)
852 * optimization: on x86, the code for access to thread-local symbol
853 values and binding/unbinding of thread-local symbols is smaller.
854 (thanks to Alastair Bridgewater)
855 * enhancement: CONSTANTP is now able to determine constantness of
856 more complex forms, including calls to constant-foldable standardized
857 functions and some special forms beyond QUOTE.
858 * fixed bug: occasional GC crashes on Solaris/x86
859 * optimization: x86-64 supports stack allocation of results of simple
860 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
861 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
862 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
863 (thanks to James Knight)
864 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
865 by the standard (thanks to Stephen Compall)
867 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
868 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
869 be used for bundling the runtime and the core file into one
870 executable binary. This feature is not currently supported on all SBCL
871 platforms. (thanks to James Bielman and NIIMI Satoshi)
872 * new feature: a generational or ephemeral garbage collector is now
873 the default on the PowerPC platform (both Linux and Darwin). The
874 old Cheney (stop and copy) collector is a build-time option.
875 (thanks to Cyrus Harmon, after Raymond Toy)
876 * minor incompatible change: the method by which SBCL finds its
877 contributed modules has changed; it no longer relies on symbolic
878 links from an $SBCL_HOME/systems directory, but searches directly
879 in the subdirectories of $SBCL_HOME.
880 * enhancement: the dynamic heap size on the Linux/PPC platform is
881 markedly larger, even using the older Cheney garbage collector.
882 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
883 certain complicated string types. (reported by Gary King)
884 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
885 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
886 * fixed bug: DOCUMENTATION on structures no longer signals an
887 error if no structure type of the right name exists. (reported by
889 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
890 applicable non-standard primary or :AROUND method on
891 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
892 no longer cause unbound variable errors. (reported by Kevin Reid)
893 * fixed bug: metacircle resolution in cases where methods have slots
894 added before the slots from STANDARD-METHOD. (reported by Jean
896 * fixed bug: the Power PC assembly code for calling into lisp
897 sometimes computed its return address wrongly (depending on the
898 15th bit of the address, fixed at link time). This used to
899 manifest itself in a segmentation violation while building PCL.
900 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
901 many others over the years)
902 * fixed bug: ROOM no longer reports silly numbers for stack usage if
903 the stack pointer (treated as a signed integer) is negative.
904 (thanks to Peter van Eynde)
906 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
907 * new platform: experimental support for the Windows operating
908 system has been added. (thanks to Alastair Bridgewater)
909 * fixed several bugs in and robustified the PPC FFI (including
910 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
911 * bug fix: allow non-simple string symbol names (reported by Paul
913 * bug fix: interrupt handling on NetBSD (thanks to Richard M
915 * bug fix: saving a core corrupted callbacks on x86/x86-64
916 * bug fix: closed a loophole in metacircularity detection and
917 grounding in the PCL implementation of CLOS.
918 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
919 (reported by tomppa on #lisp)
920 * bug fix: clear the direction flag on Lisp -> C transitions, as
921 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
922 (reported by Andrej Grozin and Hendrik Maryns)
923 * optimization: major improvements to GC efficiency on GENCGC platforms
924 * optimization: faster implementation of EQUAL
925 * optimization: emit more efficient opcodes for some common
926 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
927 * optimization: several other minor code-generation improvements on x86-64
928 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
930 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
931 * minor incompatible change: (SETF CLASS-NAME) and (SETF
932 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
933 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
934 by ANSI as a generic function, and for consistency (SETF
935 GENERIC-FUNCTION-NAME) is treated likewise.
936 * fixed bug #233.b: make constraint propagation notice when a variable
937 value is changed after it is referenced but before it is used
938 * fixed bug #296: no more arbitrary behaviour from filenames with
939 odd characters as --load arguments. (reported by Adam Warner)
940 * fixed bug #390: :CHARACTER-SET pathname components now work as
941 expected. (reported by Tim Daly Jr)
942 * fixed bug #391: complicated :TYPE intersections in slot
943 definitions no longer cause an error in PCL internals.
944 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
945 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
946 returning the number of octets which would be written to the
947 file-stream. (thanks to Robert J. Macomber)
948 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
949 arguments to the internal stream functions. (thanks to David
951 * bug fix: stack exhaustion is now handled on the x86/SunOS
952 platform. (thanks to Daisuke Homma)
953 * bug fix: more accurate ROOM results on GENCGC platforms
954 * optimization: improved type inference for arithmetic-for
955 index variables in LOOP
956 * optimization: faster floating-point SQRT on x86-64
957 * fixed some bugs revealed by Paul Dietz' test suite:
958 ** DOCUMENTATION returns NIL instead of "" for method combinations
959 that don't have a docstring
961 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
962 * minor incompatible change: (SETF CLASS-NAME) and (SETF
963 GENERIC-FUNCTION-NAME) are no longer generic functions, and
964 therefore are not customizeable by user code (as seems to be at
965 least permitted and maybe required by AMOP). As a consolation,
966 however, the SBCL implementation of these functions now calls
967 REINITIALIZE-INSTANCE as specified by AMOP.
968 * bug fix: slot-definition documentation is propagated as per ANSI
969 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
970 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
971 argument, as do eslotd initialization methods. (from Pascal
972 Costanza's "Closer" project)
973 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
974 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
976 * bug fix: it is now possible to have more than one subclass of
977 STANDARD-GENERIC-FUNCTION without causing stack overflow.
978 (reported by Bruno Haible, Pascal Costanza and others)
979 * bug fix: the dependent update protocol now works for generic
980 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
982 * bug fix: condition-class instances corresponding to
983 DEFINE-CONDITION forms are now created eagerly. (reported by
984 Kalle Olavi Niemitalo on comp.lang.lisp)
985 * bug fix: floating point printing is more accurate in some
986 circumstances. (thanks to Simon Alexander)
987 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
988 merged with *DEFAULT-PATHNAME-DEFAULTS*.
989 * bug fix: callbacks on OS X now preserve stack-alignment.
990 (thanks to Cyrus Harmon)
991 * enhancement: the x86-64 disassembler is much better at
992 disassembling SSE instructions. (thanks to Lutz Euler)
993 * enhancement: CHECK-TYPE on a variable now allows type-inference
994 (thanks to Vincent Arkesteijn)
995 * optimization: improved performance of EUC-JP external format.
996 (thanks to NIIMI Satoshi)
997 * optimization: performance improvements to IO on file streams of
998 :ELEMENT-TYPE CHARACTER
999 * optimization: much faster memory allocation on x86-64
1000 * optimization: faster garbage collections (latency and throughput)
1002 * optimization: faster bignum subtraction and division on x86 and x86-64
1003 (thanks to Lutz Euler)
1004 * optimization: more accurate type-inference for loops that use a
1005 floating point index variable or a negative step.
1007 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1008 * bug fix: add a workaround to SBCL looping infinitely at startup on
1009 Linux kernels with apparently buggy implementations of personality().
1010 (thanks to Svein Ove Aas)
1011 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1012 (thanks to David Lichteblau)
1013 * bug fix: local bindings shadow global symbol macros for
1014 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1015 * optimization: non-open coded uses of numeric comparison operators
1016 (e.g. >) no longer cons when called with more than one parameter
1017 on platforms supporting dynamic-extent allocation.
1018 * enhancement: saving cores with foreign code loaded is now supported
1019 on MIPS/Linux in addition to the previously supported platforms.
1020 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1021 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1022 * bug fix: division by zero in sb-sprof when no samples were collected
1023 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1024 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1025 (thanks to Alexander Kjeldaas)
1026 * new feature: ignoring the return values of destructive functions
1027 like NREVERSE signals a compile-time style-warning.
1028 (thanks to Kevin Reid)
1029 * new feature: the alignment of alien structure fields can be
1030 explicitly specified. (thanks to Cyrus Harmon)
1031 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1032 and dump core on SIGQUIT
1034 ** incompatible change: threads do not inherit values of specials
1035 from their parents (see manual)
1036 ** bug fix: threads stacks belonging to dead threads are freed by the
1037 next exiting thread, no need to gc to collect thread stacks anymore
1038 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1039 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1040 ** bug fix: dynamic variable and thread start related gc lossage
1041 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1043 * fixed some bugs revealed by Paul Dietz' test suite:
1044 ** SUBTYPEP is slightly more accurate on heinously complicated
1045 CONS types where some of the members have uncertain (in the
1046 NIL, NIL sense) type relationships to each other.
1048 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1049 no longer purified unless :PURIFY T is explicitly specified.
1050 ** Non-purified cores are significantly smaller than before
1052 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1053 * new feature: timers based on Zach Beane's excellent timer package
1054 * added support for the following external formats: koi8-u,
1055 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1056 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1057 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1058 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1059 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1060 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1061 * incompatible change: a threaded SBCL will no longer revert to
1062 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1063 * bug fix: interrupts are disabled until startup is complete; no
1064 more sigsegvs when receiving a signal to soon
1065 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1067 * bug fix: add a workaround for the memory randomization features in
1068 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1069 workaround will only be in effect on systems with the proc filesystem
1071 * bug fix: printing objects of type HASH-TABLE signals a
1072 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1074 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1075 have been running for over 50 days. (reported by Gilbert Baumann)
1076 * bug fix: the logic for getting names of functions gets less
1077 confused when confronded with alternate-metaclass
1078 funcallable-instances. (reported by Cyrus Harmon)
1079 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1080 consistent, even on internal alternate-metaclass objects.
1081 * bug fix: flush closure information collected by physical
1082 environment analysis prepass before the main pass. (bug reported
1084 * bug fix: compiler pack phase does not modify a hashtable, which is
1085 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1086 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1087 may be expanded. (reported by Alexander <malishev> on c.l.l)
1088 * bug fix: no more style-warnings when compiling code using LAST
1089 under high-SPEED low-SPACE settings. (reported by David Wallin
1091 * fixed some incompatibilities between SBCL's MOP and the MOP
1093 ** the METAOBJECT class is now implemented;
1094 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1095 STANDARD-OBJECT, as required;
1096 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1097 now compatible (as required by VALIDATE-SUPERCLASS); there
1098 remains a consistency requirement over the presence or absence
1099 of the FUNCTION class in the superclass list of finalized
1100 classes; see the manual for more details;
1101 ** the :AROUND method for COMPUTE-SLOTS on
1102 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1103 requested slot ordering.
1105 ** bug fix: parent thread now can be gc'ed even with a live
1107 ** bug fix: binding a special with PROGV to no value is not
1108 the same as that symbol not having been bound (thanks to
1110 ** bug fix: binding specials is thread safe (thanks to
1112 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1113 and SB-SPROF do not die with 'no handler for signal XX in
1114 interrupt_handle_now(..)' anymore
1115 ** bug fix: WITH-TIMEOUT works with multiple running threads
1116 * fixed some bugs revealed by Paul Dietz' test suite:
1117 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1118 the :method-class keyword argument.
1120 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1121 * new port: the Solaris operating system on x86 processors is now
1122 mostly supported, though some rough edges in the environment
1123 remain. (thanks to Daisuke Homma)
1124 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1125 space for its heap. (thanks to Thiemo Seufer)
1126 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1127 closures. (thanks to Thiemo Seufer)
1128 * minor incompatible change: eof selects abort in the debugger.
1129 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1130 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1132 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1133 deprecated, and will go away in a future revision of SBCL.
1134 * minor incompatible change: GC-ON and GC-OFF are no longer
1135 implemented with a counter, it does not matter how many times gc
1136 is switched on or off
1137 * bug fix: discriminating functions for generic function classes
1138 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1139 make invalid assumptions about method precedence order. (reported
1141 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1142 platforms. (thanks to Thiemo Seufer)
1143 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1144 * bug fix: correct pathname printing with printer escaping is on.
1145 (thanks to Kevin Reid)
1146 * bug fix: complex VOP definitions in "user-space" no longer trigger
1147 package locks. (reported by Zach Beane)
1148 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1149 for funcallable-instance functions, and is no different from
1151 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1152 characters in the current readtable. (reported by Nicholas Neuss)
1153 * bug fix: SUBTYPEP on various CONS types returns more a more
1154 accurate acknowledgment of its certainty.
1155 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1156 lists in some cases. This partially fixes bug 384.
1157 * flush all standard streams before prompting in the REPL and the
1159 * bug fix: signal handling and triggering gc do not conflict
1160 directly anymore, in particular a high frequency sb-sprof does
1161 not prevent gc from running
1162 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1163 approximation for timezone and DST information between the
1164 universal time epoch and the smallest negative 32-bit time_t.
1165 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1166 year 1899 so long as the encoded time is non-negative.
1167 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1168 the correct answer for non-constant rotate amounts when performing
1169 an inline 32-bit rotation.
1171 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1172 there is only one thread in the session
1173 ** bug fix: memory leak for streams created in one thread and
1174 written to in another
1175 ** bug fix: lockup when compiled with gcc4
1176 ** bug fix: race that allows the gc to be triggered when gc is
1178 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1179 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1181 ** bug fix: finished threads are gc'ed properly
1182 * fixed some bugs revealed by Paul Dietz' test suite:
1183 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1184 has arguments with a different set of applicable methods from
1185 the orignal arguments.
1186 ** The type error thrown by MAP now has a correct expected-type
1188 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1189 name a compiled function.
1190 ** Three MISC tests where a large bignum was improperly coerced to
1191 a float that couldn't represent that bignum during type
1192 derivation were fixed.
1193 ** SUBTYPEP can now handle types involving the negation of a
1194 list-form FUNCTION type.
1195 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1196 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1197 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1199 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1200 * New feature: Experimental support for bivalent streams: streams
1201 opened with :element-type :default now allow character and binary
1202 (unsigned-byte 8) I/O
1203 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1204 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1206 * bug fix: no more highly sporadic "couldn't check whether ~S is
1207 readable" when reading a stream and an interrupt hits in the middle
1208 of a select system call
1209 * compiler better recognizes complex arrays (reported by Eduardo
1211 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1212 for variables on SBCL built with linkage-tables. (reported by Luis
1214 * various error reporting improvements.
1215 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1216 (Thanks to Thiemo Seufer)
1217 * bug fix: nested reader invokations work correctly
1218 * bug fix: it is possible to have simultaneous references to foreign
1219 code and foreign data with the same name.
1221 ** added x86-64 support
1222 ** incompatible change: the threading api now works with thread
1223 objects instead of thread ids
1224 ** bug fix: threads are protected from signals and interruption when
1225 starting up or going down
1226 ** bug fix: a race where an exiting thread could lose its stack to gc
1227 ** bug fix: don't halt on infinite error in threads if possible
1228 ** fixed numerous gc deadlocks introduced in the pthread merge
1229 ** bug fix: fixed thread safety issues in read and print
1230 ** bug fix: debugger doesn't hang on session lock if interrupted at
1231 an inappropriate moment
1232 ** bug fix: run-program is now thread safe(r)
1233 ** bug fix: inner with-recursive-lock no longer releases the mutex
1234 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1235 * fixed some bugs revealed by Paul Dietz' test suite:
1236 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1237 EXPECTED-TYPE slots filled.
1238 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1240 ** MAKE-CONDITION accepts classes as type-designators.
1241 ** COMPILE may never return NIL.
1242 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1243 range before calling Unix time functions
1245 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1246 * numerous signal handling fixes to increase stability
1247 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1248 * minor incompatible change: we now correctly canonize default
1249 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1250 as per the MOP, rather than the historical (INITARG INITFUNCTION
1251 INITFORM). (reported by Bruno Haible)
1252 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1253 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1254 for more information.
1255 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1256 pathname is a directory pathname.
1257 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1258 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1260 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1261 :METACLASS argument in addition to a class name. (reported by
1262 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1263 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1264 (reported by Bruno Haible; patch from Gerd Moellmann)
1265 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1267 * bug fix: more cleanups to the floating point exception handling on
1268 x86-64 (thanks to James Knight)
1269 * bug fix: the compiler does not try to constant fold calls of
1270 COERCE and BIT-* functions when they return freshly constructed
1271 objects. (reported by Nikodemus Siivola and Paul Dietz)
1272 * optimization: improved the MIPS versions of generic (in the
1273 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1274 * optimization: direct conversion of (unsigned-byte 32) to floats on
1275 the PowerPC platform.
1276 * optimization: structure instances with raw slots now use less
1277 memory, and probably show better memory locality. (thanks to
1279 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1280 are treated as requests for stack allocation on the x86-64,
1281 Alpha-32, PPC and SPARC platforms.
1282 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1283 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1284 (thanks to Tony Martinez)
1286 ** gcing a dead thread can no longer lead to lockups
1287 ** threads block signals until they are set up properly
1288 ** errno is no longer shared by threads
1289 ** interrupt-thread restores the eflags register on x86
1290 ** fixed some lockups due to gc/thread interaction
1291 * fixed some bugs revealed by Paul Dietz' test suite:
1292 ** invalid dotted lists no longer raise a read error when
1293 *READ-SUPPRESS* is T
1294 ** signal an error if a symbol that names a declaration is used
1295 as the name of a type, or vice versa
1296 ** allow using the (declare (typespec var*)) abbreviation for
1297 (declare (type typespec var*)) with all type specifiers
1298 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1299 ** always return NIL from PPRINT-POP when OBJECT is NIL
1300 ** don't signal errors when pretty-printing malformed LABELS,
1301 FLET or MACROLET forms
1302 ** declarations in a DOLIST body are also in scope for the
1304 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1306 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1309 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1310 * fixed cross-compiler leakages that prevented building a 32-bit
1311 target with a 64-bit host compiler.
1312 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1313 opened with :IF-EXISTS :OVERWRITE.
1314 * fixed bug 281: error for an invalid qualifier in a short-form method
1315 combination method is not signalled until the faulty method is called.
1316 * bug fix: iteration variable type inferrer failed to deal with open
1317 intervals. (reported by Alan Shields)
1318 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1319 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1320 or more runtime options were provided to the sbcl binary.
1321 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1322 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1323 * optimization: there's now a fast-path for fixnum arguments in the
1324 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1325 * optimization: the code generated on x86-64 is more compact thanks
1326 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1327 * bug fix: floating-point exception handling now partly works on
1328 x86-64. (Thanks to James Knight)
1329 * improvement to the MIPS backend from Thiemo Seufer: C-style
1330 64-bit long long arguments and return values to alien functions
1332 * fixed some bugs revealed by Paul Dietz' test suite:
1333 ** the type-error signalled from WARN has a filled-in DATUM slot.
1334 ** the type-error required when a stream is not associated with
1335 a file has the stream as its datum.
1336 ** type-errors on single-floats on x86-64 no longer have
1337 :INVALID-OBJECT as the datum
1338 ** the type-errors signalled for invalid function names now have
1339 a correct expected type
1340 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1341 for structures defined with a :TYPE. Documentation strings for
1342 typed structures are no longer immediately discarded
1343 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1344 broadcast streams and synonym streams. FILE-LENGTH now also works
1345 on broadcast streams.
1347 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1348 * incompatible change: the --noprogrammer option, deprecated since
1349 version 0.7.5, has been removed. Please use the equivalent
1350 --disable-debugger option instead.
1351 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1353 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1354 inoperational for a while now) has been completely removed.
1355 * null lexical environments are now printed as #<NULL-LEXENV>,
1356 significantly reducing the amount of clutter in typical backtraces.
1357 * documentation on weak pointers, finalization, and after GC hooks
1358 has been added to the manual.
1359 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1360 as other specialized array types, is much faster. SUBSEQ and
1361 COPY-SEQ on such arrays have also been sped up.
1362 * optimization: EQL is now more efficient when at least other argument
1363 is known to be of type (OR FIXNUM (NOT NUMBER)).
1364 * fixed bug: the runtime is now less vulnerable to changes in the
1365 size of the SBCL object on OS X, and virtual memory is reserved for
1366 all spaces that need to be at a fixed address.
1367 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1368 * fixed bug: finalizers and after GC hooks that cause consing are now
1370 * fixed bug: compiler error messages and summaries are now printed to
1371 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1372 * fixed inference of the upper bound of an iteration variable.
1373 (reported by Rajat Datta).
1374 * fixed bug 211e: calling local functions with duplicated constant
1375 keyword argument no longer causes a bogus style warning about an
1377 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1378 ftype declaration to be disregarded. (reported by Dave Roberts)
1379 * fixed bug 373: caused by erronous compilation of references to alien
1380 variables in the runtime on ppc/darwin.
1381 * fixed bug 376: CONJUGATE type deriver.
1382 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1383 (reported by Baughn on #lisp)
1384 * a cross-compiler bug on non-x86 platforms has been identified and
1385 fixed. (thanks to Bruno Haible)
1386 * improvements to the MIPS runtime code for increased stability.
1387 (thanks to Thiemo Seufer)
1388 * increased the maximimum compact environment size to allow
1389 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1390 * partial workaround for bug 135: don't name the function we're
1391 calling for hairy cases of EVAL, so as not to accumulate one environment
1392 entry per call to EVAL. (reported by Kevin Reid)
1393 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1394 * optimization: type testing for non-vector arrays should be faster.
1395 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1397 * bug fix: setting 31st element of a bit vector to zero did not work
1399 * bug fix: redefining a class definition which failed due to a
1400 previous accessor / function clash now works (but see BUGS entry
1401 #380 for more problems in this area). (thanks to Zach Beane)
1402 * the long form of DEFINE-METHOD-COMBINATION disables method group
1403 checking when given a single method group with pattern *. (thanks
1405 * on x86 compiler supports stack allocation of results of simple
1406 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1407 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1408 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1409 * fixed some bugs related to Unicode integration:
1410 ** the restarts for recovering from input and output encoding
1411 errors only appear when there is in fact such an error to
1413 * fixed some bugs revealed by Paul Dietz' test suite:
1414 ** MISC.549 and similar: late transformation of unsafe type
1415 assertions into derived types caused unexpected code
1417 ** SCALE-FLOAT type deriver is less wrong.
1418 ** type derivers for EXP, LOG and similar functions compute result
1419 types for complex arguments better.
1420 ** (MISC.563) CONJUGATE type deriver works for very restricted
1422 ** out-of-line type testers for character strings are available.
1423 ** EQUAL compiler transform understands specialness of objects
1425 ** accessing double-floats stored on the stack now works on x86-64.
1426 ** debugger internals could sometimes create invalid lispobjs,
1427 resulting in GC crashes.
1428 ** MISC.548: type check weakening can convert required type into
1430 ** initialization forms for bindings are not in scope of free special
1433 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1434 * incompatible change: thread support for non-NPTL systems has
1435 been removed - locking is buggy and unreliable. A threaded
1436 SBCL build will now warn at startup and refuse to create
1437 new threads, unless futex support is detected in the kernel
1438 * incompatible change: the top level REPL now has only an ABORT
1439 restart associated with it, not TOPLEVEL and ABORT as it used to.
1440 TOP and TOPLEVEL are now available as debugger commands for
1441 returning to the top level.
1442 * incompatible change: forms evaluated in the REPL now use the
1443 global optimization policy.
1444 * incompatible change: user- and system-initialization files are
1445 no longer processed with LOAD, but by READ and EVAL; hence the
1446 global optimization policy, startup package, readtable, etc,
1448 * The .fasl file format number has been incremented because of
1449 various incompatible changes.
1450 * internal entry point details and argument counts no longer appear
1451 in backtraces unless explicitly requested by setting
1452 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1453 * built-in and standard functions no longer have names like "top
1454 level local call to FOO".
1455 * fixed bug 32: functions defined in non-null lexical environments
1456 now have more legible printed representation
1457 * fixed bug 33: functions defined in non-null lexical environemnts
1458 are now more amenable to inspection by INSPECT.
1459 * workaround for bug 354: XEPs no longer appear in backtraces unless
1460 explicitly requested.
1461 * fixed bug: receiving the signal which results from attempting to
1462 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1463 SIGBUS on other *BSDs) on architectures where the C stack is also
1464 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1465 to debugger confusion or wild instability. (reported by Cheuksan
1467 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1468 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1469 (reported by Lutz Euler)
1470 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1471 compiled code. (reported by Paul Dietz)
1472 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1473 specialized parameter to a method to something that is not TYPEP
1474 the specializer is now possible.
1475 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1476 face of package deletion.
1477 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1478 pathname. (reported by David Tolpin on c.l.l)
1479 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1480 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1481 STRING-STREAM and FILE-STREAM, respectively.
1482 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1483 than previously; in addition, its test suite is now run on build.
1484 (thanks to Luke Gorrie)
1485 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1486 * optimization: added a immediate representation for single-floats
1488 * fixed some bugs revealed by Paul Dietz' test suite:
1489 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1490 ** attempting to create a package with a colliding nickname causes
1491 correctable errors to be signalled.
1492 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1493 ** several x86-64 backend bugs related to sign-extension of immediate
1496 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1497 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1498 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1500 * fixed loading of multiply forward-referenced layouts.
1501 (thanks to Cheuksan Wang)
1502 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1503 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1504 (see COMPILE-FILE documentation for details of :PRINT option)
1505 * fixed bugs 19 and 317: fixed-format floating point printing is
1506 more accurate. This also fixes a bug reported by Adam Warner
1507 related to the ~@F format directive.
1508 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1510 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1511 dispatch table if the from-char is a dispatch macro character.
1512 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1513 by the wordsize no longer ignore the last word. (reported by Lutz
1515 * fixed bug in type checking of dynamic-extent variables. (reported
1517 * optimization: sequence traversal functions use their freedom to
1518 coerce function designators to functions.
1519 * optimization: code with many calls to CLOS methods specialized on
1520 CLOS classes has had redundant type checks removed.
1521 * optimization: added declarations to speed up operations that access
1522 the internal character database (for example STRING-UPCASE)
1523 * optimization: comparison operations between floats and sufficiently small
1524 fixnums no longer create extra rationals
1525 * fixed some bugs related to Unicode integration:
1526 ** portions of multibyte characters at the end of buffers for
1527 character-based file input are correctly transferred to the
1528 start of the buffer at the next read.
1529 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1530 passing it through to OPEN.
1531 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1532 argument given to internal calls to OPEN.
1533 ** the built-in comment readers (introduced by character sequences
1534 ";" and "#|") are more forgiving to encoding errors; they will
1535 STYLE-WARN and then attempt to resync the stream at a character
1536 boundary. (thanks to Teemu Kalvas)
1537 * fixed some bugs in the x86-64 port:
1538 ** Negative short int return values from c-calls are sign-extended
1540 ** The stack is aligned to 16-bytes for c-calls, as required by
1541 the ABI. (thanks to Cheuksan Wang)
1542 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1543 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1544 for enums. (thanks to Vincent Arkesteijn)
1545 ** Multiple small optimizations and bugfixes for floating point
1547 * fixed some bugs revealed by Paul Dietz' test suite:
1548 ** Space, Tab, Linefeed, Return and Page have the invalid
1549 secondary constituent character trait.
1550 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1552 ** WITH-INPUT-FROM-STRING should only update the index place on
1554 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1555 works more reliably.
1556 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1557 with one-dimensional arrays with fill pointers.
1558 ** TYPE-OF failed on a complex with an integer realpart and a
1560 ** compiler failure during type inference for the code of form
1561 (IF (EQL X (THE ...)) ...) (MISC.535).
1563 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1564 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1565 platform. The port supports 61-bit fixnums, large memory spaces
1566 and reloading shared object files.
1567 * enhancement: saving cores with foreign code loaded is now
1568 supported on ppc/Darwin in addition to the previously supported
1570 * enhancement: the statistical profiler now walks deeper into the
1571 call stack for more meaningful call-graphs and accrued time
1572 reports (x86/x86-64 only). It also now reports time spent in
1574 * enhancement: it is now possible to trace most individual methods
1575 of a generic function in addition to tracing the generic function
1577 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1579 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1580 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1581 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1582 produces an error. (thanks to Vincent Arkesteijn)
1583 * bug fix: NAMESTRING on pathnames with :WILD components in their
1584 directories works correctly. (thanks to Artem V. Andreev)
1585 * fixed bug 125: compiler preserves identity of closures. (reported
1587 * bug fixed: functions with &REST arguments sometimes failed with
1588 "Undefined function" when compiled with (DEBUG 3). (reported by
1590 * bug fix: overflow during compiling of setting element of a bit
1591 vector with constant index and value. (reported by Timmy Douglas)
1592 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1593 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1594 * on x86 compiler partially supports stack allocation of dynamic-extent
1596 * GO and RETURN-FROM do not check the extent of their exit points
1597 when compiled with SAFETY 0.
1598 * fixed some bugs related to Unicode integration:
1599 ** encoding and decoding errors are now much more robustly
1600 handled; it should now be possible to recover even from invalid
1601 input or output to the terminal. (thanks to Teemu Kalvas)
1602 ** provided a first cut at implementing STRING-TO-OCTETS and
1603 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1604 ** altered the SB-MD5 contributed module slightly, changing the
1605 interface just enough for it to be supportable for builds where
1606 lisp characters are not eight bits.
1607 * fixed some bugs revealed by Paul Dietz' test suite:
1608 ** the FORMATTER-generated functions for ~V[ conditionals require
1609 the correct number of arguments.
1610 ** READ-FROM-STRING returns the mandated second value when applied
1611 to displaced strings.
1612 ** the #\Rubout and #\Backspace characters are treated as invalid
1613 constituent characters by the tokenizer.
1615 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1616 * new feature: reloading changed shared object files with
1617 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1618 * new feature: references to foreign variables and functions
1619 can now be compiled and loaded before the corresponding shared
1620 object file is loaded, as long as the foreign definitions are
1621 available at runtime.
1622 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1623 supported. (thanks to Dan Debertin)
1624 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1625 just keywords. (thanks to Vincent Arkesteijn)
1626 * fixed bug #331: structure-class instances corresponding to
1627 DEFSTRUCT forms are now created eagerly.
1628 * fixed bug #345: backtraces from calls to undefined functions work
1629 on x86 as well. Related bug #61 is now also partially fixed on x86
1630 (backtraces from throws to unknown catch tags.)
1631 * bug fix: lambda-list parsing is now stricter vrt. order and number
1632 of lambda-list keywords.
1633 * bug fix: as specified by AMOP, an error is signalled if a
1634 class-option appears multiple times in a DEFCLASS form. (reported
1636 * bug fix: RANDOM can be compiled when the compiler derives the type
1637 of its numeric argument as a disjoint set of small integers.
1638 (reported by Paul Dietz)
1639 * bug fix: starting a core saved with shared objects loaded when
1640 those objects are not available no longer causes threaded SBCL to
1641 hang. (reported by Sean Ross)
1642 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1644 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1645 argument, which must always be NIL. (reported by Kalle Niemitalo)
1646 * bug fix: printing 1.0d+23 no longer results in an error.
1647 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1648 * bug fix: structure slot setters preserve evaluation order. (thanks
1650 * fixed some bugs related to Unicode integration:
1651 ** RUN-PROGRAM can allow its child to take input from a Lisp
1652 stream. (reported by Stefan Scholl)
1653 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1654 locales. (reported by Ken Causey)
1655 * fixed some bugs revealed by Paul Dietz' test suite:
1656 ** INCF, DECF and REMF evaluate their place form as specified in
1658 ** FORMATTER expands ~{ ~} iteration directives with V or #
1659 parameters correctly.
1660 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1661 consequent uses no arguments correctly.
1662 ** the system has a partial understanding of the (COMPLEX RATIO)
1664 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1667 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1668 * new feature: a build-time option (controlled by the :SB-UNICODE
1669 keyword feature, enabled by default) for building the system with
1670 support for the entire 21-bit character space defined by the
1672 * new feature: the system now has rudimentary external-format
1673 support; the primary user-visible change at this time is that
1674 characters with the high bit set (such as lower-case-e-acute) will
1675 print correctly to a terminal in a UTF-8 environment.
1676 * minor incompatible change: BASE-CHAR no longer names a class;
1677 however, CHARACTER continues to do so, as required by ANSI.
1678 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1679 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1680 has been moved to the SB-EXT package (temporarily re-exported from
1682 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1683 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1685 * fixed bug #351: better error-handlind and reporting for malformed
1687 * fixed bug #350: bignum-printing is now more memory-efficient,
1688 allowing printing of very large bignums, eg. (expt 2 10000000).
1689 (reported by Bruno Haible)
1690 * fixed bug #302: better primitive-type selection for intersection
1692 * fixed bug #308: non-graphic characters now all have names, as
1693 required. (reported by Bruno Haible)
1694 * bug fix: redefining a class with different superclasses now correctly
1695 removes it from the direct-subclasses of its previous superclasses.
1696 (reported by David Morse)
1697 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1698 new value now works. (reported by Bruno Haible)
1699 * bug fix: correct canonicalization of multiple non-standard slot
1700 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1701 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1702 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1703 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1704 now exists, an signals an error.
1705 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1706 during macroexpansion for non-iteration variables as well. (reported
1707 by Bruno Haible for CMUCL)
1708 * bug fix: Cyclic structures and unprintable objects in compiler
1709 messages no longer cause errors. (reported by Bruno Haible)
1710 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1711 and READ-FROM-STRING all now return a primary value of NIL if
1712 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1713 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1714 (reported by Bruno Haible for CMUCL)
1715 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1716 do not designate a condition. (reported by Bruno Haible for
1718 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1719 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1720 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1721 permitting resolution in favour of any of the conflicting symbols.
1722 (reported by Bruno Haible for CMUCL)
1723 * bug fix: EQUAL compiler optimizations is less aggressive on
1724 strings which can potentially compare true despite having distinct
1725 specialized array element types.
1726 * bug fix: unit enumerations can be defined without dividing by
1727 zero. (thanks to Vincent Arkesteijn)
1728 * FORMAT compile-time argument count checking has been enhanced.
1729 (report from Bruno Haible for CMUCL)
1730 * a partial workaround for the bug 262: the compiler does not try to
1731 inline-expand a local function doing RETURN-FROM from a deleted
1732 BLOCK. (thanks to Peter Denno for the bug report and to David
1733 Wragg for the simple test case)
1734 * fixed some bugs revealed by Paul Dietz' test suite:
1735 ** RENAME-PACKAGE allows all package designators as new package
1737 ** constraint propagation and lambda variable substitution are
1738 more cautious in dealing with partially deleted code.
1739 ** compiler sometimes lost reoptimization passes.
1740 ** CERROR, given a condition as condition designator, passes its
1741 remaining arguments to the continue format control without
1743 ** Case-altering FORMAT directives work correctly on non-ASCII
1745 ** The REAL type specifier handles bounds outside the floating
1746 point ranges without signalling FLOATING-POINT-OVERFLOW.
1747 ** Functions with IR1-transformations can create intercomponent
1748 references to global functions.
1749 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1751 ** FORMAT ~R treats a nil value for its first parameter correctly.
1753 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1754 * enhancement: saving cores with foreign code loaded is now
1755 supported on x86/NetBSD and sparc/Linux in addition to the previously
1756 supported platforms.
1757 * bug fix: on some platforms repeated installations caused multiple
1758 copies of HTML documentation to be installed -- should not happen
1759 any more. (reported by Stefan Scholl)
1760 * bug fix: parsing self-recursive alien record types multiple times
1761 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1762 original patch by Helmut Eller for CMUCL)
1763 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1764 (thanks to Richard Kreuter)
1765 * bug fix: defining classes whose accessors are methods on existing
1766 generic functions in other (locked) packages no longer signals
1767 bogus package lock violations. (reported by François-René Rideau)
1768 * bug fix: special variables as DEFMETHOD parameters no longer have
1769 associated bogus type declarations. (reported by David Wragg and
1771 * bug fix: read-write consistency on streams of element-type
1772 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1773 * bug fix: redefiniton of the only method of a generic function with
1774 no DEFGENERIC no longer emits a full WARNING. In addition,
1775 redefinition of generic functions with no DEFGENERIC to an
1776 incompatible lambda list now signals an error. (thanks to Zach
1778 * bug fix: DEFGENERIC now works even when there's a function of the
1779 same name in an enclosing lexical environment. (thanks to Zach
1781 * fixed compiler failure, caused by instrumenting code during
1782 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1783 * optimization: added loop analysis and improved register allocation
1784 * fixed some bugs revealed by Paul Dietz' test suite:
1785 ** POSITION on displaced vectors with non-zero displacement
1786 returns the right answer.
1787 ** (SIMPLE-STRING) is a valid type specifier for sequence
1789 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1791 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1792 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1794 ** The justification version of the FORMAT ~< directive treats
1795 non-zero minpad parameter correctly.
1797 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1798 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1799 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1800 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1801 the supported interface.
1802 * new feature: Single-stepping of code compiled with DEBUG 2 or
1803 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1804 * new feature: saving cores with foreign code loaded is now
1805 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1806 Timothy Moore's work for CMUCL)
1807 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1808 parameters to * instead of NIL if no initform is supplied.
1809 (reported by Johan Bockgaard)
1810 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1811 correctly when FUNCALL appears as the car of the form. Note:
1812 despite this FUNCALL forms are not currently subject to
1813 compiler-macro expansion. (port of Raymond Toy's fix for the
1814 same from CMUCL, reported by Johan Bockgaard)
1815 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1816 (thanks for Teemu Kalvas)
1817 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1818 case instead of terminating the iteration (thanks for Julian
1819 Squires, Sean Champ and Raymond Toy)
1820 * bug fix: incorrect expansion of defgeneric that caused a style
1821 warning. (thanks for Zach Beane)
1822 * bug fix: slot accessor effective method computation works properly
1823 for classes with multiple non-standard applicable methods on
1824 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1825 * on x86 compiler supports stack allocation of results of LIST and
1826 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1827 CMUCL implementation by Gerd Moellmann)
1828 * fixed some bugs revealed by Paul Dietz' test suite:
1829 ** FORMAT strings with both the ~<~:;~> form of the justification
1830 directive and pretty-printing directives cause an error.
1832 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1833 * incompatible change: the internal functions
1834 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1835 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1836 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1837 instead of the old functions.
1838 * new feature: on platforms where "dladdr" is available foreign
1839 function names now appear in backtraces. (based on Helmut Eller's
1841 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1842 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1844 * documentation: ASDF manual is now distributed with SBCL.
1845 * bug fix: SBCL can now load its contributed modules with REQUIRE
1846 even if the system-provided entries have been removed from
1847 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1849 * bug fix: backtraces involving undefined functions or assembly
1850 routines are more informative. (thanks to Brian Downing)
1851 * bug fix: mutually referent alien structures now work correctly.
1852 (reported by Rick Taube)
1853 * bug fix: structures defined by WITH-ALIEN can be referred to
1854 within other definitions in the same WITH-ALIEN.
1855 * bug fix: division operators (MOD, TRUNCATE and the like) with
1856 constant zero divisors and integer dividends no longer generate
1858 * bug fix: provide default methods for INPUT-STREAM-P and
1859 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1860 * bug fix: improve the detection and resolution of MOP metacycles
1861 (where effective method computation depends on the generic function
1862 having its effective method computed). (reported by Bruno Haible)
1863 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1864 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1865 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1866 * bug fix: pathnames with relative directory components can be
1867 represented relative to default pathnames.
1868 * optimization: in taking the GCD of bignums, reduce the two bignums
1869 to approximately the same size (using Euclid's algorithm) before
1870 applying the more sophisticated binary GCD. (thanks to Juho
1872 * optimization: COUNT on bitvectors now operates word-at-a-time.
1873 * optimization: ASH with a positive, but not necessarily constant,
1874 (leftwards) shift, when in a modular context, is compiled to a
1876 * fixed some bugs revealed by Paul Dietz' test suite:
1877 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1878 if the corresponding argument is NIL.
1879 ** FORMAT directives accepting a minpad parameter treat negative
1881 ** literal commas as character parameters to FORMAT directives are
1883 ** literal spaces directly after ~<Newline> directives within a
1884 format-logical-block (~:< ~@:>) do not induce :FILL-style
1885 conditional newlines.
1886 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1887 as specified: it no longer includes conditional newlines).
1888 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1890 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1891 :PER-LINE-PREFIX argument does not evaluate to a string.
1893 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1894 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1895 the manual for details; for now, package locks can be disabled by
1896 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1897 if no major problems are found then it is likely that they will be
1898 compiled in unconditionally.
1899 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1900 unsupported operators on all platforms. To load a shared library
1901 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1902 object file, link it into a shared library outside of SBCL and
1903 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1905 * bug workaround: ROOM T and the GC were not working together
1906 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1907 reporting facility aren't preserved. That reporting has been
1908 disabled, so now until and unless someone figures out how to make it
1909 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1910 * minor incompatible change: as threatened around sbcl-0.8.0, the
1911 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1912 The symbols are also exported from SB-PCL for backwards
1913 compatibility, but more so than before SB-PCL should be treated as
1914 an implementation-internal package.
1915 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1917 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1918 It is known as of this release not to work on the Alpha, however.
1919 * fixed bug #167: errors signalled due to illegal syntax in method
1920 bodies are now more legible.
1921 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1922 designators and can hence be used with TYPEP.
1923 * fixed bug #333: CHECK-TYPE now ensures that the type error
1924 signalled, if any, has the right object to be accessed by
1925 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1926 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1927 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1929 * fixed bug #334: programmatic addition of slots using specialized
1930 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1931 and :ALLOCATION :CLASS slots.
1932 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1933 float exponents. (rereported by Peter Seibel)
1934 * fixed a bug: #\Space (and other whitespace characters) are no
1935 longer considered to be macro characters in standard syntax by
1936 GET-MACRO-CHARACTER.
1937 * fixed bug: initialization of condition class metaobjects no longer
1938 causes an instance of the condition to be created. (reported by
1940 * fixed bug: it is now possible to have slots such that
1941 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1942 description is neither :INSTANCE nor :CLASS.
1943 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1944 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1945 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1946 non-standard SLOT-DEFINITION-ALLOCATION values.
1947 * fixed bugs: various race conditions handling exiting threads.
1948 CL-PPCRE's thread torture test now passes.
1949 * fixed arguably-a-bug: GC time proportional to number of threads in
1950 system even when most of them are idle
1951 * optimization: improved performance of BIT and SBIT on bit-vectors.
1952 * .fasl file incompatibility: The fasl file version number has
1953 been incremented because of changes associated with package locks.
1955 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1956 * minor incompatible change: the system no longer provides
1957 optimization hints (conditions which are TYPEP
1958 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1959 mode; these hints are still emitted when higher SPEED optimization
1961 * new contrib module: a sampling profiler (profiling by statistical
1962 sampling, rather than by instrumenting functions) is available as
1963 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1964 * the behaviour of the standard function ED is now customizeable by
1965 third parties through a hook variable: see ED's documentation
1966 string for information on the protocol.
1967 * the compiler no longer emits efficiency notes for (FUNCALL X)
1968 when the type of X is uncertain under default optimization
1970 * fixed bug 276: mutating a binding of a specialized parameter to a
1971 method to something that is not TYPEP the specializer is now
1973 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1974 exhibit the required behaviour on the x86 platform. (thanks to
1975 Peter van Eynde, Eric Marsden and Bruno Haible)
1976 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1977 even for difficult arguments. (reported by Peter Graves)
1978 * fixed bug 141a: the backquote printer now descends quoted
1980 * fixed another bug in backquote printing: no more destructive
1981 modification of the form's list structure. (reported by Brian
1983 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1984 move between its address being taken and the call to
1985 interrupt_thread, fixing a crashing race condition.
1986 * the SB-POSIX contrib implementation has been adjusted so that it
1987 no longer exhibits ridiculously poor performance when constructing
1988 instances corresponding to C structs.
1990 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1991 * minor incompatible change: the sb-grovel contrib now treats C
1992 structures as alien (in the sense of SB-ALIEN) objects rather than
1993 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1994 has implications for memory management of client code
1995 (sb-grovel-returned objects must now be manually managed) and for
1996 type safety (alien objects now have full types).
1997 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1998 used to control emission of compiler diagnostics, rather than the
1999 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2000 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2001 quality should be considered deprecated.
2002 * install.sh now installs the user manual as well
2003 * (not quite a new documentable feature, but worth considering in
2004 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2005 beginnings of a semantically meaningful condition hierarchy is
2006 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2008 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2009 designator as the defaults argument.
2010 * fixed bug: Displaced arrays whose displaced-to array has become
2011 too small now cause ARRAY-DIMENSION to signal an error, providing
2012 for safer bounds-checking. (reported by Bruno Haible)
2013 * fixed bug: DEFCLASS slot definitions with identical :READER and
2014 :WRITER names now signal a reasonable error. (reported by Thomas
2016 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2018 * fixed bug: Invalid slot specification errors now print correctly.
2019 (thanks to Zach Beane)
2020 * fixed bug 320: Shared to local slot value transfers in class
2021 redefinitions now happen corrently from superclasses as
2022 well. (reported by Bruno Haible)
2023 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2025 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2026 improper lists correctly. (reported by Bruno Haible)
2027 * fixed bug 313: source-transform for <fun-name> was erroneously
2028 applied to a call of a value of a variable with name <fun-name>.
2029 (reported by Antonio Menezes Leitao)
2030 * fixed bug 307: The obsolete instance protocol ensures that
2031 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2032 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2033 remove dead unknown-values globs from the middle of the stack.
2034 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2035 easier to resume long computations after using *BREAK-ON-SIGNALS*
2036 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2037 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2038 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2039 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2040 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2041 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2043 * fixed a (fairly theoretical) bug in string printing: if
2044 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2045 does not have array-element-type equal to the most general string
2047 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2048 function name. (thanks to Nikodemus Siivola)
2049 * fixed bug: compile-time format string checker failed on
2050 non-closed ~{. (reported by Thomas F Burdick)
2051 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2052 don't warn on legal loop code involving more than one aggregate
2053 boolean. (thanks to Nikodemus Siivola)
2054 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2055 in keyword position to MAKE-PACKAGE was not being reported as
2057 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2058 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2060 * fixed bugs 280 and 312: the checking for multiple definitions in a
2061 file is less likely to become confused by uses of inline
2063 * fixed bug: the #S reader macro performs the keyword coercion
2064 specified for slot names. (reported by Kalle Niemitalo)
2065 * fixed bug: lambda lists may contain symbols whose names start with
2066 & but are not lambda-list-keywords; their occurrence triggers a
2068 * fixed bug 321: define-method-combination argument lambda lists do
2069 not cause invalid code to be generated when &optional or &aux
2070 variables have default values. (reported by Bruno Haible)
2071 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2072 have CLOS classes; this appears also to have fixed reported
2073 instability in stack exhaustion detection.
2074 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2075 MUFFLE-WARNING when no associated restart is present is now
2077 * optimization: rearranged the expansion of various defining macros
2078 so that each expands into only one top-level form in a
2079 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2081 * optimization: used a previously little-used slot in symbols to
2082 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2084 * fixed some bugs revealed by Paul Dietz' test suite:
2085 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2087 ** Fixed an optimization bug related to inheritance of initforms
2088 from local to shared slots.
2089 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2090 ** CAST optimizer forgot to flush argument derived type.
2091 ** print/read consistency on floats is now orders of magnitude
2092 more likely. (thanks also to Bruno Haible for a similar report
2094 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2095 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2096 for objects of type REAL. Make it so.
2097 ** SXHASH is specified (infelicitously) to respect similarity,
2098 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2099 it so. (thanks to Markus Ziegler)
2100 ** on the Alpha, the compiler succeeds in compiling functions
2101 returning a known number of arguments greater than 63.
2102 ** fixed handling of invalid NIL arguments in keyword position
2104 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2106 ** provided a readably-printable representation for RANDOM-STATE
2108 ** ensured that pathnames, where they have a namestring, always
2109 print using #P"..." syntax.
2111 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2112 * Support for the forthcoming 2.0 version of the NetBSD kernel
2113 running on x86 hardware has been added. (thanks to Perry
2114 E. Metzger most immediately, and others for their past work)
2115 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2116 current release version; SBCL's OpenBSD support had been broken
2117 since about the time of OpenBSD's switch to ELF binary format.)
2118 * [placeholder for DX summary]
2119 ** user code with &REST lists declared dynamic-extent, under high
2120 speed or space and low safety and debug optimization policy.
2121 * The manual has been converted to Texinfo format and the debugger
2122 chapter from the cmucl manual has been added.
2123 * A facility has been added to extract documentation strings from
2124 sbcl and store them as Texinfo-formatted snippets for inclusion in
2125 the manual (via Texinfo's @include directive)
2126 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2127 the test case to Dave Roberts)
2128 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2129 pointers. (reported by Sean Ross)
2130 * bug fix: PROFILE output is printed nicely even for large numerical
2131 values. (thanks to Zach Beane)
2132 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2133 greater than 32 handle EOF correctly.
2134 * bug fix: on X86 an immediate argument of the IMUL instruction is
2135 correctly printed in disassembly. (thanks to Lutz Euler)
2136 * bug fix: class slots in redefined classes preserve their old
2137 values. (thanks to Bruno Haible and Nikodemus Siivola)
2138 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2139 succeeds. (reported by Marco Baringer)
2140 * fixed some bugs revealed by Paul Dietz' test suite:
2141 ** READ-SEQUENCE now works on ECHO-STREAMs.
2142 ** RATIONALIZE works more according to its specification. (thanks
2145 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2146 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2147 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2148 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2149 should matter to you only if you rebind the printer control
2150 variables and then find you want different bindings in the
2151 debugger than in the ordinary execution of your program.)
2152 * The runtime build system has been tweaked to support building
2153 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2154 assembler and linker. (thanks to Nikodemus Siivola)
2155 * Unbound, undefined, undeclared variables now trigger full
2156 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2157 more often programmer error than deliberate exploitation of undefined
2159 * optimization: the hash algorithm for strings has changed to one
2160 that is less vulnerable to spurious collisions. (thanks to Juho
2162 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2163 needless bounds checking. (thanks to Juho Snellman)
2164 * optimization: implemented multiplication as a modular
2165 (UNSIGNED-BYTE 32) operation on the PPC backend.
2166 * fixed some bugs revealed by Paul Dietz' test suite:
2167 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2169 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2170 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2171 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2173 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2175 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2177 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2178 their output stream on EOF from read.
2179 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2180 have been read to end-of-file.
2181 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2183 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2184 description of determination of which consecutive characters
2186 ** Printing the "Space" character with escaping on now yields "#\\ ",
2187 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2188 ** Reading floating-point numbers with *READ-BASE* set to a number
2189 less than 10 works correctly.
2190 ** Reading floating-point numbers with *READ-BASE* set to a number
2191 more than 10 works correctly.
2192 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2193 the readtable currently in effect.
2195 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2196 * minor incompatible change: parsing of namestrings on a physical
2197 (Unix) host has changed; numbers after the final #\. in a
2198 namestring are no longer interpreted as a version field. This is
2199 intented to be largely invisible to the user, except that the
2200 meaning of the namestring "*.*.*" has changed: it now refers to a
2201 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2202 should usually be replaced by
2203 (make-pathname :name :wild :type :wild :version :wild)
2204 with the added benefit that this is more likely to be portable.
2205 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2206 option to OPEN now signals an error if the file being opened
2207 exists; this may have an impact on existing code.
2208 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2209 signals received, on the PowerPC platforms (both Linux and
2210 Darwin). (thanks to Pierre Mai for pointing out the location of
2212 * several fixes on OS X: The system now builds and runs cleanly on
2213 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2214 (thanks to Brian Mastenbrook)
2215 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2216 second-resolution: integer multiples of 1/3600 between -24 and 24.
2217 (thanks to Vincent Arkesteijn)
2218 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2219 type when called with 1 argument; PEEK-CHAR checked type of
2220 PEEK-TYPE only after having read first character from a
2221 stream. (reported by Peter Graves)
2222 * bug fix: the garbage collector now has much better locality
2223 behaviour, and in particular no longer treats all memory as being
2224 exhausted when many small objects point to each other in a deeply
2226 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2227 recognized as being TYPEP their class.
2228 * bug fix: the PUSHNEW documentation string has been corrected.
2229 (thanks to Vincent Arkesteijn)
2230 * bug fix: defaulting of the value for the last of an atypically
2231 large number of multiple values being bound was not being
2232 performed correctly on the Alpha or PPC platforms
2233 * optimization: implemented multiplication as a modular
2234 (UNSIGNED-BYTE 32) operation on the x86 backend.
2235 * optimization: SEARCH on simple-base-strings can now be open-coded.
2236 (see also contrib/compiler-extras.lisp for inspiration for
2237 teaching the compiler about the Boyer-Moore algorithm).
2238 * value, returned by MAX (and MIN) called with several EQUALP, but
2239 not EQL, arguments now does not depend on compiler settings.
2240 * fixed some bugs revealed by Paul Dietz' test suite:
2241 ** in stack analysis liveness information is propagated from
2242 non-local entry points.
2243 ** pathwise CAST removing failed when the CAST node did not start
2245 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2246 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2248 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2249 host is already defined.
2250 ** RENAME-FILE works on streams instead of signalling an internal
2252 ** PEEK-CHAR uses the current readtable when determining whether
2253 or not a character is whitespace.
2254 ** MERGE-PATHNAMES handles the case when the pathname does not
2255 specify a name while the default-pathname specifies a version.
2256 ** Pathnames now stand a better chance of respecting print/read
2258 ** Attempting to use standardized file system operators with a
2259 pathname with invalid :DIRECTORY components signals a
2261 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2263 ** DIRECTORY on logical pathnames is more correct.
2264 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2265 signal a TYPE-ERROR if their argument is not a stream.
2266 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2267 designator argument does not designate a stream.
2268 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2269 examining the synonym.
2270 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2272 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2273 element-type larger than ([UN]SIGNED-BYTE 32).
2275 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2276 * When built with the :SB-FUTEX feature, threaded builds now take
2277 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2278 for faster/more reliable mutex and condition variable support.
2279 * Incompatible change (but one you probably shouldn't have been using
2280 anyway): the interface and code for arbitrating between multiple
2281 threads in the same user session has been redesigned.
2282 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2283 error when called without an explicit environment argument.
2284 (thanks to Nikodemus Siivola)
2285 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2286 confused about their position. (thanks to Adam Warner and Gerd
2288 * bug fix: Pretty printing backquoted forms with unquotations in the
2289 argument list position of various code constructs such as LAMBDA
2290 now works correctly. (reported by Paul Dietz)
2291 * bug fix: Pretty printing unquotations no longer loses all
2292 stream position information.
2293 * optimization: performance of string output streams is now less
2294 poor for multiple small sequence writes.
2295 * optimization: performance of CSUBTYPEP in the presence of complex
2296 expressions involving CONS and NOT many times has been improved.
2297 (reported by Paul Dietz)
2298 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2300 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2301 streams for the REPL, for the debugger, and so on. (thanks to
2303 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2304 * fixed some bugs revealed by Paul Dietz' test suite:
2305 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2306 is ignored if it is too small, rather than propagating through
2307 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2308 ** extremely complex negations of CONS types were not being
2309 sufficiently canonicalized, leading to inconsistencies in
2311 ** VALUES tranformer lost derived type.
2313 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2314 * fixed a bootstrapping bug: the build process no longer assumes
2315 that the various BOOLE-related constants have the same value in
2316 host and target lisps. (noted by Paul Dietz' test suite on an
2317 SBCL binary built from CLISP)
2318 * The system can now be dynamically linked on the MIPS platform,
2319 which enables dynamic loading of foreign code from Lisp. (thanks
2320 to Ralf Baechle for discussions on the MIPS ABI)
2321 * The system now records debugging information for its own source
2322 files in a filesystem-position-independent manner, relative to
2323 the "SYS" logical host.
2324 * fixed a compiler bug: MV-LET convertion did not check references
2325 to the "max args" entry point. (reported by Brian Downing)
2326 * tweaked disassembly notes to be less confident about proclaiming
2327 some instruction as an LRA. (thanks to Brian Downing)
2328 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2329 now each have their own history, command character, and other
2330 characteristics. (thanks to David Lichteblau)
2331 * fixed some bugs revealed by Paul Dietz' test suite:
2332 ** compiler failure in compiling LOGAND expressions including a
2334 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2335 shift greater than 32.
2336 ** FLUSH-DEST did not mark blocks for type check regeneration.
2337 ** HANDLER-CASE failed to accept declarations in handler clauses
2338 in some circumstances.
2340 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2341 * New code in contrib/sb-introspect (still probably not entirely
2342 stable yet) provides some support for smart Lisp development
2343 environments like SLIME.
2344 * The conditions signalled for errors occurring when loading .fasl
2345 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2346 in a way which should help ASDF recover gracefully.
2347 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2348 stuff has been cleaned up. If you code contrib/ stuff, this might
2349 affect you, and you can look at contrib/README, contrib/STANDARDS,
2350 and/or the 0.8.4.27 diff to check.
2351 * In full calls the compiler now does not generate checks for declared
2352 argument types for all arguments.
2353 * various threading fixes
2354 ** and some experimental patches which didn't make it into
2355 the main tree for this release, but which are shipped in
2356 contrib/experimental-thread.patch as a possible fix for some
2357 failures (deadlock, spinning...) in GC-intensive multithreaded
2359 * fixed PPC build problem (source code incompatibility of different
2360 library versions): added offsetof() hackery which attempts to divine
2361 where glibc maintainers put uc_mcontext today
2362 * fixed bug 282: compiler does not trust type assertions while passing
2363 arguments to a full call.
2364 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2365 &OPTIONAL VALUES type parameter.
2366 * fix bug 214: algorithm for noting rejected templates is now more
2367 similar to that of template seletion. (also reported by rydis on
2369 * fixed bug 141b: printing backquoted information readably and prettily
2370 inserts a space where necessary.
2371 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2372 (CHARACTER 10) are now reported as errors, rather than propagated
2373 as unknown types. (reported by piso on #lisp)
2374 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2375 with values NIL and :ERROR. (thanks to Milan Zamazal)
2376 * fixed bug 191c: CLOS now does proper keyword argument checking as
2377 described in CLHS 7.6.5 and 7.6.5.1.
2378 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2379 longer raise an error; further, using a list as a for-as-arithmetic
2380 counter now raises a meaningful error.
2381 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2382 now understood by sequence creation functions such as MAKE-SEQUENCE
2384 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2385 a STRING-INPUT-STREAM.
2386 * compiler enhancement: SIGNUM is now better able to derive the type
2388 * type declarations inside WITH-SLOTS are checked. (reported by
2390 * fixed some bugs revealed by Paul Dietz' test suite:
2391 ** incorrect optimization of TRUNCATE for a positive first
2392 argument and negative second.
2393 ** compiler failure in let-convertion during flushing dead code.
2394 ** compiler failure while deriving type of TRUNCATE on an
2395 interval, containing 0.
2396 ** ASH of a negative bignum by a negative bignum count now returns
2398 ** intersection of CONS types now canonicalizes properly, fixing
2399 inconsistencies in SUBTYPEP.
2401 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2402 * incompatible change: The --disable-debugger command line
2403 option now clobbers the debugger at a more fundamental
2404 level, by redefining #'INVOKE-DEBUGGER instead of by
2405 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2406 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2407 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2408 enter the debugger REPL and then suffer recursive errors
2409 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2410 while under the new system, BREAK in --disable-debugger mode
2411 terminates the system just as an unhandled error would.
2412 * fixed compiler performance when processing loops with a step >1;
2413 * bug fix: DOCUMENTATION now retrieves generic function
2414 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2415 support has been systematized, and now supports the methods
2416 specified by ANSI, along with a default method and a method for
2417 slot documentation. (reported by Nathan Froyd)
2418 * bug fix: effective methods associated with a generic function are
2419 no longer cached over a change of that generic function's method
2420 combination. (reported by Andreas Fuchs)
2421 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2422 names a non-existent directory. (thanks to Andreas Fuchs)
2423 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2424 the PPC platform, lead to stack corruption; this has been fixed.
2425 (reported by Rainer Joswig)
2426 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2427 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2428 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2429 arithmetic in general, is now much more reliable on the PPC
2431 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2432 platform now returns the right answer.
2433 * optimization: restored some effective method precomputation in
2434 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2435 precomputation is now tunable.
2436 * optimization: compiler-internal data structure use has been
2437 reviewed, and changes have been made that should improve the
2438 performance of the compiler by about 20%.
2439 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2440 simple-base-strings and simple-bit-vectors is improved.
2441 * optimization: the optimization of 32-bit logical and arithmetic
2442 functions introduced in version 0.8.3 on the x86 has been
2443 implemented on the mips, ppc and sparc platforms; an
2444 implementation of the same facility, but for 64-bit arithmetic,
2445 has been added for the alpha.
2446 * microoptimization: the compiler is better able to make use of the
2447 x86 LEA instruction for multiplication by constants.
2448 * bug fix: in some situations compiler did not report usage of
2449 generic arithmetic in (SPEED 3) policy.
2450 * bug 145b fix: compiler used wrong type specifier while converting
2451 MEMBER-types to numeric.
2452 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2454 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2455 index argument now works properly on the Alpha platform.
2456 * bug fix: floating point exception treatment on the Alpha platform
2458 * bug fix: FILE-POSITION works much better on string input and
2459 output streams. (thanks to Nikodemus Siivola)
2460 * bug fix: many threading/garbage collection symptoms sorted.
2461 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2462 might be pseudo-atomic.
2463 * internal change: Stopping for GC is now done with signals not
2464 ptrace. GC is now done in whichever thread wanted it, instead of
2466 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2468 * bug fix: result form in DO is not contained in the implicit
2470 * incompatible change: ICR structure is changed; the value part of
2471 CONTINUATION is now called LVAR; corresponding functions are
2472 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2473 * added type deriver for ISQRT (thanks to Robert E. Brown).
2474 * bug fix: better support for loading from the command line when an
2475 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2477 * fixed some bugs revealed by Paul Dietz' test suite:
2478 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2479 ** ROUND and FROUND now give the right answer when given very
2480 small float arguments.
2481 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2483 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2484 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2485 ** LCM should return a non-negative integer.
2486 ** PARSE-INTEGER returned the index of a terminator instead of the
2487 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2488 ** PARSE-INTEGER returned an incorrect index being applied to a
2490 ** LCM with two arguments of 0 returns 0 rather than signalling
2492 ** unsigned addition of a 32-bit constant with the high bit set no
2493 longer causes an internal compiler error on the x86.
2494 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2495 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2496 with negative last argument.
2497 ** byte specifiers with zero size and position no longer cause
2498 an error during type derivation.
2499 ** bignum multiplication on the Alpha platform now returns the
2501 * porting: The system now builds on SuSE AMD64, although it still
2502 generates a 32-bit binary.
2503 * .fasl file incompatibility: The fasl file version number has
2504 been incremented (because of the changes to internal compiler
2505 data structures referred to above).
2507 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2508 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2509 more accurately, on the Darwin kernel running on PowerPC hardware.
2510 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2511 * Compiler code deletion notes now signal a condition of type
2512 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2513 an associated MUFFLE-WARNING restart.
2514 * The compiler now performs limited argument count validation of
2515 constant format strings in FORMAT, and where appropriate in ERROR,
2516 CERROR and WARN. (thanks to Gerd Moellmann)
2517 * New ASDF-INSTALL contrib can be used for automatic download and
2518 installation of third-party Lisp code from CCLAN or other sites
2520 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2521 which forces another thread to execute a function supplied by the
2523 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2524 now accept and act upon their :ELEMENT-TYPE keyword argument.
2525 (reported by Martin Atzmueller, Edi Weitz)
2526 * bug fix: FILE-POSITION now accepts position designators up to
2527 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2528 is the greater. (thanks to Patrik Nordebo)
2529 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2530 Kalle Olavi Niemitalo)
2531 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2532 printable. (reported by Eric Marsden)
2533 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2534 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2535 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2536 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2537 (thanks to Stig Sandoe)
2538 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2539 UNREAD-CHAR now correctly includes the unread character in the
2540 target sequence. (thanks to Gerd Moellmann)
2541 * bug fix in threaded builds: the system can now be suspended and
2542 resumed by shell job control with minimal disruption.
2543 * bug fixes in times and timezones >2038AD
2544 * better handling of "where is GNU make?" problem in build scripts
2545 (thanks to Nikodemus Siivola)
2546 * new optimization: inside a named function any reference to a
2547 function with the same name is considered to be a self-reference;
2548 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2549 optimization quality.
2550 * new optimization on x86: logical functions and + now have
2551 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2552 used when the result is truncated to 32 bits.
2553 * VALUES declaration is partially enabled.
2554 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2555 * fixed some bugs revealed by Paul Dietz' test suite:
2556 ** The system now obeys the constraint imposed by
2557 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2558 types form a lattice under type intersection.
2559 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2560 ** ASSOC now ignores NIL elements in an alist.
2561 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2562 and (1+ MOST-POSITIVE-FIXNUM) answers.
2563 ** The addition of a method with invalid qualifiers to a generic
2564 function does not cause an error to be signalled immediately;
2565 a warning is signalled, and the error is generated only on
2566 calling the generic function.
2567 * changed .fasl file version number, in part to add type codes for
2568 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2569 obscure ANSI requirements
2571 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2572 * fixed bug 148: failure to inline-expand a local function left
2573 garbage, confusing the compiler.
2574 * fixed bugs 3cd: structure slot readers perform type check if the
2575 slot can have an invalid value (i.e. it is either not initialized
2576 or can be written with a less specific slot writer).
2577 * bug fix: the compiler now traps array references to elements off
2578 the end of an array; previously, the bounds checking in some
2579 circumstances could go off-by-one.
2580 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2582 * type declarations for array element types now obey the description
2583 on the CLHS page "Declaration TYPE", as per discussions on
2584 sbcl-help around 2003-05-08. This means that a declaration
2585 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2586 declaration, all references to BAR will be asserted or assumed
2587 (with THE, so dependent on compiler policy) to involve objects of
2588 type FOO. Note that no such declaration is implied in
2589 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2590 * declared types of functions from the "Conditions"
2591 chapter. (reported by Paul Dietz)
2592 * bug fix: CERROR accepts a function as its first argument.
2593 * bug fix: NTH an NTHCDR accept a bignum as index
2594 arguments. (reported by Adam Warner)
2595 * optimization: character compare routines now optimize comparing
2596 against a constant character. (reported by Gilbert Baumann)
2597 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2598 argument now works properly on the MIPS platform.
2599 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2600 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2601 declared to have a functional type, does not cause an error
2603 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2604 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2605 * fixed bug: if last continuation of a deleted block has a
2606 destination, this destination should be deleted too. (reported by
2608 * fixed a bug in the bootstrap process: the host compiler's values
2609 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2610 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2611 test case from Patrik Nordebo)
2612 * improved the ability of the disassembler on the PPC platform to
2613 provide helpful disassembly notes.
2614 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2615 the class in more cases than previously.
2616 * bug fix: FILE-POSITION now understands :START and :END for
2617 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2618 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2619 8) data. (thanks to David Lichteblau)
2620 * bug fix: it is possible to add a method to a generic function
2621 without lambda list.
2622 * bug fix: reader failed to signal END-OF-FILE inside an
2623 object representation. (reported by Nikodemus Siivola)
2624 * fixed some bugs revealed by Paul Dietz' test suite:
2625 ** LAST and [N]BUTLAST should accept a bignum.
2626 ** condition slot accessors are methods.
2627 ** (VECTOR NIL) is a subtype of STRING.
2629 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2630 * minor incompatible change: some nonsensical specialized lambda
2631 lists (used in DEFMETHOD) which were previously ignored now signal
2633 * minor incompatible change: the system is now aware of the types of
2634 variables in the COMMON-LISP package, and will signal errors for
2635 most violations of these type constraints (where previously they
2636 were silently accepted).
2637 * minor incompatible change: COMPILE-FILE now uses the freedom
2638 afforded (ANSI 3.2.2.3) to use derived function types for
2639 functions defined in the same file. This also permits the system
2640 to warn on static type mismatches and function
2641 redefinition. (Currently it does not work with high DEBUG level.)
2642 * minor incompatible change: VALUES declaration is disabled.
2643 * When issuing notes, the compiler now signals a condition of type
2644 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2645 restart for use in user handlers. It is expected that the
2646 COMPILER-NOTE condition will eventually become a condition
2647 supertype to a hierarchy of note types, which will then be
2648 handleable in a similar fashion. However, at the moment, no such
2649 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2650 but it's an implementation detail, not a classification for the
2652 * Changes in type checking closed the following bugs:
2653 ** type checking of unused values (192b, 194d, 203);
2654 ** template selection based on unsafe type assertions (192c, 236);
2655 ** type checking in branches (194bc).
2656 * A short form of VALUES type specifier has ANSI meaning (it has
2657 increased the number of situations when SBCL cannot perform type
2659 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2660 keywords or constants is permissible.
2661 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2662 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2663 classes. (thanks to Antonio Martinez)
2664 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2665 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2666 argument to a pretty-print stream. Instead, it leaves any such
2667 operations to DESCRIBE-OBJECT methods.
2668 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2670 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2671 lambda lists. (thanks to David Lichteblau)
2672 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2673 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2674 (thanks to Gerd Moellmann)
2675 * fixed bug 246: increased compilation speed of long
2676 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2678 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2680 * DEFCONSTANT now throws a condition of type
2681 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2682 constant to a non-EQL value; CONTINUE and ABORT restarts
2683 respectively change and preserve the value.
2684 * fixed bug 63: The code walker, part of the implementation of CLOS,
2685 is now better at handling symbol macros.
2686 * bug fix: There is no longer an internal implementation type named
2687 CL:LENGTH. (reported by Raymond Toy)
2688 * bug fix: In macro-like defining macros/special operators the
2689 implicit block does not enclose the lambda list.
2690 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2691 enclosed in parentheses) are not suitable as type specifiers, and
2692 their use properly signals an error now.
2693 * bug fix: An argument count mismatch for a type specifier in code
2694 being compiled no longer causes an unhandled error at compile
2695 time, but signals a compile-time warning.
2696 * fixed simple vector readable printing
2697 * bug fix: DESCRIBE takes more care over whether the class
2698 precedence list slot of a class is bound before accessing it.
2699 (reported by Markus Krummenacker)
2700 * bug fix: FORMATTER can successfully compile pretty-printer format
2701 strings which use variants of the ~* directive inside.
2702 * bug fix: SEARCH now applies its TEST predicate to the elements of
2703 the arguments in the correct order. (thanks to Wolfhard Buss)
2704 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2705 in type inference. (thanks to Robert E. Brown)
2706 * bug fix: Reading in symbols with an explicit package name of ""
2707 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2708 * fixed some bugs revealed by Paul Dietz' test suite:
2709 ** NIL is now allowed as a structure slot name.
2710 ** Arbitrary numbers, not just REALs, are allowed in certain
2711 circumstances in LOOP for-as-arithmetic clauses.
2712 ** Multiple class redefinitions before slot access no longer
2713 causes a type error.
2714 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2715 association between the name and a class.
2716 ** Generic functions with non-standard method-combination and over
2717 five methods all of which return constants no longer return NIL
2718 after the first few invocations. (thanks to Gerd Moellmann)
2719 ** CALL-NEXT-METHOD with no arguments now passes the original
2720 values of the arguments, even in the presence of assignment.
2721 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2723 ** DISASSEMBLE works with closures and funcallable instances.
2724 ** ADD-METHOD now returns the generic function, not the new method.
2725 ** FIND-METHOD signals an error if the lengths of the specializers
2726 is incompatible with the generic function, even if the ERRORP
2728 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2729 which its argument is a member.
2730 ** DEFCLASS only redefines the class named by its class-name
2731 argument if that name is the proper name of the class;
2732 otherwise, it creates a new class.
2733 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2734 of the UNBOUND-SLOT condition to the name of the slot.
2735 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2736 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2737 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2738 SLOT-MAKUNBOUND in the specified fashion.
2740 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2741 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2742 cross-compilation host. As a consequence, we can now bootstrap our
2743 way up to SBCL starting with a bare gcc toolchain and human-readable
2744 source code (first the source to CLISP, then the source to SBCL).
2745 * A contributed module containing a partial implementation of the
2746 simple-streams interface has been included. (thanks to Rudi
2748 * A contributed module implementing the RFC1321 Message Digest
2749 Algorithm, known as MD5, has been included.
2750 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2751 no longer has any effect, as the code controlled by this feature
2752 has been deleted. (As far as we know, no-one has ever built using
2753 this feature, and its semantics were confused in any case).
2754 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2755 work on conditions (as required by the ANSI specification),
2756 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2757 expected behaviour on conditions. Users should note, however,
2758 that such behaviour is not required by the ANSI specification,
2759 and so use of this behaviour may render their code unportable.
2760 * fixed some bugs revealed by Paul Dietz' test suite:
2761 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2763 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2764 ** SXHASH on condition objects no longer returns NIL.
2765 ** :ALLOCATION :CLASS slots are better treated; their values are
2766 updated on class redefinition, and initforms inherited from
2767 superclasses are applied.
2768 ** REMOVE-METHOD returns its generic function argument even when
2769 no method was removed.
2770 ** SHARED-INITIALIZE now initializes the values of the requested
2771 slots, including those with :ALLOCATION :CLASS.
2772 ** ALLOCATE-INSTANCE now works on structure classes defined via
2773 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2775 ** SLOT-EXISTS-P now works on conditions, as well as structures
2777 ** MAKE-LOAD-FORM now has the required methods on
2778 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2779 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2780 keyword, and now implements the SLOT-NAMES argument.
2781 ** methods with &OPTIONAL arguments no longer allow too many
2782 arguments to be passed in the call without error.
2783 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2784 option is consistent with the required arguments of the generic
2785 function lambda list.
2786 * bug fix: REQUIRE accepts a string designator. (Thanks to
2788 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2789 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2790 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2791 * bug fix: The compiler now checks for duplicated variables in macro
2793 * bug fix: SETQ on globals returns the correct value.
2794 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2795 not a COMPILER-ERROR (followed by some other strange error on
2796 choosing the CONTINUE restart).
2797 * bug fix: make.sh and friends are now more consistent in the way that
2798 they look for GNU "make".
2800 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2801 * experimental native threads support (on x86 Linux >=2.4 only).
2802 This is not compiled in by default: you need to add :SB-THREAD to
2803 the target features. See the "Beyond ANSI" chapter of the manual
2805 * fix for longstanding nonANSIism: The old distinction between
2806 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2807 The return value from CL:FIND-CLASS is now a CLOS class, and
2808 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2809 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2810 * An interface to the MetaObject Protocol, as described in Kiczales,
2811 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2812 MIT Press, 1991, is available from the SB-MOP package.
2813 * incompatible change: the SB-PCL package should now be considered
2814 a private implementation detail, and no longer a semi-private MOP
2816 * minor incompatible change: due to rearrangement for threads, the
2817 control stack and binding stack are now allocated at arbitrary
2818 addresses instead of being hardcoded per-port. Users affected by
2819 this probably have to be doing advanced things with shared
2820 libraries, and will know who they are.
2821 * minor incompatible change: Previously, all --eval forms used were
2822 processed with READ before any of them were processed with EVAL.
2823 Now each --eval form is processed with both READ and EVAL before
2824 the next --eval form is processed. (Thus package operations like
2825 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2826 work as the user might reasonably expect.)
2827 * minor incompatible change: *STANDARD-INPUT* is now only an
2828 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2830 * minor incompatible change: Y-OR-N-P is now character-oriented,
2831 not line oriented. Also, YES-OR-NO-P now works without errors.
2832 (thanks to Antonio Martinez)
2833 * sb-aclrepl module improvements: an integrated inspector, added
2834 repl features, and a bug fix to :trace command.
2835 * Known functions, which cannot be open coded by the backend, are
2836 considered to be able to check types of their arguments. (fixing
2837 a bug report by Nathan J. Froyd)
2838 * fixed a bug in computing method discriminating functions: It is
2839 now possible to define methods specialized on classes which have
2840 forward-referenced superclasses. (thanks to Gerd Moellmann)
2841 * fixed evaluation order in optional entries (reported by Gilbert
2843 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2844 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2845 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2846 required-by-AMOP NAME argument, as well as CLASS and
2847 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2848 * fixed bug 20: DEFMETHOD can define methods using names that are
2849 not the proper names of classes to designate class specializers.
2850 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2851 Unix file descriptors, instead of blowing up. (thanks to Antonio
2853 * Garbage collection refactoring: user-visible change is that a
2854 call to the GC function during WITHOUT-GCING will not do garbage
2855 collection until the end of the WITHOUT-GCING. If you were doing
2856 this you were probably losing anyway.
2857 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2858 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2859 TYPEP the latter but not the former.
2860 * The compiler issues a full WARNING for calls to undefined functions
2861 with names from the CL package.
2862 * MAP-INTO for a vector destination is open coded. (reported by
2863 Brian Downing on c.l.l)
2864 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2865 documentation string.
2866 * fixed some bugs revealed by Paul Dietz' test suite:
2867 ** COPY-ALIST now signals an error if its argument is a dotted
2869 ** Condition slots are now accessed more correctly in the presence
2870 of multiple initargs for a given slot.
2871 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2872 exclude restarts of the same name associated with a different
2874 ** DEFCLASS of forward-referenced classes with another
2875 forward-referenced class in the superclasses list no longer
2877 ** Condition slots are now initialized once each, not multiple
2878 times. (thanks to Gerd Moellmann)
2879 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2881 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2882 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2883 arguments contain duplicated elements.
2884 ** RESTART-CASE understands local macros.
2885 ** RESTART-CASE associates exactly its own restarts with a condition.
2886 ** ENDP in safe mode checks its argument to be of type LIST.
2887 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2888 in question is unbound.
2889 ** Optimized MAKE-INSTANCE functions no longer cause internal
2890 assertion failures in the presence of duplicate initargs.
2891 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2892 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2893 propagated even in the ctor optimized implementation of
2895 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2897 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2898 * a better implementation of SXHASH on (simple) bit vectors,
2899 measured both in execution speed and in distribution of results
2900 over the positive fixnums, has been installed. Likewise, a better
2901 implementation of EQUAL for simple bit vectors is now available.
2902 * fixed CEILING optimization for a divisor of form 2^k.
2903 * fixed bug 240 (emitting extra style warnings "using the lexical
2904 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2905 by Antonio Martinez)
2906 * fixed SXHASH, giving different results for NIL depending on type
2907 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2908 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2909 variable. (found by Rolf Wester)
2910 * SBCL does not ignore type declarations for special
2911 variables. (reported by rif on c.l.l 2003-03-05)
2912 * some bug fixes in contrib/sb-aclrepl/
2913 * fixed some bugs revealed by Paul Dietz' test suite:
2914 ** a bug in the CONS type specifier, whereby the CAR and CDR
2915 types got intertwined, has been fixed;
2916 ** the type system is now able to reason about the interaction
2917 between INTEGER and RATIO types more completely;
2918 ** APPEND, [N]REVERSE and NRECONC check that those their
2919 arguments, which must be proper lists, are really so;
2920 ** An array specialized to be unable to hold elements has been
2921 implemented, as required -- yes, really -- by ANSI;
2922 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2923 on malformed property lists;
2925 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2926 * incompatible packaging change: in line with Unix convention,
2927 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2928 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2929 * REQUIRE and PROVIDE are now optionally capable of doing something
2930 useful. See the documentation string for REQUIRE.
2931 * infrastructure for a managed SBCL contrib system: contributed
2932 modules in this release include:
2933 ** the ASDF system definition facility;
2934 ** an interface to the BSD Sockets API;
2935 ** an ACL-like convenience interface to the repl;
2936 (thanks to Kevin Rosenberg)
2937 ** an implementation of ROTATE-BYTE, with efficient implementation
2939 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2940 gives a closer approximation to the right answer than previously.
2941 (thanks to Raymond Toy)
2942 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2943 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2944 optional environment arguments, as required by ANSI.
2945 * fixed bugs in other functions taking environment objects, allowing
2946 calls with an explicit NIL environment argument to be compiled
2948 * fixed bug 228: primary return values from
2949 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2950 COMPILE or FUNCTION.
2951 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2952 :TYPE will no longer signal errors on innocuous objects.
2953 * fixed bug 231b: SETQ is better at respecting type declarations in
2954 the lexical environment.
2955 * fixed a bug in DEFCLASS: classes named by symbols with no or
2956 unprintable packages can now be defined.
2957 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2958 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2959 * fixed compiler failure related to checking types of functions.
2960 (reported by Robert E. Brown)
2961 * the compiler is now much more consistent in its error-checking
2962 treatment of bounding index arguments to sequence functions: in
2963 (SAFETY 3) code, errors will be signalled in almost all cases if
2964 invalid sequence bounding indices are passed to functions defined
2965 by ANSI to operate on sequences.
2966 * fixed a bug in the build procedure: documentation of SBCL-specific
2967 packages is now preserved and available in the final Lisp image.
2968 * lifted FDEFINITION lookup out of loops in the implementation of
2969 many list operations. (thanks to Robert E. Brown)
2970 * fixed a bug in the reader: the #n# reader macro now works for
2971 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2972 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2973 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2974 E. Brown and Tony Martinez respectively)
2975 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2976 SLOT-BOUNDP: in particular, it is now able to optimize them much
2977 better, and is now not vulnerable to having packages renamed.
2978 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2979 always called when a slot is not present in an instance. (thanks
2981 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2982 longer signal an error on unreading a character following EOF on
2983 the previous constituent stream. (thanks to Tony Martinez)
2984 * fixed some bugs revealed by Paul Dietz' test suite:
2985 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2986 not just nonnegative fixnums;
2987 ** the logical bit-array operators such as BIT-AND now accept an
2988 explicit NIL for their "opt-arg" argument (to indicate a
2989 freshly-consed result bit-array);
2990 ** ELT now signals an error on an invalid sequence index in safe
2992 ** the type system is now cleverer about negations of numeric
2993 types, and consequently understands the BIGNUM and RATIO types
2995 ** the type system is now cleverer about the interaction between
2996 INTEGER and RATIO types: while bugs still remain, many more
2997 cases are accurately computed;
2998 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2999 if it is in the last clause;
3000 ** CONSTANTLY now correctly returns a side-effect-free function in
3002 ** DECLARE is no longer treated as a special-operator; in
3003 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3004 * incremented fasl file version number due to the change in the
3005 DEFSTRUCT-SLOT-DESCRIPTION structure.
3007 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3008 * minor incompatible change: code processed by the "interpreter" or
3009 EVAL now has a compilation optimization policy of (DEBUG 2)
3010 (changed from (DEBUG 1)) to improve debuggability of interactive
3011 development, and to allow the use of the debug RETURN command in
3013 * an experimental implementation of the RETURN command for the
3014 debugger has been included. (thanks to Frederik Kuivinen)
3015 * fixed bug 62: constraints were not propagated into a loop.
3016 * fixed bug in embedded calls of SORT (reported and investigated by
3018 * fixed some bugs revealed by Paul Dietz' test suite:
3019 ** printing and reading of arrays with some dimensions having
3020 length 0 (thanks to Gerd Moellmann);
3021 ** BOA constructor with &AUX argument without a default value does
3022 not cause a type error;
3023 ** CONSTANTP now returns true for all self-evaluating objects.
3025 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3026 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3027 accessors that are related by inheritance, as specified in the
3028 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3029 Valtteri Vuorikoski)
3030 * The compiler is now able to inline functions that were defined in
3031 a complex lexical environment (e.g. inside a MACROLET).
3032 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3034 * Support for the upcoming FreeBSD-5.0 release has been included.
3035 (thanks to Dag-Erling Smorgrav)
3036 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3037 effect when it is not in a toplevel context.
3038 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3039 stand a better chance of being correct. (thanks to Gerd
3041 * fixed bug in COERCE, which now signals an error on coercing a
3042 rational to a bounded real type which excludes the expected
3044 * The compiler is now able to derive types more accurately from the
3045 COERCE and COMPILE functions.
3046 * fixed bug 223: functional binding is considered to be constant
3047 only for symbols in the CL package.
3048 * fixed bug 231: SETQ did not check the type of a variable being set
3049 (reported by Robert E. Brown)
3050 * A new optimization for MAKE-INSTANCE has been included, fixing
3051 various bugs (including relating to :ALLOCATION :CLASS slots and
3052 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3054 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3055 ** As required by ANSI, LOOP now disallows anonymous collection
3056 clauses such as COLLECT I in conjunction with aggregate boolean
3057 clauses such as THEREIS (= I 1);
3058 ** LOOP now signals an error when any variable is reused in the
3059 same loop (including the potentially useful construct analogous
3060 to WITH A = 1 WITH A = (1+ A);
3061 ** IT is only a special loop symbol within the first clause of a
3062 conditional loop clause;
3063 ** LOOP with a typed iteration variable over a hashtable now
3064 signals a type error iff it should.
3065 * fixed some other bugs revealed by Paul Dietz' test suite:
3066 ** FILE-STREAM now names the class previously known as FD-STREAM;
3067 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3068 argument) no longer signals an error;
3069 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3070 of the slot symbol, rather than using the current package
3071 ((:CONC-NAME "") continues to intern the slot's name in the
3073 * incremented fasl file version number, because of the incompatible
3074 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3075 of the new implementation of DEFINE-COMPILER-MACRO.
3077 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3078 * Support for building SBCL for MIPS platforms running in
3079 little-endian mode has now been checked in, and basic
3080 functionality on said platforms verified.
3081 * minor incompatible change: PCL now records the pathname of a file
3082 in which methods and the like are defined, rather than its
3084 * minor incompatible change: TRUENAME now considers the truename of
3085 a file naming a directory to be the pathname with :DIRECTORY
3086 component indicating that directory.
3087 * minor incompatible change: a NAMED clause in the extended form of
3088 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3089 reason for the previous behaviour is unclear.
3090 * more systematization and improvement of CLOS and MOP conformance
3091 in PCL (thanks to Gerd Moellman and Pierre Mai):
3092 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3094 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3095 primary methods with no specializers;
3096 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3098 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3099 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3100 FORWARD-REFERENCED-CLASSes; error reporting on
3101 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3103 ** SXHASH on CLOS instances now uses a slot internal to the
3104 instance to return different numbers on distinct instances,
3105 while preserving the same return value through invocations of
3107 ** DEFMETHOD signals errors when methods with longer incongruent
3108 lambda lists are added to generic functions;
3109 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3110 CLASS, as specified in AMOP;
3111 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3112 on the order returned by the primary method for classes of
3113 class STANDARD-CLASS;
3114 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3115 * fixed some bugs shown by Paul Dietz' test suite:
3116 ** DOLIST puts its body in TAGBODY;
3117 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3119 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3120 value producing form;
3121 ** if more variables are given to PROGV than values, extra
3122 variables are bound and made to have no value;
3123 ** NSUBSTITUTE on list arguments gets the right answer with
3125 ** ELT signals an error of type TYPE-ERROR when the index argument
3126 is not a valid sequence index;
3127 ** LOOP signals (at macroexpansion time) an error of type
3128 PROGRAM-ERROR when duplicate variable names are found;
3129 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3130 ** FUNCALL of special-operators now cause an error of type
3132 ** PSETQ now works as required in the presence of side-effecting
3133 symbol-macro places;
3134 ** NCONC accepts any object as its last argument;
3135 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3137 ** loop-for-as-package does not require a package to be explicitely
3139 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3141 * fixed bug 166: compiler preserves "there is a way to go"
3142 invariant when deleting code.
3143 * fixed bug 172: macro lambda lists with required arguments after
3144 &REST arguments now cause an error to be signalled. (thanks to
3146 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3147 bug. (thanks to Matthew Danish)
3148 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3150 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3151 arguments are lexically rebound. (thanks to Gerd Moellmann and
3153 * fixed bug 194: error messages are now more informative when there
3154 is no primary method applicable in a call to a generic
3155 function. (thanks to Gerd Moellmann)
3156 * fixed bug in command line argument checking (thanks to Julian
3158 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3159 (thanks to Matthew Danish)
3160 * incremented fasl file version number, because of the
3161 SXHASH-related changes in the layout of CLOS data structures
3163 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3164 * minor incompatible change: The runtime (the Unix executable named
3165 "sbcl") is now much pickier about the .core files it will load.
3166 Essentially it now requires .core files to descend from the same
3167 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3168 as the runtime does. (The intent is to prevent the crashes which
3169 can occur, and which can even be reported as mysterious failures,
3170 when people patch the sources or change the build parameters
3171 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3172 sbcl and .core files.)
3173 * fixed bug: VALUES-LIST is no longer optimized away.
3174 * fixed bug 142: The FFI conversion of C string values to Lisp
3175 string values no longer conses excessively. (thanks to Nathan
3176 Froyd porting Raymond Toy's fix to CMU CL)
3177 * began to systematize and improve MOP conformance in PCL (thanks to
3178 Nathan Froyd, Gerd Moellman and Pierre Mai):
3179 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3181 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3182 ** FINALIZE-INHERITANCE is now called on class finalization;
3183 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3184 argument precedence order.
3185 * fixed bug 202: The compiler no longer fails on functions whose
3186 derived types contradict their declared type.
3187 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3188 so it can be non-toplevel.
3189 * The fasl file version number has changed (because of the new
3190 implementation of DEFMACRO).
3191 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3192 safe code, that any length requirement by their type-specifier
3193 argument is valid. The exceptions to this are described in bug
3195 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3196 functions now check the types of their inputs as required by ANSI.
3197 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3198 introduced symbol is DECLAREd to be SPECIAL.
3199 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3200 * fixed bug: :COUNT argument to sequence functions may be negative
3201 * fixed bug: body of DO-SYMBOLS may contain declarations
3202 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3203 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3204 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3205 symbol macro only once
3206 * fixed printing of call frame when argument list is unavailable
3207 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3208 * compiler no longer signals WARNING on unknown keyword
3211 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3212 * A beta-quality port to the mips architecture running Linux,
3213 based on the old CMUCL backend, has been made. It has been tested
3214 on a big-endian kernel, and works sufficiently well to be able to
3215 rebuild itself; it has not been tested in little-endian mode.
3216 * fixed an inconsistency between gencgc.c and purify.c which made
3217 dumping/loading .core files unreliable
3218 * fixed bug 120a: The compiler now deals correctly with IFs where
3219 the consequent is the same as the alternative, instead of
3220 misderiving the return type. (thanks to Alexey Dejneka)
3221 * fixed bug 113: Logical pathnames are now dumpable (the logical
3222 host is resolved at load-time, throwing an error if it is not
3224 * fixed bug 174: FORMAT's error message is slightly clearer when a
3225 non-printing character is used in a format directive.
3226 * fixed several bugs in compiler checking of type declarations, i.e.
3227 violations of the Python "declarations are assertions" principle
3228 (thanks to Alexey Dejneka)
3229 * fixed several bugs in PCL's error checking (thanks to Gerd
3231 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3233 * fixed bug in compilation of functions as first class values
3234 (thanks to Antonio Martinez-Shotton)
3235 * The compiler's handling TYPE-ERRORs which it can prove will
3236 inevitably happen at runtime has been cleaned up and corrected
3237 in several ways. (thanks to Alexey Dejneka)
3238 * improved argument type checking for various basic arithmetic
3239 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3240 lavished on them in the past that they can be compiled in many
3241 ways in different special cases
3242 * fixed bug 181: compiler checks validity of user supplied type
3244 * cleaned up code flushing in optimization: Function calls which
3245 should signal errors for safety purposes (e.g. which ANSI says
3246 should signal errors when their arguments are of incorrect type)
3247 are no longer optimized away.
3248 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3249 * incremented fasl file version number, because changes in the
3250 implementation of sequence functions like COERCE caused
3251 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3252 in old inline expansions) to become undefined. (Actually these
3253 changes were later undone, so we might very well be binary
3254 compatible with 0.7.7 after all, but leaving the version number
3255 incremented seemed like the simplest and most conservative
3258 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3259 * An alpha-quality port to the parisc architecture running Linux,
3260 based on the old CMUCL backend, has been made. This, even more so
3261 than the other backends, should be considered still a work in
3262 progress; known problems include that the Linux kernel in 64-bit
3263 mode does not propagate the correct sigcontext structure to
3264 userspace, and consequently SBCL on a parisc64 kernel will not
3266 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3267 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3268 suppressed.) Also now that the compiler is looking at declarations
3269 in the environment, it checks optimization declarations as well,
3270 and suppresses inlining when (> DEBUG SPEED).
3271 * More fixes have been made to treatment of floating point exception
3272 treatment and other Unix signals. In particular, floating point
3273 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3274 * The detection and handling of control stack exhaustion (infinite
3275 or very deeply nested recursion) has changed. Stack exhaustion
3276 detection is now done by write-protecting pages at the OS level
3277 and applies at all optimization settings; when found, a
3278 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3279 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3280 be caught using IGNORE-ERRORS.
3281 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3282 correctly when transforming recursive LABELS functions to LETs.
3283 (thanks to Alexey Dejneka)
3284 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3285 that are names of constants or global variables.
3286 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3287 alien routines with docstrings.
3288 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3289 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3291 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3292 (thanks to Eric Marsden)
3293 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3294 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3295 * Bug fix: The SPARC backend can now compile functions involving
3296 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3297 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3298 to a functions expecting a PATHNAME-DESIGNATOR.
3299 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3300 lambda lists. (thanks to Alexey Dejneka)
3301 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3302 functions so that EXPORTing the name of the class doesn't cause
3303 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3304 to Antonio Martinez for reporting this)
3305 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3306 constant-folded without failing an assertion. (thanks to Einar
3307 Floystad Dorum for reporting this)
3308 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3309 types are now dealt with more correctly by the compiler.
3310 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3311 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3312 behaviour analogous to RENAME-FILE. This puts its behaviour more
3313 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3315 * The fasl file version number has changed again. (because of the
3316 bug fix involving the names of PCL MAKE-INSTANCE functions)
3318 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3319 * bug fix: Floating point exceptions are treated much more
3320 consistently on the x86/Linux and PPC/Linux platforms.
3321 * Array initialization with :INITIAL-ELEMENT is now much faster for
3322 cases when the compiler cannot open code the array creation, but
3323 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3324 array accesses have also seen a speed increase.
3325 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3326 specified. (This was at the root of some bad interactions between
3327 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3329 * bug fix: Internal error arguments for undefined functions are now
3330 computed correctly on the PPC/Linux platform.
3331 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3332 Raymond Toy's patch for CMU CL)
3333 * Support for the Solaris 9 operating environment has been included
3334 (thanks to Daniel Merritt)
3335 * A very ugly but hopefully complete draft of the missing FFI chapter
3336 of the manual has been created by reformatting the corresponding
3337 CMU CL manual chapter into (currently very ugly and incoherent)
3338 DocBook and bringing it up to date for SBCL behavior. Thus, the
3339 manual is now essentially complete, at least by my extreme
3340 once-and-only-once standards, whereby it's acceptable to refer to
3341 the doc strings of SB-EXT functions as the primary documentation.
3342 * The fasl file version number has changed again, due to cleanup of
3343 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3344 is no longer a static symbol.)
3346 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3347 * SBCL now builds with OpenMCL (version 0.12) as the
3348 cross-compilation host; also, more progress has been made toward
3349 bootstrapping under CLISP.
3350 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3352 * bug 158 fixed: The compiler can now deal with integer loop
3353 increments different from 1; fixing this turned out also to fix
3355 * bug 169 fixed: no more bogus warnings about using lexical bindings
3356 despite the presence of perfectly good SPECIAL declarations (thanks
3357 to David Lichteblau)
3358 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3359 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3360 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3362 * bug 180 fixed: Method combination specifications no longer ignore
3363 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3364 * bug fix: Structure type predicate functions now check their argument
3365 count as they should.
3366 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3367 correctly. (thanks to Pierre Mai)
3368 * minor incompatible change: The --noprogrammer option is deprecated
3369 in favor of the new --disable-debugger option, which is very similar.
3370 (The major difference is that it takes effect at a slightly different
3371 time at startup, causing handling of errors in --sysinit and
3372 --userinit files will be affected differently.) The
3373 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3374 been added to allow this functionality to be controlled from ordinary
3375 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3376 maintainers, who might want to run non-interactive scripts to
3377 build SBCL cores which will later be used interactively.)
3378 * minor incompatible change: The LOAD function no longer, when given
3379 a wild pathname to load, loads all files matching that pathname.
3380 Instead, an error of type FILE-ERROR is signalled.
3382 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3383 * bug 147 fixed: The compiler preserves its block link/count
3384 invariants more correctly now so that it doesn't crash. (thanks
3386 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3388 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3389 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3390 no-value-for-this-character as NIL (as specified by ANSI).
3391 * HOST-NAMESTRING on physical pathnames now returns a string that is
3392 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3393 (thanks to Christophe Rhodes)
3394 * The Alpha port handles icache flushing more correctly. (thanks to
3396 * More progress has been made toward bootstrapping under CLISP. (thanks
3397 to Christophe Rhodes)
3398 * The fasl file format has changed again, because dynamic loading
3399 on OpenBSD (which has non-ELF object files) motivated some cleanups
3400 in the way that foreign symbols are transformed and passed around.
3401 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3402 is no longer treated as whitespace by the reader, but instead as
3403 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3404 |A
\7fB|, instead of A as it used to.
3406 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3407 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3408 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3409 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3410 this as of 0.7.1.45, but was left out of the previous news section
3411 (thanks to Dan Barlow)
3412 * SBCL now runs on the Solaris operating system on SPARC architectures
3413 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3414 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3415 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3416 * SPARC backend cleanups, allowing builds of cores optimized for V8
3417 and V9 SPARCS, and also emission of code targeted to a particular
3418 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3420 * SBCL is closer to bootstrapping under CLISP, thanks to various
3421 fixes by Christophe Rhodes.
3422 * The fasl file format has changed again, to allow the compiler's
3423 INFO database to support symbol macros.
3424 * The user manual (in doc/) is formatted into HTML more nicely.
3425 (thanks to coreythomas)
3426 * The system is smarter about SUBTYPEP relationships, especially
3427 those involving NOT types (including types such as ATOM which are
3428 represented internally using NOT types). Thus SUBTYPEP is less
3429 likely to return (VALUES NIL NIL) in general, and in particular
3430 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3432 * The fasl file format has changed again, because the internal
3433 representation of types now includes a new slot to support the new
3434 SUBTYPEP-of-NOT-types logic.
3435 * (not a change in the main branch of SBCL, but a related prototype
3436 which can hopefully be merged into the main branch of SBCL in the
3437 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3438 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3439 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3440 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3441 dispatch macro character combinations. (thanks to Alexey Dejneka)
3442 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3443 that superficially look like logical namestrings correctly.
3444 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3446 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3447 default supplied-p) for &optional and &key arguments. (thanks to
3450 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3451 * incompatible change: The compiler is now less aggressive about
3452 tail call optimization, doing it only when (> SPACE DEBUG) or
3453 (> SPEED DEBUG). (This is an incompatible change because there are
3454 programs which relied on the old CMU-CL-style behavior to optimize
3455 away their unbounded recursion which will now die of stack overflow.)
3456 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3457 for non-GENCGC systems has been increased to 20M (since that
3458 seems much closer to the likely performance optimum for modern
3459 systems than the old 4M value was)
3460 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3461 and *DEBUG-PRINT-LEVEL*
3462 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3463 of CMU CL's support for SPARC, and various endianness and other
3464 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3465 * new syntactic sugar for the Unix command line: --load foo.bar is now
3466 an alternate notation for --eval '(load "foo.bar")'.
3468 ** The system now detects stack overflow and handles it gracefully,
3469 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3470 optimization settings. (This is a good thing in general, and
3471 its introduction in this version should be particularly timely
3472 for anyone whose code fails because of suppression of tail
3474 ** The system now hunts for the C variable "environ" in a more
3475 devious way, to avoid segfaults when the C library version differs
3476 between compile time and run time. (thanks to Christophe Rhodes)
3477 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3478 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3479 ** The compiler no longer issues bogus style warnings for undefined
3480 classes in the same source file as the DEFCLASSes which defined
3481 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3483 ** fixes in CONDITION class precedence list for undefined function
3484 errors (thanks to Alexei Dejneka)
3485 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3486 correctly. (thanks to Dan Barlow)
3487 ** portability fixes aiming at bootstrapping under CLISP (thanks
3488 to Dave McDonald and Christophe Rhodes)
3489 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3490 ** fixes in type translation and and type inference (thanks to
3492 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3493 ** various fixes in the new ports (thanks to Dan Barlow)
3494 * several changes related to debugging:
3495 ** suppression of tail recursion, as noted above
3496 ** stack overflow detection, as noted above
3497 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3498 is now the default. (For some time encapsulation has been more
3499 reliable than the breakpoint-based :ENCAPSULATE NIL
3500 implementation, at least on X86 systems; and I just noticed that
3501 encapsulation also seems closer to the spirit of the ANSI
3504 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3506 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3507 up properly again. (There was a packaging bug in 0.7.0 which
3508 left their definitions in SB-SYS::LOAD-FOREIGN and
3509 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3510 vital for most things which interface to C-level interfaces,
3511 like extensions working with sockets or databases or
3512 Perl-compatible regexes or whatever, and the need to fix
3513 this bug is the main reason that 0.7.1 was released so
3515 ** DEFGENERIC is now choosier about the methods it redefines, so that
3516 reLOADing a previously-LOADed file containing DEFGENERICs does
3517 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3518 cycle now works as it should. (thanks to Alexey Dejneka)
3519 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3520 fixed by Christophe Rhodes. (He actually submitted this patch
3521 months ago, and I delayed until after 0.7.0.)
3522 ** Bug 111 (internal compiler confusion about runtime checks on
3523 FUNCTION types) was fixed by Alexey Dejneka.
3524 * Some internal cleanups (getting rid of variables which aren't
3525 needed now that the byte interpreter is gone) caused the fasl
3526 file format number to change again.
3528 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3529 * major incompatible change: The default fasl file extension, i.e. the
3530 default extension for files produced by COMPILE-FILE, has changed
3531 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3533 ** There are many changes in the implementation of the compiler.
3534 SBCL is now essentially a compiler-only implementation of ANSI
3535 Common Lisp. EVAL still "interprets" a few special cases, but
3536 almost all the interesting cases are handled by creating
3537 a LAMBDA expression, calling COMPILE on it, then calling
3538 FUNCALL on the result.
3539 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3540 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3541 Since the code is newer, there might still be some new bugs
3542 (though not as many as before Martin Atzmueller's fixes:-). But
3543 the new code is substantially simpler and clearer, and hopefully
3544 any remaining bugs will be simpler, less fundamental, and more
3545 fixable then the bugs in the old code.
3546 ** The revised compiler is still a little unsteady on its feet.
3548 *** The debugging information it produces (particularly the names
3549 of FUNCTION objects) is sometimes much less useful than what
3550 the old compiler produced.
3551 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3552 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3553 within a MACROLET) has been temporarily weakened.
3554 ** There are new compiler optimizations for various functions:
3555 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3556 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3557 *** the math functions TRUNCATE, FLOOR, and CEILING
3558 *** the function-of-all-trades COERCE
3559 Mostly these should be transparent, but there's one
3560 potentially-annoying problem (bug 117): when the compiler
3561 inline-expands a function and does type analysis on the result,
3562 it can create control paths which have type mismatches, and
3563 when it can't prove that those control paths aren't taken,
3564 it will issue WARNINGs about the type mismatches. This is
3565 a particular problem in practice for the new sequence functions.
3566 It's not clear how this should be fixed, and for now, a
3567 workaround is given in the entry for 117 in the BUGS file.
3568 ** (Because of the interaction between the two previous items --
3569 occasional inlining problems and new inline expansions -- some
3570 of the new sequence function optimizations won't really kick in
3571 completely until debugging information, and then inlining, are
3572 straightened out in some future version.)
3573 * minor incompatible changes:
3574 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3575 DIRECTORY no longer implicitly promotes NIL slots of its
3576 pathname argument to :WILD. In particular, when you ask for the
3577 contents of a directory (which you used to be able to do without
3578 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3579 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3580 ** changes in behavior that ANSI explicitly defines to be
3581 implementation dependent:
3582 *** The new compiler-only implementation still conforms with ANSI,
3583 but acts a little different than before. Besides the obvious
3584 changes in performance tradeoffs (that the cost per form passed
3585 to EVAL has gone up, and the cost per form executed by EVAL
3586 has gone down), the behavior of the system changes a little
3587 because there are no longer any interpreted function objects.
3588 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3589 e.g. doing COMPILE on the output of interactive DEFUN is
3591 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3593 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3594 has changed from (:CL) to NIL.
3595 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3596 e.g. #\Newline and #\Tab, don't have names specified in the
3597 ANSI Common Lisp standard, is now based on their ASCII symbolic
3598 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3599 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3600 are no longer used for output.
3601 ** changes in internal implementation constants:
3602 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3603 4 million. (If your application spends a lot of time GCing and
3604 you have a lot of RAM, you might want to experiment with
3605 increasing it even more.)
3606 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3607 However, almost all old code should still continue to work without
3608 immediate update, as SB-C-CALL is now a (deprecated) nickname
3610 ** Old operator names in the style DEF-FOO are now deprecated in
3611 favor of new corresponding names DEFINE-FOO, for consistency with
3612 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3613 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3614 internal symbols, but a few supported extensions like
3615 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3616 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3617 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3618 etc. as you get deeper into recursive calls to the debugger
3619 command loop, instead of the old "5]", "5]]", "5]]]"
3620 sequence. (I was motivated to do this when squabbles between
3621 ILISP and SBCL left me very deeply nested in the debugger. In the
3622 short term, this change will probably provoke more ILISP/SBCL
3623 squabbles, but hopefully it will be an improvement in the long run.)
3624 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3625 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3626 function, since declaiming return types involving aliens is
3627 (1) annoyingly messy to do by hand and (2) vital to efficient
3628 compilation of code which calls such functions.
3629 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3630 longer reexported by the SB-EXT package. They're solely useful
3631 for alien code, so it seems more logical that you should get
3632 them from the SB-ALIEN package, not in SB-EXT.
3633 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3634 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3635 features. Instead, the code that they used to control is always
3636 built into the system.
3637 * many other bug fixes
3638 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3639 advantage of the new EVAL-WHEN stuff and to clean them up in
3640 general, and they are now more ANSI-compliant in a number of
3641 ways. Martin Atzmueller is responsible for a lot of this.
3642 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3644 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3645 *** correct ERROR type for various file operations
3646 *** some fixes for Lisp streams
3647 *** DEFMETHOD syntax checking
3648 *** changing old weird representation of debug information as
3649 strings (which, among their other deficiencies, don't transform
3650 correctly when you rename packages, and don't change their
3651 print representation when you change things like *PACKAGE*
3652 and *PRINT-LENGTH*) to symbols and lists of symbols
3653 He also made several improvements and fixed several bugs in DESCRIBE.
3654 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3656 *** misbehavior of WRITE-STRING/WRITE-LINE
3657 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3658 and several other LOOP problems as well
3659 *** DIRECTORY when similar filenames are present
3660 *** DEFGENERIC with :METHOD options
3661 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3662 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3663 *** argument ordering in FIND with :TEST option
3664 *** mishandled package designator argument in APROPOS-LIST
3665 *** various problems in the backquote readmacro
3666 *** a bug in APROPOS
3667 *** probably some others that I'm not describing very well here,
3668 since the CVS log documents them by reference to sbcl-devel
3669 messages, and the SourceForge archives aren't working well.:-(
3670 ** Dan Barlow improved the Alpha port (and is making progress on the
3671 PPC port, for those of you who think different).
3672 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3673 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3674 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3675 continues to work on the SPARC port (for those of you in a position
3676 to look down upon our little PC-compatible boxes from a great height).
3677 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3678 and uses that copy, rather than the current dynamic value, when
3679 it's trying to decide whether to truncate output. Thus e.g.
3680 (let ((*print-lines* 50))
3681 (pprint-logical-block (stream nil)
3683 (let ((*print-lines* 8))
3684 (print (aref possiblybigthings i) stream)))))
3685 should now truncate the logical block only at 50 lines, instead of
3686 often truncating it at 8 lines, as it did before.
3687 * The doc/cmucl/ directory, containing old CMU CL documentation
3688 from the time of the fork, is no longer part of the base system.
3689 SourceForge has shut down its anonymous FTP service, and with it
3690 my original plan for distributing the old CMU CL documentation
3691 there. For now, if you need these files you can download an old
3692 SBCL source release and extract them from it.
3693 * The fasl file version number changed again, for dozens of reasons,
3694 some of which are apparent above.
3696 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3697 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3698 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3699 patch, so that DISASSEMBLE output is much nicer.
3700 * The code in the SB-PROFILE package now seems reasonably stable.
3701 I still haven't decided what the final interface should look like
3702 (I'd like PROFILE to interact cleanly with TRACE, since both
3703 facilities use function encapsulation) but if you have a need
3704 for profiling now, you can probably use it successfully with
3705 the current CMU-CL-style interface.
3706 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3707 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3708 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3709 link now returns the dangling link itself, and for similar
3710 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3711 link itself. (In these cases the old code signalled an error and
3712 looped endlessly, respectively.) Thus, DIRECTORY now works even
3713 in the presence of dangling and cyclic symbolic links.
3714 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3715 is now a supported extension again, since the consensus on
3716 sbcl-devel was that it can be useful for ordinary development
3717 work, not just for debugging SBCL itself.
3718 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3719 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3720 that currently-defined functions might be redefined later with
3721 different return types.
3722 * Hash tables can be printed readably, as inspired by CMU CL code
3723 of Eric Marsden and SBCL code of Martin Atzmueller.
3724 * better error handling in CLOS method combination, thanks to
3725 Martin Atzmueller porting Pierre Mai's CMU CL patches
3726 * more overflow fixes for >16Mbyte I/O buffers
3727 * A bug in READ has been fixed, so that now a single Ctrl-D
3728 character suffices to cause end-of-file on character streams.
3729 In particular, now you only need one Ctrl-D at the command
3730 line (not two) to exit SBCL.
3731 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3733 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3734 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3735 does the right thing.
3736 * The implementation of some type tests, especially for CONDITION
3737 types, is now tidier and maybe faster, due to CMU CL code
3738 originally by Douglas Crosher, ported by Martin Atzmueller.
3739 * Some math functions have been fixed, and there are new
3740 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3741 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3742 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3743 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3744 likely to be built into sbcl-0.7.0.)
3745 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3746 CL, ported to SBCL by Martin Atzmueller)
3747 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3748 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3749 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3750 * There's a new slam.sh hack to shorten the edit/compile/debug
3751 cycle for low-level changes to SBCL itself, and a new
3752 :SB-AFTER-XC-CORE target feature to control the generation of
3753 the after-xc.core file needed by slam.sh.
3754 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3755 COMPILE-FILE is no longer supported, so that now every function
3756 gets an entry point, so that block compilation looks a little
3757 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3758 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3759 returns the number of bytes consed since the system started,
3760 rather than the number consed since the first time the function
3761 was called. (The new definition parallels ANSI functions like
3762 CL:GET-INTERNAL-RUN-TIME.)
3763 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3764 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3765 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3766 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3767 * Fasl file version numbers are now independent of the target CPU,
3768 since historically most system changes which required version
3769 number changes have affected all CPUs equally. Similarly,
3770 the byte fasl file version is now equal to the ordinary
3773 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3774 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3775 is no longer recognized. I apologize for this, because it was
3776 listed in SB-EXT as a supported extension, but I found that
3777 its existing behavior was poorly specified, as well as incorrectly
3778 specified, and it looked like too much of a mess to straighten it
3779 out. I have enough on my hands trying to get ANSI stuff to work..
3780 * many patches ported from CMU CL by Martin Atzmueller, with
3781 half a dozen bug fixes in pretty-printing and the debugger, and
3782 half a dozen others elsewhere
3783 * fixed bug 13: Floating point infinities are now supported again.
3784 They might still be a little bit flaky, but thanks to bug reports
3785 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3786 as flaky as they were.
3787 * The --noprogrammer command line option is now supported. (Its
3788 behavior is slightly different in detail from what the old man
3789 page claimed it would do, but it's still appropriate under the
3790 same circumstances that the man page talks about.)
3791 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3792 are now supported, and enabled by default. Thus, the compiler can
3793 handle many floating point and complex operations much less
3794 inefficiently. (Thus e.g. you can implement a complex FFT
3796 * The compiler now detects type mismatches between DECLAIM FTYPE
3797 and DEFUN better, and implements CHECK-TYPE more correctly, and
3798 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3799 thanks to patches from Martin Atzmueller.
3800 * various fixes to make the cross-compiler more portable to
3801 ANSI-conforming-but-different cross-compilation hosts (notably
3802 Lispworks for Windows, following bug reports from Arthur Lemmens)
3803 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3804 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3805 fixed thanks to Pierre Mai's CMU CL patches.
3806 * fixes to keep the system from overflowing internal counters when
3807 it tries to use i/o buffers larger than 16M bytes
3808 * fixed bug 45a: Various internal functions required to support
3809 complex special functions have been merged from CMU CL sources.
3810 (When I was first setting up SBCL, I misunderstood a compile-time
3811 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3812 * improved support for type intersection and union, fixing bug 12
3813 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3814 more obscure bugs as well
3815 * some steps toward byte-compiling non-performance-critical
3816 parts of the system, courtesy of patches from Martin Atzmueller
3817 * Christophe Rhodes has made some debian packages of sbcl at
3818 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3819 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3820 stable, but are nonetheless usable. When he's ready, I'd be happy
3821 to add them to the SourceForge "File Releases" section. (And if
3822 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3823 * new fasl file format version number (because of changes in
3824 internal representation of (OR ..) types to accommodate the new
3825 support for (AND ..) types, among other things)
3827 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3828 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3830 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3832 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3833 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3834 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3835 * DESCRIBE now works on structure objects again.
3836 * Most function call argument type mismatches are now handled as
3837 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3838 know whether the function will be redefined before the call is
3839 executed. (The compiler could flag local calls with full WARNINGs,
3840 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3841 it doesn't keep track of enough information to know whether calls
3842 are local in this sense.)
3843 * Compiler output is now more verbose, with messages truncated
3844 later than before. (There should be some supported way for users
3845 to override the default verbosity, but I haven't decided how to
3846 provide it yet, so this behavior is still controlled by the internal
3847 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3848 src/compiler/ir1util.lisp.)
3849 * Fasl file format version numbers have increased again, because
3850 support for the Gray streams extension changes the layout of the
3851 system's STREAM objects.
3852 * The Gray subclassable streams extension now works, thanks to a
3853 patch from Martin Atzmueller.
3854 * The full LOAD-FOREIGN extension (not just the primitive
3855 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3856 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3857 but like most other programs, it defaults to copying the Unix
3858 environment from the original process instead of starting the
3859 new process in an empty environment.
3860 * Extensions which manipulate the Unix environment now support
3861 an :ENVIRONMENT keyword option which doesn't smash case or
3862 do other bad things. The CMU-CL-style :ENV option is retained
3863 for porting convenience.
3864 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3865 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3867 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3869 * A patch from Martin Atzmueller seems to have solved the SIGINT
3870 problem, and as far as we know, signal-handling now works cleanly.
3871 (If you find any new bugs, please report them!)
3872 * The system no longer defaults Lisp source file names to types
3873 ".l", ".cl", or ".lsp", but only to ".lisp".
3874 * The compiler no longer uses special default file extensions for
3875 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3876 seems to expect a single default extension for all compiled code,
3877 and there's no compelling reason to try to stretch the standard
3878 to allow two different extensions.) Instead, byte-compiled files
3879 default to the same extension as native-compiled files.
3880 * Fasl file format version numbers have increased again, because
3881 a rearrangement of internal implementation packages made some
3882 dumped symbols in old fasl files unreadable in new cores.
3883 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3884 many fewer weird special cases.
3885 * Bug #17 (differing COMPILE-FILE behavior between logical and
3886 physical pathnames) has been fixed, and some related misbehavior too,
3887 thanks to a patch from Martin Atzmueller.
3888 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3889 by Tim Moore, ported to SBCL by Martin Atzmueller.
3890 * Martin Atzmueller fixed several filesystem-related problems,
3891 including bug #36, in part by porting CMU CL patches, which were
3892 written in part by Paul Werkowski.
3893 * More compiler warnings in src/runtime/ are gone, thanks to
3894 more patches from Martin Atzmueller.
3895 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3898 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3900 * DESCRIBE now works on CONDITION objects.
3901 * The debugger now handles errors which arise when trying to print
3902 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3904 * The build system now uses an additional file, customize-target-features.lisp,
3905 to allow local modifications to the target *FEATURES* list. (The point of
3906 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3907 debugging features enabled, without having to worry about propagating it
3908 into everyone's system when I do a "cvs update".) When no
3909 customize-target-features.lisp file exists, the target *FEATURES* list
3910 should be constructed the same way as before.
3911 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3912 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3913 warn about behavior which is undefined under the ANSI spec. Note: This
3914 is specified by ANSI, but it's not very popular with programmers.
3915 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3916 macro in the SBCL sources for an example of a workaround which you
3917 might use to make portable ANSI-standard code which does what you want.
3918 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3919 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3920 when it's used as a non-toplevel form. (This is required in order
3921 to implement the DEFCONSTANT-EQX macro.)
3922 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3923 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3924 funny things, due to bugs in the implementation of EVAL-WHEN
3925 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3926 non-toplevel DEFCONSTANTs are uncommon.)
3927 * The core file version number and fasl file version number have been
3928 incremented, because the old noncompliant DEFCONSTANT behavior involved
3929 calling functions which no longer exist, and because I also took the
3930 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3931 * fixed bug 1 (error handling before read-eval-print loop starts), and
3932 redid debugger restarts and related debugger commands somewhat while
3934 ** The QUIT debugger command is gone, since it did something
3935 rather different than the SB-EXT:QUIT command, and since it never
3936 worked properly outside the main toplevel read/eval/print loop.
3937 Invoking the new TOPLEVEL restart provides the same functionality.
3938 ** The GO debugger command is also gone, since you can just invoke
3939 the CONTINUE restart directly instead.
3940 ** The TOP debugger command is also gone, since it's redundant with the
3941 FRAME 0 command, and since it interfered with abbreviations for the
3943 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3945 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3946 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3947 some porting work by Martin Atzmueller)
3948 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3949 with a boatload of patches which he ported to SBCL. Now that those
3951 ** The system tries to make sure that its low-priority messages
3952 are prefixed by semicolons, to help people who like to use
3953 syntax highlighting in their ILISP buffer. (This patch
3954 was originally due to Raymond Toy.)
3955 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3956 patches originally written by Raymond Toy.
3957 ** The compiler understands coercion between single-value and
3958 multiple-VALUES type expressions better, getting rid of some very
3959 weird behavior, thanks to patches originally by Robert MacLachlan
3960 and Douglas Crosher.
3961 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3962 lambda lists, thanks to a patch originally by Pierre Mai.
3963 ** The system no longer bogusly warns about "abbreviated type
3965 ** The compiler gets less confused by inlining and RETURN-FROM,
3966 thanks to some patches originally by Tim Moore.
3967 ** The system no longer hangs when dumping circular lists to fasl
3968 files, thanks to a patch originally from Douglas Crosher.
3969 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3970 undefined function error.
3971 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3972 in structure slot initforms) for now, documented workaround instead:-|
3973 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3974 * fixed bug 5: added stubs for various Gray stream functions called
3975 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3976 installed, at least appropriate type errors are generated
3977 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3978 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3979 consistently in DEFMETHOD forms.
3980 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3981 it doesn't seem to affect SBCL after all
3982 * The C runtime system now builds with better optimization and many
3983 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3985 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3987 * The system is now under CVS at SourceForge (instead of the
3988 CVS repository on my home machine).
3989 * The new signal handling code has been tweaked to treat register
3990 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3991 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3992 downstream, e.g. in debug-int.lisp, has implicit dependencies
3993 on the unsignedness of integer representation of machine words,
3994 and that caused the system to bomb out with infinite regress
3995 when trying to recover from type errors involving signed values,
3996 e.g. (BUTLAST '(1 2 3) -1).)
3997 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3998 (This was one of the bugs Peter Van Eynde reported back in July.)
3999 * The system now uses code inspired by Colin Walters' O(N)
4000 implementation of MAP (from the cmucl-imp@cons.org mailing
4001 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4002 inline the MAP operation, and there is more than one
4003 sequence argument to the MAP call (so that it can't just
4004 do ETYPECASE once and for all based on the type of the
4005 single sequence argument). (The old non-inline implementation
4006 of the general M-argument sequence-of-length-N case required
4007 O(M*N*N) time when any of the sequence arguments were LISTs.)
4008 * The QUIT :UNIX-CODE keyword argument has been renamed to
4009 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4010 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4012 * Raymond Wiker's patches to port dynamic loading from Linux to
4013 FreeBSD have been added.
4014 * The BUGS file is now more nearly up to date, thanks in large part
4015 to Martin Atzmueller's review of it.
4016 * The debugger now flushes standard output streams before it begins
4017 its output ("debugger invoked" and so forth).
4018 * The core version number and fasl file version number have both
4019 been incremented, because of incompatible changes in the layout
4021 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4022 used to not be called for a saved Lisp image.)
4023 * Martin Atzmueller's version of a patch to fix a compiler crash,
4024 as posted on sbcl-devel 13 September 2000, has been installed.
4025 * Instead of installing Martin Atzmueller's patch for the
4026 compiler transform for SUBSEQ, I deleted the compiler transform,
4027 and transforms for some similar consing operations.
4028 * A bug in signal handling which kept TRACE from working on OpenBSD
4030 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4031 away by constant folding
4032 * The system now defines its address space constants in one place
4033 (in the Lisp sources), and propagates them automatically elsewhere
4034 (through GENESIS and the sbcl.h file). Therefore, patching the
4035 address map is less unnecessarily tedious and error-prone. The
4036 Lisp names of address space constants have also been systematized.
4037 * CVS tags like dollar-Header-dollar have been removed from
4038 the sources, because they have never saved me trouble and
4039 they've been source of trouble working with patches and other
4040 diff-related operations.
4041 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4042 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4044 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4046 * The system has been ported to OpenBSD.
4047 * The system now compiles with a simple "sh make.sh" on the systems
4048 that it's supported on. I.e., now you no longer need to tweak
4049 text in the target-features.lisp-expr and symlinks in src/runtime/
4050 by hand, the make.sh takes care of it for you.
4051 * The system is no longer so grossly inefficient when compiling code
4052 involving vectors implemented as general (not simple) vectors (VECTOR T),
4053 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4054 which dares to use the various sequence functions on non-simple
4055 vectors, takes less of a performance hit.
4056 * There is now a primitive type predicate VECTOR-T-P
4057 to test for the (VECTOR T) type, so that e.g.
4058 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4059 can now be compiled with some semblance of efficiency. (The old code
4060 turned the type declaration into a full call to %TYPEP at runtime!)
4061 * AREF on (VECTOR T) is still not fast, since it's still compiled
4062 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4063 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4064 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4065 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4066 without making full calls to SUBTYPEP at runtime.
4067 (Some analogous efficiency issues for non-simple vectors specialized to
4068 element types other than T, or for non-simple multidimensional arrays,
4069 have not been addressed. They could almost certainly be handled the
4070 same way if anyone is motivated to do so.)
4071 * The changes in array handling break binary compatibility, so
4072 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4073 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4074 instead of (VALUES T T).
4075 * By following the instructions that Dan Barlow posted to sbcl-devel
4076 on 2 July 2000, I was able to enable primitive dynamic object
4077 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4078 functionality is not implemented (since it calls ld to resolve
4079 library references automatically, requiring RUN-PROGRAM for its
4080 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4081 try to resolve library references) is now supported.
4082 * The system now flushes the standard output streams when it terminates,
4083 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4084 them at several other probably-convenient times, e.g. in each pass of
4085 the toplevel read-eval-print loop, and after evaluating a form given
4086 as an "--eval" command-line option. (These changes were motivated by a
4087 discussion of stream flushing issues on cmucl-imp in August 2000.)
4088 * The source transform for TYPEP of array types no longer assumes
4089 that an array whose element type is a not-yet-defined type
4090 is implemented as an array of T, but instead punts, so that the
4091 type will be interpreted at runtime.
4092 * There is now some support for cross-compiling in make.sh: each of
4093 the phases of make.sh has its own script. (This should be transparent
4094 to people doing ordinary, non-cross-compile builds.)
4095 * Since my laptop doesn't have hundreds of megabytes of memory like
4096 my desktop machine, I became more motivated to do some items on
4097 my to-do list in order to reduce the size of the system a little:
4098 ** Arrange for various needed-only-at-cold-init things to be
4099 uninterned after cold init. To support this, those things have
4100 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4101 symbols with such names are now uninterned after cold init).
4102 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4103 fasl files for cold load.
4104 ** Remove the old compiler structure pooling code (which used to
4105 be conditional on the target feature :SB-ALLOC) completely.
4106 ** Redo the representation of some data in cold init to be more compact.
4107 (I also looked into supporting byte compiled code at bootstrap time,
4108 which would probably reduce the size of the system a lot, but that
4109 looked too complicated, so I punted for now.)
4110 * The maximum signal nesting depth in the src/runtime/ support code has
4111 been reduced from 4096 to 256. (I don't know any reason for the very
4112 large old value. If the new smaller value turns out to break something,
4113 I'll probably just bump it back up.)
4114 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4116 * Many, many bugs reported by Peter Van Eynde have been added to
4117 the BUGS list; some have even been fixed.
4118 * While enabling dynamic object file loading, I tried to make the
4119 code easier to understand, renaming various functions and variables
4120 with less ambiguous names, and changing some function calling
4121 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4122 * While trying to figure out how to do the OpenBSD port, I tried to
4123 clean up some of the code in src/runtime/. In particular, I dropped
4124 support for non-POSIX signal handling, added various comments,
4125 tweaked the code to reduce the number of compilation warnings, and
4126 renamed some files to increase consistency.
4127 * To support the new automatic configuration functionality in make.sh,
4128 the source file target-features.lisp-expr has been replaced with the
4129 source file base-target-features.lisp-expr and the machine-generated
4130 file local-target-features.lisp-expr.
4131 * fixed a stupid quoting error in make.sh so that using CMU CL
4132 "lisp -batch" as cross-compilation host works again
4134 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4136 * DESCRIBE no longer tries to call itself recursively to describe
4137 bound/fbound values, so that it no longer fails on symbols which are
4138 bound to themselves (like keywords, T, and NIL).
4139 * DESCRIBE now works on generic functions.
4140 * The printer now prints less-screwed-up representations of closures
4141 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4142 * A private symbol is used instead of the :EMPTY keyword previously
4143 used to mark empty slots in hash tables. Thus
4144 (DEFVAR *HT* (MAKE-HASH-TABLE))
4145 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4146 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4147 now does what ANSI says that it should. (You can still get
4148 similar noncompliant behavior if bang on the hash table
4149 implementation with all the symbols you get back from
4150 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4151 This breaks binary compatibility, since tests for equality to
4152 :EMPTY are wired into things like the macroexpansion of
4153 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4155 * There's now a minimal placeholder implementation for CL:STEP,
4156 as required by ANSI.
4157 * An obscure bug in the interaction of the normal compiler, the byte
4158 compiler, inlining, and structure predicates has been patched
4159 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4160 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4162 * Missing ordinary arguments in a macro call are now detected even
4163 when the macro lambda list contains &KEY or &REST.
4164 * The debugger no longer complains about encountering the top of the
4165 stack when you type "FRAME 0" to explicitly instruct it to go to
4166 the top of the stack. And it now prints the frame you request even
4167 if it's the current frame (instead of saying "You are here.").
4168 * As specified by ANSI, the system now always prints keywords
4169 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4171 * The default initial SIZE of HASH-TABLEs is now smaller.
4172 * Type information from CLOS class dispatch is now propagated
4173 into DEFMETHOD bodies, so that e.g.
4174 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4176 is now basically equivalent to
4177 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4178 (DECLARE (TYPE SINGLE-FLOAT X))
4180 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4181 operation, without having to do run-time type dispatch.
4182 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4183 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4184 * Since I don't seem to be making big file reorganizations very often
4185 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4186 is overflowing my ability to conveniently back them up), I've finally
4187 checked the system into CVS. (The CVS repository is on my home system,
4188 not at SourceForge -- putting it on SourceForge might come later.)
4189 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4190 high-level GC-NOTIFY-FOO functions send their output. (There's
4191 still very little control of where low-level verbose GC functions
4192 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4193 less than it used to -- the GC-NOTIFY-FOO functions are now under
4194 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4195 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4196 in only one place in the source code, and propagates it automatically
4197 everywhere that it's needed. Thus e.g. when I bump the version from
4198 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4199 * The C source files now include boilerplate legalese and documentation
4200 at the head of each file (just as the Lisp source files already did).
4201 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4202 to his page will be replaced with a link to his new CLiki service.
4204 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4206 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4207 * The build process now looks for GNU make under the default name "gmake",
4208 instead of "make" as it used to. If GNU make is not available as "gmake"
4209 on your system, you can change this default behavior by setting the
4210 GNUMAKE environment variable.
4211 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4212 can build without error under CMU CL.
4214 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4216 * There is now a partial SBCL user manual (with some new text and some
4217 text cribbed from the CMU CL manual).
4218 * The beginnings of a profiler have been added (starting with the
4219 CMU CL profiler and simplifying and cleaning up). Eventually the
4220 main interface should be through the TRACE macro, but for now,
4221 it's still accessed through vaguely CMU-CL-style functions and macros
4222 exported from the package SB-PROFILE.
4223 * Some problems left over from porting CMU CL to the new
4224 cross-compilation bootstrap process have been cleaned up:
4225 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4226 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4227 part when getting rid of the compiler *BACKEND* stuff.)
4228 ** The value of *NULL-TYPE* was screwed up, because it was
4229 being initialized before the type system knew the final
4230 definition of the 'NULL type. This screwed up several key
4231 optimizations in the compiler, causing inefficiency in all sorts
4232 of places. (I found it because I wanted to understand why
4233 GET-INTERNAL-RUN-TIME was consing.)
4234 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4235 PROCLAIM FTYPE information. Unfortunately this broke binary
4236 compatibility again, since now the forms output by DEFGENERIC
4237 to refer to functions which didn't exist in 0.6.3.
4238 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4239 can use the new (as of 0.6.3) transform for SOME into MAP into
4241 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4242 values to (INTEGER 0 1000000), so that the time code will no longer
4243 occasionally get blown up by Linux returning 1000000 microseconds
4244 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4245 its output conform to the ANSI spec. (Alas, this makes its output
4246 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4247 * A full call to MAP NIL with a single sequence argument no longer conses.
4248 * fixes to problems pointed out by Martin Atzmueller:
4249 * The manual page no longer talks about multiprocessing as though
4250 it were currently supported.
4251 * The ILISP support patches have been removed from the distribution,
4252 because as of version 5.10.1, ILISP now supports SBCL without us
4253 having to maintain patches.
4254 * added a modified version of Raymond Toy's recent CMU CL patch for
4255 EQUALP comparison of HASH-TABLE
4257 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4259 * The system still can't cross-compile itself with
4260 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4261 entails), but at least it can compile more of itself that way
4262 than it used to be able to, and various buglets which were uncovered
4263 by trying to cross-compile itself that way have now been fixed.
4264 * This release breaks binary compatibility again. This time
4265 at least I've incremented the FASL file format version to 2, so that the
4266 problem can be detected reliably instead of just causing weird errors.
4267 * various new style warnings:
4268 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4269 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4270 ** using the lexical binding of a variable named in the *FOO* style
4271 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4272 as specified by ANSI.
4273 * *RANDOM-STATE* is no longer automatically initialized from
4274 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4275 default behavior of the system is to repeat its behavior every time
4276 it's run. If you'd like to change this behavior, you can always
4277 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4278 old convention there was no comparably easy way to get the system to
4279 repeat its behavior every time it was run.
4280 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4281 FTYPE declarations has been removed, in favor of their ANSI
4282 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4283 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4284 the types of their sequence arguments can be determined at compile time.
4285 This is done through a new open code expansion for MAP which eliminates
4286 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4287 argument types can be determined at compile time.
4288 * The optimizer now transforms COERCE into an identity operation when it
4289 can prove that the coerced object is already of the correct type. (This
4290 can be a win for machine generated code, including the output of other
4291 optimization transforms, such as the MAP transform above.)
4292 * Credit information has been moved from source file headers into CREDITS.
4293 * Source file headers have been made more standard.
4294 * The CASE macro now compiles without complaining even when it has
4297 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4299 * (Note that the way that the PCL macroexpansions were rewritten
4300 to accommodate the change in DEFGENERIC below breaks binary
4301 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4302 not run under sbcl-0.6.2. Once we get out of alpha releases,
4303 i.e. hit release 1.0.0, we'll probably try to maintain binary
4304 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4305 and sbcl-1.4.4. Until then, however, it might be fairly common
4306 for maintenance releases to break binary compatibility.)
4307 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4309 * The reporting of unhandled signals has been changed to print some
4310 explanatory text as well as the report form. (Previously only
4311 the report form was printed.)
4312 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4313 it defines, so that the compiler no longer issues undefined function
4314 warnings for compiled-but-not-yet-loaded generic functions.
4315 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4316 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4317 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4318 section of the ANSI spec are supported.
4319 * The "" nickname for the "KEYWORD" package has been removed.
4320 The reader still handles symbol tokens which begin with a package marker
4321 as keywords, but it doesn't expose its mechanism for doing so in the
4322 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4323 * The system now issues STYLE-WARNINGs for contradictory TYPE
4324 proclamations. (Warnings for contradictory FTYPE proclamations would
4325 be nice too, but those can't be done usefully unless the type system
4326 is made smarter about FUNCTION types.)
4327 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4328 systematized, so that "*target-*.lisp is supposed to exist only on the
4329 target and imply that there's a related file which exists on the
4330 host, and *host-*.lisp is supposed to exist only on the host and imply
4331 that there's a related file which exists on the target. This involves a
4332 lot of renaming. Hopefully the acute confusion caused by the renaming
4333 will be justified by the reduction in chronic confusion..
4334 ** runtime-type.lisp -> early-target-type.lisp
4335 ** target-type.lisp -> late-target-type.lisp
4336 ** early-host-format.lisp -> early-format.lisp
4337 ** late-host-format.lisp -> late-format.lisp
4338 ** host-error.lisp -> misc-error.lisp
4339 ** early-error.lisp -> early-target-error.lisp
4340 ** late-error.lisp -> late-target-error.lisp
4341 ** host-defboot.lisp -> early-defboot.lisp
4342 ** code/misc.lisp -> code/target-misc.lisp
4343 ** code/host-misc.lisp -> code/misc.lisp
4344 ** code/numbers.lisp -> code/target-numbers.lisp
4345 ** code/early-numbers.lisp -> numbers.lisp
4346 ** early-host-type.lisp -> early-type.lisp
4347 ** late-host-type.lisp -> late-type.lisp
4348 ** host-typep.lisp -> typep.lisp
4349 ** load.lisp -> target-load.lisp
4350 ** host-load.lisp -> load.lisp
4351 ** host-disassem.lisp -> disassem.lisp
4352 ** host-insts.lisp -> insts.lisp
4353 ** byte-comp.lisp -> target-byte-comp.lisp
4354 ** host-byte-comp.lisp -> byte-comp.lisp
4355 ** host-signal.lisp -> signal.lisp
4356 ** host-defstruct.lisp -> defstruct.lisp
4357 ** late-target-type.lisp -> deftypes-for-target.lisp
4358 Furthermore, several other previously target-only files foo.lisp (e.g.
4359 hash-table.lisp and random.lisp) have been split into a target-and-host
4360 foo.lisp file and a target-only target-foo.lisp file, with their key type
4361 definitions in the target-and-host part, so that the cross-compiler will
4362 know more about target types.
4363 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4364 gone away. In their place are various *BACKEND-FOO* variables
4365 corresponding to the slots of the old structure.
4366 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4367 into the target SBCL has been fixed.
4368 * The chill.lisp system for loading cold code into a running SBCL
4370 * Support for the CMU CL "scavenger hook" extension has been removed.
4371 (It was undocumented and unused in the CMU CL sources that SBCL was
4372 derived from, and stale in sbcl-0.6.1.)
4373 * Various errors in the cross-compiler type system were detected
4374 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4375 (enabling various consistency checks). Many of them were fixed,
4376 but some hard problems remain, so the compiler is back to
4377 running without *TYPE-SYSTEM-INITIALIZED* for now.
4378 * As part of the cross-compiler type system cleanup, I implemented
4379 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4380 * I have started adding UNCROSS calls throughout the type system
4381 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4382 on cross-compiler input files will be able to go away, and various
4384 * CONSTANTP now returns true for quoted forms (as explicitly required
4387 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4389 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4390 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4391 (MAKE-LIST -1). In the longer term, it ought to become true
4392 that declarations are assertions even at SAFETY 1. For now, it's not
4393 quite true even at SAFETY 3, but it's at least more nearly true..
4394 (Note that this change seems to increases the size of the system by
4395 O(5%) and to decrease the speed of the compiler by 20% or more.)
4396 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4397 for the problem of printing dozens of lines of distracting information
4398 about low-level system machinery as part of the top stack frame
4399 on entry to the debugger when an undefined function was called.
4400 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4401 is not reset to COMMON-LISP-USER.
4402 * Compilation of stuff related to dyncount.lisp has been made conditional
4403 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4404 smaller. The various dyncount-related symbols have been moved into
4405 a new "SB-DYNCOUNT" package.
4406 * tty-inspect.lisp has been renamed to inspect.lisp.
4407 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4408 feature has gone away. (When we eventually port to other flavors of
4409 libc and/or Unix, we'll try to make the differences between flavors
4410 invisible at the user level.)
4411 * Various other *FEATURES* tags, and/or their associated conditionals,
4412 have been removed if obsolescent, or given better documentation, or
4413 sometimes given more-mnemonic names.
4415 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4417 * tidied up "make.sh" script
4418 * tidied up system directory structure
4419 * better "clean.sh" behavior
4420 * added doc/FOR-CMUCL-DEVELOPERS
4421 * many many small tweaks to output format, e.g. removing possibly-confusing
4422 trailing #\. character in DESCRIBE-INSTANCE
4423 * (EQUALP #\A 'A) no longer signals an error.
4424 * new hashing code, including EQUALP hashing
4425 * tidied up Lisp initialization and toplevel
4426 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4427 * command line argument processing
4428 * added POSIX-GETENV function to deal with Unix-ish environment variables
4429 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4430 terminating SBCL on EOF
4431 * non-verbose GC by default
4432 * There is no more "sbcl" shell script; the sbcl file is now the C
4433 runtime executable (just like CMU CL).
4434 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4436 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4437 symbol and package information as Lisp native symbol and package objects
4438 instead of strings naming symbols and strings naming packages. This way,
4439 whenever packages are renamed (as in warm init), debug information is
4440 transformed along with everything else.
4441 * tweaked the optimization policy declarations which control the building
4442 of SBCL itself. Now, among other things, the system no longer saves
4443 source location debugging information. (This helps two problems at once
4444 by reducing SBCL size and by keeping SBCL from trying to look for its
4445 sources -- which may not exist -- when reporting errors.)
4446 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4447 debugging and testing purposes
4448 * cleaned up printing, making the printer call PRINT-OBJECT for
4449 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4450 methods, giving nearly-ANSI behavior
4451 * converted almost all special variables to use *FOO* naming convention
4452 * deleted PARSE-TIME functionality, since it can be done portably
4453 * moved some files out of cold init into warm init
4454 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4456 * regularized formatting of source files
4457 * added an install.sh script
4458 * fixed ridiculous memory usage of cross-compiler by making
4459 compiler/alloc.lisp not try to do pooling unless it can hook
4460 itself into the GC of the cross-compilation host. Now the system
4461 builds nicely on my old laptop.
4462 * added :SB-ALLOC in target-features.lisp-expr
4463 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4464 was not implemented)
4465 * re-did condition handling and note reporting in the compiler. Notes
4466 are no longer handled by signalling conditions. Style warnings
4467 and warnings are handled more correctly and reported in such a way
4468 that it's easy to find one or the other in your output (so that you
4469 can e.g. figure out which of many problems caused COMPILE-FILE to
4471 * changed the severity of several compiler warnings from full WARNING
4472 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4473 compiler note reporting so that it doesn't use the condition system
4474 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4475 COMPILE-FILE command)
4476 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4477 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4478 no longer supported, since their implementation was deeply intertwingled
4479 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4480 * removed bogus "support" for compiler macros named (SETF FOO), and
4481 removed the compiler macro for SETF INFO (but only after making a fool
4482 of myself on the cmucl-imp mailing list by posting a bogus patch for
4483 DEFINE-COMPILER-MACRO..)
4484 * Compiled files containing forms which have side effects on the Lisp
4485 reader (such as DEFPACKAGE forms) are now handled more correctly.
4486 (Compiler queuing of top level lambdas has been suppressed by setting
4487 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4488 * deleted various currently-unused source files, e.g. gengc.lisp. They
4489 may be added back at some point e.g. when porting to other architectures,
4490 but until they are it's distracting to distribute them and to try to
4492 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4493 there were so many of them they're just distractions, and UNCROSS is
4494 known to be able to handle the current sources
4495 * moved PROFILE functionality into TRACE, so that it will be clear
4496 how the wrapping and unwrapping of functions when you profile them
4497 interacts with the wrapping and unwrapping of functions when you
4498 trace them. (Actually, the functionality isn't there yet, but at least
4499 the interface specification is there. Hopefully, the functionality will
4500 arrive with some future maintenance release.)
4501 * removed host-oops.lisp
4502 * changed signature of QUIT function to allow UNIX-CODE argument
4503 * fixed READ-SEQUENCE bug
4504 * tweaked verbose GC output so that it looks more like the progress
4505 output that ANSI specifies for functions like LOAD
4506 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4507 * added <http://sbcl.sourceforge.com> to the banner information printed by