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 * bug fix: SB-PROFILE will no longer report extra consing for nested
11 calls to profiled functions.
12 * bug fix: if file compilation is aborted, the partial fasl is now
13 deleted, and COMPILE-FILE returns NIL as the primary value.
14 * bug fix: number of thread safety issues relating to SBCL's internal
15 hash-table usage have been fixed.
16 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
17 be returned from its body when the values were being returned
18 using unknown-values return convection and the W-P-O was wrapped
19 inside an UNWIND-PROTECT.
20 * bug fix: sb-posix should now compile again under Windows, enabling
23 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
24 * incompatible change: hash-table accessor functions are no longer
25 automatically protected by locks. Concurrent accesses on the same hash-table
26 from multiple threads can give inconsistent results or even corrupt the
27 hash-table completely. Multi-threaded applications should do their own
28 locking at the correct granularity. In the current implementation it is
29 still safe to have multiple readers access the same table, but it's not
30 guaranteed that this property will be maintained in future releases.
31 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
32 and will signal an error at runtime.
33 * enhancement: SB-THREAD package now exports a semaphore interface.
34 * enhancement: CONS can now stack-allocate on x86 and
35 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
37 * enhancement: nested lists can now be stack allocated on
38 platforms providing stack allocation support.
39 * enhancement: dynamic-extent support has been extended to support
40 cases where there are multiple possible sources for the stack
42 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
43 if the mutex is uncontested on Linux.
44 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
45 as specified by the CLHS. (thanks to Tobias Rittweiler)
46 * bug fix: NaN comparison now works on x86-64.
47 * bug fix: CLOSE :ABORT T on a stream with pending output now
49 * bug fix: instances of non-standard metaclasses using standard
50 instance structure protocol sometimes missed the slot type checks
52 * bug fix: known functions can be inlined in high-debug code.
54 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
55 * minor incompatible change: the MSI installer on Windows no longer
56 associates .lisp and .fasl files with the installed SBCL.
57 * minor incompatible change: :UNIX is no longer present in *FEATURES*
58 on Windows. (thanks to Luis Oliviera)
59 * new feature: SB-CLTL2 contrib module now implements
60 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
61 * optimization: scavenging weak pointers is now more efficient,
62 requiring O(1) instead of O(N) per weak pointer to identify
63 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
64 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
65 method bodies using constant slot-names when the first argument is
66 a specializer parameter for the method.
67 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
68 names, when the class of the instance is a direct instance
69 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
70 as slow as the constant slot-name case.)
71 * optimization: member type construction is now O(N) instead
73 * optimization: UNION and NUNION are now O(N+M) for large
74 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
76 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
77 now more readable in environments like Slime which display it.
78 (thanks to Tobias C. Rittweiler)
79 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
80 was able to infer the correct argument type for the object on which
81 the CAS operation was being performed.
82 * bug fix: copy propagation interfered with parallel assignment
83 semantics in local calls. (reported by Paul Khuong)
84 * bug fix: the signed modular fixnum shift compiled to wrong code on
85 x86 and x86-64. (spotted by a slight modification to some of
88 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
89 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
90 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
91 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
93 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
94 and generic function now print the object identity for anonymous
95 instances. (thanks to Attila Lendvai)
96 * enhancement: as an extension to MOP, SBCL now supports SETF of
97 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
98 (thanks to Attila Lendvai)
99 * enhancement: sb-cover has an alternate annotation mode which puts
100 the form annotations at the car of the form, instead of annotating
101 it completely. The benefit of this mode is that it shows better
102 which forms were instrumented by the compiler.
103 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
104 by printer control variables. (reported by Dan Corkill)
105 * bug fix: system leaked memory when delayed output was performed by
106 the OS in smaller chunks then expected. (thanks to David Smith)
107 * bug fix: system leaked memory when file streams were not closed
109 * bug fix: large objects written to slow streams that were modified
110 after the write could end up with the modified state written to
111 the underlying file descriptor.
112 * bug fix: multiple threads operating in parallel on the same stream
113 could cause buffer-overflows.
114 * bug fix: source location information is stored correctly
115 (broken since 1.0.6). This bug would generally show up as the
116 Slime debugger higlighting the wrong form when the sldb-show-source
118 * bug fix: Branch forms are again annotated as branches in the sb-cover
120 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
121 on x86-64. (reported by Gregory Vanuxem)
122 * bug fix: EQUALP could return wrong results for structures with raw
123 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
124 word). (reported by Vjacheslav Fyodorov)
125 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
128 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
129 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
130 atomic compare-and-swap operations on threaded platforms.
131 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
132 allows assining a global minimum value to optimization qualities
133 (overriding proclamations and declarations).
134 * enhancement: closed over variables can be stack-allocated on x86
136 * performance bug fix: GETHASH and (SETF GETHASH) are once again
138 * optimization: slot definition lookup is now O(1). This speeds up
139 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
140 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
141 * optimization: ASSOC and MEMBER can now be open-coded for all
142 combinations of keyword arguments when second argument is constant
143 and SPEED >= SPACE. In other cases a specialized version is
145 * bug fix: using obsoleted structure instances with TYPEP and
146 generic functions now signals a sensible error.
147 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
148 (reported by Kristoffer Kvello)
149 * bug fix: backtrace construction is now more careful when making
150 lisp-objects from pointers on the stack, to avoid creating bogus
151 objects that can be seen by the GC.
152 * bug fix: defaulting of values in contexts expecting more than 7
153 variables now works on x86-64. (reported by Christopher Laux)
154 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
156 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
157 as the property-list of a symbol.
158 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
159 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
160 situations requested, are once again file-compileable. (reported
163 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
164 * MOP improvement: support for user-defined subclasses of
165 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
166 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
167 the expansion of DEFMETHOD, is responsible for generating a form
168 which creates a list of specializers when evaluated. Additional
169 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
170 debugging and introspective support.
171 * minor incompatible change: the (unsupported) spinlock interface
172 has changed: free spinlock now has the value NIL, and a held spinlock
173 has the owning thread as its value.
174 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
175 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
177 * enhancement: name of a socket-stream is now "a socket" instead of
179 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
180 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
181 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
182 for generic functions. (Thanks to Tobias C. Rittweiler)
183 * optimization: bignum printing speed has been improved by 20-40%
184 (depending on the bignum size.)
185 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
187 * bug fix: the cache used by the CLOS to store precomputed effective
188 methods, slot offsets, and constant return values is now thread and
190 * bug fix: generic function dispatch function updating is now thread
191 and interrupt safe (in the sense that the known issues have been
193 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
194 * bug fix: interning EQL-specializers is now thread and interrupt safe.
195 * bug fix: asdf systems with dependencies to the SB-POSIX or
196 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
197 * bug fix: interrupt safety of applicable method computation has been
200 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
201 * new contrib: sb-cover, an experimental code coverage tool, is included
203 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
204 significantly faster.
205 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
206 produce more efficient code on x86-64 if the compiler can determine
207 that (+ INDEX <constant>) does not require a bounds check and FOO
208 has an element type at least 8 bits wide.
209 * enhancement: a new, experimental synchronous timeout facility is
210 provided. Refer to SB-SYS:WITH-DEADLINE for details.
211 * enhancement: when a symbol name conflict error arises, the
212 conflicting symbols are always printed with a package prefix.
213 (thanks to Kevin Reid)
214 * enhancement: stepping is now once again supported on the SPARC and
215 MIPS platforms. (It is also now more likely to work on CheneyGC
217 * enhancement: sb-sprof can now also track and report accurate call
219 * bug fixes: the treatment of non-standard subclasses of
220 SB-MOP:SPECIALIZER is more correct.
221 * incompatible change: PURIFY no longer copies the data from the
222 dynamic space into the static and read-only spaces on platforms
223 that use the generational garbage collector
224 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
226 * bug fix: GC race condition occasionally resulting in crashes with
227 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
229 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
231 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
232 a STYLE-WARNING to be signalled (regression from 1.0.4.)
233 * bug fix: an asynchronous interrupt could previously leave the
234 system running with GC inhibited.
235 * bug fix: a DECLARE form evaluated at top-level now causes an error
236 rather than silently (or verbosely) returning NIL.
237 * bug fix: trying to dissassemble functions compiled at high DEBUG could
238 result in a error being signalled due to source form lookup errors.
239 (reported by Peter Graves)
241 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
242 * incompatible change: removed writer methods for host-ent-name,
243 host-ent-addresses -- changing the values did not update the DNS
245 * minor incompatible change: changed experimental JOIN-THREAD interface
246 * documentation: the manual now lists reader and writer methods
247 in class slot documentation sections. (thanks to Richard M Kreuter)
248 * documentation: unwinding from asyncronous events has been
249 documented as unsafe.
250 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
251 in multithreaded application code.
252 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
253 platforms. (thanks to James Anderson for the optimization hint)
254 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
256 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
257 variants no longer cons.
258 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
259 their NOT- variants no longer cons.
260 * optimization: EQUAL hash tables no longer use SXHASH for objects
261 of all data types, but instead use an EQL hash for types for which
262 EQUAL is the same as EQL.
263 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
264 are significantly faster.
265 * optimization: new STRING-OUTPUT-STREAM implementation is much
266 faster for large outputs and conses 30% less on average.
267 * enhancement: XREF information is now collected for references made
268 to global variables using SYMBOL-VALUE and a constant argument.
269 * enhancement: SIGINT now causes a specific condition
270 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
271 * bug fix: FILE-POSITION used to signal an error in some cases where
272 ANSI requires it to return NIL.
273 * bug fix: ADJUST-ARRAY is now interrupt-safe.
274 * bug fix: adding and removing fd-handlers is now interrupt-safe.
275 * bug fix: inlined calls to C now ensure 16byte stack alignment on
277 * bug fix: bad type declaration in the CLOS implementation has
278 been fixed. (reported by James Anderson)
279 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
280 (thanks to Lutz Euler)
281 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
282 required by the CLHS. (reported by Samium Gromoff)
283 * bug fix: dead unbound variable references now signal an error.
284 (reported by Marco Monteiro)
285 * bug fix: / with an unused value was being deleted in safe code.
286 (thanks to Marco Monteiro and Kevin Reid)
287 * bug fix: number of characters that can be written onto a single
288 line in a file is unlimited.
289 * bug fix: some GC deadlocks caused by asynchronous interrupts have
290 been fixed by inhibiting interrupts for when GC is disbled.
291 * bug fix: some interrupt safety issues with GETHASH, (SETF
292 GETHASH), CLRHASH and REMHASH have been fixed.
293 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
294 specifier no longer causes infinite recursion.
295 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
296 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
297 * bug fix: an &environment argument with the correct variable information
298 is passed to macros that are expanded during byte compilation.
299 (reported by Samium Gromoff)
300 * bug fix: a logic error was causing the hash function for bignums to
301 have excessive amounts of collisions. (reported by Faré Rideau)
302 * bug fix: modifying the contents of an array could change the return
303 value of SXHASH on that array, which is only allowed for strings
304 and bit vectors. (bug introduced in 0.9.16)
305 * improvement: the x86-64/darwin port now passes all tests (except
306 for the debugger tests) but should still be considered
307 experimental until this is fixed.
308 * improvement: a style-warning is signaled for CASE (etc) clauses with
309 duplicate keys. (thanks to Kevin Reid)
310 * improvement: macos/x86 and macos/x86-64 now use mach exception
311 handlers for dealing with illegal instructions (for trapping and
312 error handling) and memory protection violations (for GC).
314 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
315 * new platform: experimental support for x86-64/darwin (MacOS).
316 * incompatible change: the thread-safe (on most platforms) getaddrinfo
317 and getnameinfo sockets functions are used instead of gethostbyaddr
318 and gethostbyname, on platforms where the newer functions are available.
319 As a result, the ALIASES field of HOST-ENT will always be NIL on these
321 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
322 don't print the startup banner, but behave as if --noinform was passed
323 as a command line argument. (thanks to Kevin Reid)
324 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
325 * optimization: code using alien values with undeclared types is much faster.
326 * optimization: the compiler is now able to open code SEARCH in more cases.
327 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
328 * bug fix: using standardized COMMON-LISP special variables as loop
329 variables no longer signals bogus package lock violations. (reported
331 * bug fix: declaring local loop variables to be of a range-limited type
332 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
333 (reported by Andras Simon)
334 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
335 bugs remain on x86-64.)
336 * bug fix: the #= and ## reader macros now interact reasonably with
337 funcallable instances.
338 * bug fix: type-checks for function arguments were compiled using the
339 compiler policy settings of the wrong lexical environment (in 1.0.2
341 * bug fix: SHADOW accepts characters as string designators, as required
342 by the spec (thanks to Eric Marsden)
343 * bug fix: fixed GC safety issues when foreign functions are called with
344 non-base strings as arguments
345 * bug fix: more consistent error output for fatal-compiler-warnings, like
347 * bug fix: functions with proper names were showing up as NIL in
348 backtraces (reported by Edi Weitz, regression in 1.0.2)
350 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
351 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
352 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
353 produce more efficient code on the x86 if the compiler can determine
354 that (+ INDEX <constant>) does not require a bounds check and FOO
355 has an element type at least 8 bits wide.
356 * bug fix: references to undefined variables are handled the same way
357 in toplevel forms as in the normal compiler
358 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
360 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
361 (thanks to Jon Buller)
362 * improvement: faster compilation times for complex functions
363 * improvement: added readlink support to SB-POSIX (thanks to Richard
366 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
367 * improvement: experimental support for mach exception handling on
368 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
370 * improvement: support for GBK external format.
371 (thanks to Chun Tian (binghe))
372 * improvement: the debugger now displays variables that have been closed
373 over, in code compiled with (DEBUG 2) or higher
374 * improvement: support for executable cores on NetBSD (thanks to
376 * new feature: added a RESTART-FRAME debugger command
377 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
378 be used to provide an implementation for FILE-POSITION on Gray streams
379 (thanks to Eric Marsden)
380 * new feature: add syslog, openlog and closelog support to SB-POSIX
381 (thanks to Richard Kreuter)
382 * optimization: the function call overhead in code compiled with
383 a high DEBUG optimization setting is significantly
384 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
385 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
386 (thanks to Tony Martinez)
387 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
388 evaluated once (reported by Kevin Reid)
389 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
390 works again (thanks to Kilian Sprotte)
391 * bug fix: an error is signaled for tagbodies with duplicate tags
392 (thanks to Stephen Wilson)
393 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
394 * bug fix: Win32 port can now handle foreign code unwinding Lisp
395 stack frames from alien callbacks.
396 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
397 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
398 65k pages (thanks to David Woodhouse)
399 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
401 changes in sbcl-1.0.1 relative to sbcl-1.0:
402 * new platform: FreeBSD/x86-64, including support for threading.
403 * new feature: the compiler stores cross-referencing information
404 abount function calls (who-calls), macroexpansion (who-macroexpands)
405 and special variables (who-binds, who-sets, who-references) for code
406 compiled with (< SPACE 3). This information is available through the
407 sb-introspect contrib.
408 * new feature: users may subclass SEQUENCE, and have instances of
409 these classes interoperate with standard Common Lisp functions if
410 a number of methods are defined. (This feature is experimental
411 and the interface subject to change based on feedback from SBCL
412 users and the general community)
413 * improvement: sb-sprof traces call stacks to an arbitrary depth on
414 x86/x86-64, rather than the previous fixed depth of 8
415 * improvement: another pthread back-end of mutex "pthread-futex".
416 * bug fix: non-ascii command-line arguments are processed correctly
417 (thanks to Yaroslav Kavenchuk)
418 * bug fix: non-required arguments were not passed correctly when a method
419 defined using DEFMETHOD was called from a mop-generated method using
420 CALL-NEXT-METHOD (reported by Pascal Costanza)
421 * bug fix: recursion is now permitted in accessors through
422 SLOT-UNBOUND. (reported by Pascal Costanza)
423 * bug fix: an error was signaled at startup if the HOME environment
424 variable was defined, but had an empty value (reported by Peter Van Eynde)
425 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
426 signal a PROGRAM-ERROR, not a BUG.
427 * bug fix: SB-POSIX:READDIR works when built with large file support.
428 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
430 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
431 for better automatic indentation support. (thanks to Matt Pillsbury)
432 * optimization: loading generic functions no longer takes O(n^2) time,
433 proportional to the amount of methods in the generic function
434 (reported by Todd Sabin and Jeremy Brown)
435 * optimization: the FIND and POSITION family of sequence functions
436 are significantly faster on arrays whose element types have been
438 * improvements to the Windows port:
439 ** Intermittent heap corruption problems have been fixed. (thanks
440 to Alastair Bridgewater)
441 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
443 ** Lisp is able to unwind foreign exception frames from alien
444 callbacks. (thanks to Alastair Bridgewater)
446 changes in sbcl-1.0 relative to sbcl-0.9.18:
447 * improvement: experimental support for threading on FreeBSD/x86.
448 (thanks to NIIMI Satoshi)
449 * improvement: runtime option --dynamic-space-size can be used
450 to set the size of the dynamic space reserved on startup.
451 * improvement: floating point modes in effect are now saved in
452 core, and restored on startup.
453 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
454 startup, not time since first call to GET-INTERNAL-REAL-TIME.
455 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
456 threads are not running after *SAVE-HOOKS* have run.
457 * improvement: writes to CLOS instance slots are type-checked in code
458 compiled with (SAFETY 3)
459 * improvement: floating-point exception handling on FreeBSD (thanks to
461 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
462 (thanks to Zach Beane)
463 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
465 * improvement: added support for the Shift-JIS external format.
466 (contributed by NIIMI Satoshi)
467 * improvement: callbacks are supported on Linux/PPC. (thanks to
469 * bug fix: compiler bug triggered by a (non-standard) VALUES
470 declaration in a LET* was fixed. (reported by Kaersten Poeck)
471 * bug fix: file compiler no longer confuses validated and already
472 dumped structurres. (reported by Kaersten Poeck)
473 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
474 fill-pointer signals a type-error as required. (thanks to
476 * bug fix: disassemly of funcallable instances works.
477 * bug fix: single stepping on PPC.
478 * bug fix: fix thread-safety problems in the type system (generally
479 manifesting as nonsensical errors like "STRING is a bad type specifier
480 for sequences" or "The value 1 is not of type FIXNUM")
481 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
482 (reported by Josip Gracin)
483 * bug fix: an error is signaled for attempts to displace arrays with
484 incompatible element types (thanks to Mario Mommer)
485 * bug fix: more correct handling of wide characters in debug info
486 (bug reported by Attila Lendvai and fixed by Juho Snellman)
487 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
488 and don't cause extra consing
489 * optimization: MAP and MAP-INTO are significantly faster on vectors
490 whose elements types have been declared.
491 * Improvements to SB-SPROF:
492 ** Support for allocation profiling
493 ** Reduced profiling overhead, especially for long profiling runs
494 * Improvements to the Windows port:
495 ** floating point exceptions are now reported correctly.
496 ** stack exhaustion detection works partially.
497 ** more accurate GET-INTERNAL-REAL-TIME.
498 ** better RUN-PROGRAM behaviour in case of errors in the child
500 ** PROBE-FILE now simplifies pathnames correctly.
501 ** DIRECTORY now works correctly with :WILD-INFERIORS.
502 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
503 1st 1970 (although time timezone data will be unreliable).
504 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
506 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
507 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
508 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
510 * bug fix: two potential GC deadlocks affecting threaded builds.
511 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
512 returns T (reported by Anton Kazennikov)
513 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
514 with non-variable places
515 * bug fix: remove a race condition in the setting of
516 funcallable-instance functions, this should make threaded CLOS
517 code more stable against memory faults.
518 * bug fix: corruption of specials when unbinding is interrupted by an
519 asynchronous unwind (reported by Hannu Koivisto)
520 * improvement: the debugger will now also display local variables that
521 are only used once, for code compiled with a DEBUG optimization quality
524 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
525 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
526 * incompatible change: External-format support for FFI calls. The
527 SB-ALIEN:C-STRING no longer implies an ASCII
528 external-format. Instead the string is subject to external-format
529 conversion. Additionally return values of type SB-ALIEN:C-STRING
530 are by default of type (SIMPLE-ARRAY CHARACTER), not
531 SIMPLE-BASE-STRING. For an alien type that behaves like the old
532 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
533 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
534 * incompatible change: SB-EXT package no longer contains the
535 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
536 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
538 * incompatible change: the single-stepper is no longer available
539 on Alpha, Mips and Sparc platforms.
540 * minor incompatible change: the direct superclasses of
541 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
542 not (STANDARD-OBJECT FUNCTION). This makes the
543 class-precedence-lists of GENERIC-FUNCTION and
544 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
546 * minor incompatible change: the default stream external format on
547 non-unicode SBCL is no longer detected from the locale, but is always
549 * new feature: Add a version of evaluator that uses an interpreter instead
550 of the compiler. EVAL still uses the compiler by default, to switch it
551 to use the interpreter, set the value of the variable
552 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
553 * minor incompatible change: the single-stepper REPL has been merged
554 with the normal debugger (see the "Stepping" heading of the debugger help
555 for more details). The debugger command STEP will no longer switch
556 to the single-stepper REPL.
557 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
558 for a type now works.
559 * bug fix: loading of default sysinit file works. (thanks to Leonid
561 * bug fix: better detection of circularities in the file-compiler.
562 (reported by Marco Monteiro)
563 * bug fix: the CL pathname functions now work with files that have
564 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
565 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
566 systems with Unix98 pty semantics.
567 * bug fix: ASDF-INSTALL will now work with bsd tar.
568 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
570 * bug fix: timers expiring in dead threads no longer cause a
571 type-error (reported by Paul "Nonny Mouse").
572 * bug fix: thanks to more lightweight single-stepper instrumentation,
573 code compiled with (DEBUG 3) will compile and execute significantly faster,
574 and will have more accurate type-inferencing than before
575 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
576 whose bindings are modified
577 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
578 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
579 ** replace dummy implementations of CL:MACHINE-INSTANCE and
580 CL:SOFTWARE-VERSION with proper version
582 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
583 * feature: implemented the READER-METHOD-CLASS and
584 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
585 as specified by AMOP.
586 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
588 * optimization: faster LOGCOUNT implementation on x86 and x86-64
589 (thanks to Lutz Euler)
590 * optimization: hashing of general arrays and vectors has been
591 improved. (reported by Any Fingerhut)
592 * enhancement: SB-INTROSPECT is now able to find definitions of
593 profiled functions. (thanks to Troels Henriksen)
594 * enhancement: compiler-macro expansion applies now to FUNCALL forms
595 as well, allowing compiler-macros for SETF-functions to expand.
596 * enhancement: step-instrumentation no longer wraps known
597 single-value functions with multiple-value context, allowing
598 better type inference.
599 * fixed bug #337: use of MAKE-METHOD in method combination now works
600 even in the presence of user-defined method classes. (reported by
601 Bruno Haible and Pascal Costanza)
602 * fixed bug #339(c): if there are applicable methods not part of any
603 long-form method-combination group, call INVALID-METHOD-ERROR.
604 (reported by Bruno Haible)
605 * fixed bug #361: the :FUNCTION initarg in the protocol for
606 initialization of methods can now be used to override
607 internally-produced optimized functions. (reported by Bruno
609 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
610 system-provided lambda expression no longer cause warnings about
611 unbound #:|pv-table| symbols.
612 * bug fix: improved the handling of type declarations and the
613 detection of violations for keyword arguments with non-constant
615 * bug fix: potentially erronous calls to PATHNAME and
616 MERGE-PATHNAMES were being flushed in some cases.
617 (reported by Richard Kreuter)
618 * bug fix: compiled calls to TYPEP were mishandling obsolete
619 instances. (reported by James Bielman and Attila Lendvai)
620 * bug fix: format strings with ~> without matching ~< no longer
621 trigger an AVER, but signal an understandable error instead.
622 (reported by Antonio Martinez)
623 * bug fix: specifying an output-file without a pathname-type for
624 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
625 of input-file instead of "fasl". (reported by Robert Dodier)
626 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
627 (reported by James Y Knight).
628 * bug fix: compiler-macros-function did not consider the environment
629 argument for shadowing by local functions.
630 * bug fix: compiler-macros expansion was inhibited by local INLINE
632 * bug fix: inline expansions of known functions were subject to
633 step-instrumentation in high debug policies, leading to problems
635 * bug fix: compiler failed to differentiate between different CONS
637 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
638 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
639 * bug fix: #368: incorrect use of expressed vs. upgraded array
641 * bug fix: #306a: more precise unions of array types.
642 * thread-safety improvements:
643 ** CONDITION-WAIT could return early on Linux, if the thread was
644 interrupted and subsequently continued with SIGCONT.
645 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
647 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
648 * added support for the ucs-2 external format. (contributed by Ivan
650 * minor incompatible change: pretty printing of objects of type
651 (cons symbol) is, in the default pprint-dispatch-table, now
652 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
654 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
655 called later in a class's lifetime, possibly as late as when the
656 first instance of the class is created. Previously,
657 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
658 class became finalizeable.
659 * fixed bug: FILE-POSITION sometimes returned inconsistent results
660 for multibyte external-format streams. (thanks to "vbzoli")
661 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
662 with :ALLOCATION :CLASS inherited from superclasses of the
664 * fixed bug: anonymous classes can now be created using the :NAME
665 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
666 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
667 * fixed bug: core-files saved with :EXECUTABLE T can again be
668 executed when SBCL_HOME isn't set. (reported by James Knight)
669 * fixed bug: toplevel LOCALLY forms with declarations could
670 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
671 * fixed bug: printing from several different threads using different
672 values of *print-case* could cause invalid output, due to
673 some internal special variables of the printer not being bound
674 thread-locally (reported by Max Mikhanosha)
675 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
676 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
677 (reported by Pascal Costanza)
678 * fixed bug: FINALIZE-INHERITANCE is called from
679 REINITIALIZE-INSTANCE on classes when the class has previously
680 been finalized, as required by AMOP.
681 * minor code generation optimizations:
682 ** better register allocation in CLOS dispatching functions
683 ** overflow detection when coercing signed bytes to fixnums on x86-64
684 ** is now implemented with one IMUL instruction instead of three shifts
685 ** more efficient bit-vector access on x86 and x86-64
686 ** more efficient access to raw structure slots on x86 and x86-64
687 * fixed some bugs revealed by Paul Dietz' test suite:
688 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
691 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
692 * feature: thread support on Solaris/x86, and experimental thread support
694 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
696 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
697 default method for SLOT-UNBOUND.
698 * minor incompatible change: prevent the user from specializing the
699 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
700 somewhat counter to the intent of the protocol, I (CSR) think, and
701 additionally it just doesn't work in SBCL as currently
702 implemented, thanks to optimizations (that are always valid for
703 the other three SLOT-VALUEish functions, but not for the setter).
704 * minor incompatible change: the :SB-LDB feature is now enabled by
705 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
706 the low-level debugger.
707 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
708 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
709 down to an SB-SYS:FD-STREAM, allowing direct communication in
710 more cases. Temporary files and pipes are still used as a fallback
712 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
713 on the non-reentrant C functions localtime(3) and gmtime(3).
714 * bug fix: global optimization policy was not visible in LOCALLY and
716 * bug fix: class objects can be used as specializers in methods.
717 (reported by Pascal Costanza)
718 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
719 with a type error. (reported by blitz_ on #lisp)
720 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
721 failed with a type error. (reported by Pascal Bourguignon)
722 * bug fix: merging pathnames against defaults with :DIRECTORY
723 starting with '(:RELATIVE :BACK) should preserve the :BACK.
724 (reported by James Y Knight)
725 * bug fix: saving large (>2GB) cores on x86-64 now works
726 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
727 constant index and a (simple-array (signed-byte 32)) array
728 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
729 error (patch by Robert J. Macomber)
730 * feature: TIME now displays the amount of run-time spent in GC
731 * bug fix: The debugger now does a better job of respecting
732 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
733 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
734 workaround for bug 403.)
735 * bug fix: floating point exception handling now works on FreeBSD
736 (thanks to NIIMI Satoshi)
737 * fixed some bugs revealed by Paul Dietz' test suite:
738 ** MISC.641: LET-conversion were not supposed to work in late
740 * improvements to the Win32/x86 port:
741 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
742 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
744 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
745 * new feature: source path information is generated for macro-expansion
746 errors for use in IDE's like Slime (thanks to Helmut Eller)
747 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
748 * bug fix: compiler does not loop forever on an invalid type in
750 * improvement: compilation of most CLOS applications is significantly
752 * optimization: added a limited bytecode compiler for simple toplevel
753 forms, speeding up compilation and FASL loading
754 * bug fix: the statistical profiler now properly distinguishes anonymous
757 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
758 * minor incompatible change: in sbcl-0.9.11 (but not earlier
759 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
760 into subclasses of the finalized class. Now user calls to
761 FINALIZE-INHERITANCE finalize just the one class, and calls by the
762 system return before any subclasses are finalized.
763 * minor incompatible change: The reader no longer ignores errors
764 regarding non-existent packages in #+ and #- feature tests.
765 * new feature: command line options --no-sysinit, --no-userinit to
766 inhibit loading the corresponding init files
767 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
768 for sending data through UDP sockets (thanks to François-René Rideau)
769 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
770 error is signalled instead (thanks to François-René Rideau)
771 * minor incompatible change: Error signalling behaviour of lexical
772 operations violating package locks has changed slightly. Refer to
773 documentation on package locks for details.
774 * enhancement: EVAL can process IF-expressions without resorting to the
776 * bug fix: Some operations involving SIN, COS, or TAN weren't being
777 constant-folded properly due to missing out-of-line functions.
778 (reported by Mika Pihlajamäki)
779 * bug fix: LISTEN sometimes returned T even in cases where no data was
780 immediately available from the stream
781 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
782 were reversed. (reported by Levente Mészáros)
783 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
784 when given a vector displaced to an adjustable array. (reported
786 * bug fix: derivation of float boundaries from numbers outside the
787 appropriate float range (reported by John Wiseman)
788 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
790 * bug fix: precomputing cache entries for generic functions with
791 some subclasses of specializers as yet invalid does not attempt to
792 fill a cache line with a negative offset. (reported by Levente
794 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
795 allow condition class objects as arguments to DOCUMENTATION and
796 (SETF DOCUMENTATION); only find and set documentation for
797 structure names for the STRUCTURE doc-type. (suggested by Gary
799 * improvements to the Win32/x86 port:
800 ** (user-homedir-pathname) and default initialization file
801 locations now know about the user's "Documents and Settings"
802 directory (thanks to Yaroslav Kavenchuk)
803 ** run-program is implemented (thanks to Mike Thomas)
804 ** sockets support (thanks to Timothy Ritchey)
805 ** better backtrace support (thanks to Alastair Bridgewater)
806 ** sb-grovel supported
807 ** asdf-install and sb-posix work somewhat
808 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
809 * improvements to the Solaris/x86 port:
810 ** works on Solaris 11/Solaris Express
811 ** floating-point exception handling support
812 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
813 * fixed some bugs revealed by Paul Dietz' test suite:
814 ** REMOVE-METHOD returns its generic function argument even when
815 the method is not one of the generic functions' methods.
816 ** objects declared as MEMBER types can be admissible arguments to
818 ** printing characters should simply be printed by the FORMAT ~:C
820 ** compiler failure when compiling functions with hairy constant
821 defaults for optional parameters.
822 ** compiler produces wrong code when MAYBE-INLINE-expanding a
823 function, which is already optimized.
825 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
826 * new platform: experimental support for SBCL x86/Darwin, including
827 MacOS X 10.4.5 on Intel.
828 * new feature: Unicode character names are now known to the system
829 (through CHAR-NAME and NAME-CHAR).
830 * new feature: the filesystem location of SBCL's core file is
831 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
832 * minor incompatible change: the contrib modules SB-POSIX and
833 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
834 this change is to make it easier to distribute
835 location-independent binaries.
836 * bug fix: as implied by AMOP, standardized classes no longer have
837 slots named by external symbols of public packages. (reported by
839 * optimization: calling functions via constant symbols -- as in
840 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
841 via the function object as in (FUNCALL #'FOO).
842 * optimization: on x86 and x86-64, the calling convention for the
843 case of dealing with an unknown number of values has been altered
844 to be friendlier to the prediction heuristics implemented,
845 particularly with reference to CALL and RET pairing. (thanks to
846 Alastair Bridgewater)
847 * optimization: on x86, the code for access to thread-local symbol
848 values and binding/unbinding of thread-local symbols is smaller.
849 (thanks to Alastair Bridgewater)
850 * enhancement: CONSTANTP is now able to determine constantness of
851 more complex forms, including calls to constant-foldable standardized
852 functions and some special forms beyond QUOTE.
853 * fixed bug: occasional GC crashes on Solaris/x86
854 * optimization: x86-64 supports stack allocation of results of simple
855 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
856 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
857 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
858 (thanks to James Knight)
859 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
860 by the standard (thanks to Stephen Compall)
862 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
863 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
864 be used for bundling the runtime and the core file into one
865 executable binary. This feature is not currently supported on all SBCL
866 platforms. (thanks to James Bielman and NIIMI Satoshi)
867 * new feature: a generational or ephemeral garbage collector is now
868 the default on the PowerPC platform (both Linux and Darwin). The
869 old Cheney (stop and copy) collector is a build-time option.
870 (thanks to Cyrus Harmon, after Raymond Toy)
871 * minor incompatible change: the method by which SBCL finds its
872 contributed modules has changed; it no longer relies on symbolic
873 links from an $SBCL_HOME/systems directory, but searches directly
874 in the subdirectories of $SBCL_HOME.
875 * enhancement: the dynamic heap size on the Linux/PPC platform is
876 markedly larger, even using the older Cheney garbage collector.
877 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
878 certain complicated string types. (reported by Gary King)
879 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
880 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
881 * fixed bug: DOCUMENTATION on structures no longer signals an
882 error if no structure type of the right name exists. (reported by
884 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
885 applicable non-standard primary or :AROUND method on
886 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
887 no longer cause unbound variable errors. (reported by Kevin Reid)
888 * fixed bug: metacircle resolution in cases where methods have slots
889 added before the slots from STANDARD-METHOD. (reported by Jean
891 * fixed bug: the Power PC assembly code for calling into lisp
892 sometimes computed its return address wrongly (depending on the
893 15th bit of the address, fixed at link time). This used to
894 manifest itself in a segmentation violation while building PCL.
895 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
896 many others over the years)
897 * fixed bug: ROOM no longer reports silly numbers for stack usage if
898 the stack pointer (treated as a signed integer) is negative.
899 (thanks to Peter van Eynde)
901 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
902 * new platform: experimental support for the Windows operating
903 system has been added. (thanks to Alastair Bridgewater)
904 * fixed several bugs in and robustified the PPC FFI (including
905 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
906 * bug fix: allow non-simple string symbol names (reported by Paul
908 * bug fix: interrupt handling on NetBSD (thanks to Richard M
910 * bug fix: saving a core corrupted callbacks on x86/x86-64
911 * bug fix: closed a loophole in metacircularity detection and
912 grounding in the PCL implementation of CLOS.
913 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
914 (reported by tomppa on #lisp)
915 * bug fix: clear the direction flag on Lisp -> C transitions, as
916 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
917 (reported by Andrej Grozin and Hendrik Maryns)
918 * optimization: major improvements to GC efficiency on GENCGC platforms
919 * optimization: faster implementation of EQUAL
920 * optimization: emit more efficient opcodes for some common
921 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
922 * optimization: several other minor code-generation improvements on x86-64
923 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
925 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
926 * minor incompatible change: (SETF CLASS-NAME) and (SETF
927 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
928 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
929 by ANSI as a generic function, and for consistency (SETF
930 GENERIC-FUNCTION-NAME) is treated likewise.
931 * fixed bug #233.b: make constraint propagation notice when a variable
932 value is changed after it is referenced but before it is used
933 * fixed bug #296: no more arbitrary behaviour from filenames with
934 odd characters as --load arguments. (reported by Adam Warner)
935 * fixed bug #390: :CHARACTER-SET pathname components now work as
936 expected. (reported by Tim Daly Jr)
937 * fixed bug #391: complicated :TYPE intersections in slot
938 definitions no longer cause an error in PCL internals.
939 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
940 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
941 returning the number of octets which would be written to the
942 file-stream. (thanks to Robert J. Macomber)
943 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
944 arguments to the internal stream functions. (thanks to David
946 * bug fix: stack exhaustion is now handled on the x86/SunOS
947 platform. (thanks to Daisuke Homma)
948 * bug fix: more accurate ROOM results on GENCGC platforms
949 * optimization: improved type inference for arithmetic-for
950 index variables in LOOP
951 * optimization: faster floating-point SQRT on x86-64
952 * fixed some bugs revealed by Paul Dietz' test suite:
953 ** DOCUMENTATION returns NIL instead of "" for method combinations
954 that don't have a docstring
956 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
957 * minor incompatible change: (SETF CLASS-NAME) and (SETF
958 GENERIC-FUNCTION-NAME) are no longer generic functions, and
959 therefore are not customizeable by user code (as seems to be at
960 least permitted and maybe required by AMOP). As a consolation,
961 however, the SBCL implementation of these functions now calls
962 REINITIALIZE-INSTANCE as specified by AMOP.
963 * bug fix: slot-definition documentation is propagated as per ANSI
964 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
965 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
966 argument, as do eslotd initialization methods. (from Pascal
967 Costanza's "Closer" project)
968 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
969 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
971 * bug fix: it is now possible to have more than one subclass of
972 STANDARD-GENERIC-FUNCTION without causing stack overflow.
973 (reported by Bruno Haible, Pascal Costanza and others)
974 * bug fix: the dependent update protocol now works for generic
975 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
977 * bug fix: condition-class instances corresponding to
978 DEFINE-CONDITION forms are now created eagerly. (reported by
979 Kalle Olavi Niemitalo on comp.lang.lisp)
980 * bug fix: floating point printing is more accurate in some
981 circumstances. (thanks to Simon Alexander)
982 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
983 merged with *DEFAULT-PATHNAME-DEFAULTS*.
984 * bug fix: callbacks on OS X now preserve stack-alignment.
985 (thanks to Cyrus Harmon)
986 * enhancement: the x86-64 disassembler is much better at
987 disassembling SSE instructions. (thanks to Lutz Euler)
988 * enhancement: CHECK-TYPE on a variable now allows type-inference
989 (thanks to Vincent Arkesteijn)
990 * optimization: improved performance of EUC-JP external format.
991 (thanks to NIIMI Satoshi)
992 * optimization: performance improvements to IO on file streams of
993 :ELEMENT-TYPE CHARACTER
994 * optimization: much faster memory allocation on x86-64
995 * optimization: faster garbage collections (latency and throughput)
997 * optimization: faster bignum subtraction and division on x86 and x86-64
998 (thanks to Lutz Euler)
999 * optimization: more accurate type-inference for loops that use a
1000 floating point index variable or a negative step.
1002 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1003 * bug fix: add a workaround to SBCL looping infinitely at startup on
1004 Linux kernels with apparently buggy implementations of personality().
1005 (thanks to Svein Ove Aas)
1006 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1007 (thanks to David Lichteblau)
1008 * bug fix: local bindings shadow global symbol macros for
1009 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1010 * optimization: non-open coded uses of numeric comparison operators
1011 (e.g. >) no longer cons when called with more than one parameter
1012 on platforms supporting dynamic-extent allocation.
1013 * enhancement: saving cores with foreign code loaded is now supported
1014 on MIPS/Linux in addition to the previously supported platforms.
1015 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1016 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1017 * bug fix: division by zero in sb-sprof when no samples were collected
1018 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1019 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1020 (thanks to Alexander Kjeldaas)
1021 * new feature: ignoring the return values of destructive functions
1022 like NREVERSE signals a compile-time style-warning.
1023 (thanks to Kevin Reid)
1024 * new feature: the alignment of alien structure fields can be
1025 explicitly specified. (thanks to Cyrus Harmon)
1026 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1027 and dump core on SIGQUIT
1029 ** incompatible change: threads do not inherit values of specials
1030 from their parents (see manual)
1031 ** bug fix: threads stacks belonging to dead threads are freed by the
1032 next exiting thread, no need to gc to collect thread stacks anymore
1033 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1034 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1035 ** bug fix: dynamic variable and thread start related gc lossage
1036 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1038 * fixed some bugs revealed by Paul Dietz' test suite:
1039 ** SUBTYPEP is slightly more accurate on heinously complicated
1040 CONS types where some of the members have uncertain (in the
1041 NIL, NIL sense) type relationships to each other.
1043 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1044 no longer purified unless :PURIFY T is explicitly specified.
1045 ** Non-purified cores are significantly smaller than before
1047 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1048 * new feature: timers based on Zach Beane's excellent timer package
1049 * added support for the following external formats: koi8-u,
1050 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1051 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1052 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1053 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1054 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1055 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1056 * incompatible change: a threaded SBCL will no longer revert to
1057 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1058 * bug fix: interrupts are disabled until startup is complete; no
1059 more sigsegvs when receiving a signal to soon
1060 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1062 * bug fix: add a workaround for the memory randomization features in
1063 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1064 workaround will only be in effect on systems with the proc filesystem
1066 * bug fix: printing objects of type HASH-TABLE signals a
1067 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1069 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1070 have been running for over 50 days. (reported by Gilbert Baumann)
1071 * bug fix: the logic for getting names of functions gets less
1072 confused when confronded with alternate-metaclass
1073 funcallable-instances. (reported by Cyrus Harmon)
1074 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1075 consistent, even on internal alternate-metaclass objects.
1076 * bug fix: flush closure information collected by physical
1077 environment analysis prepass before the main pass. (bug reported
1079 * bug fix: compiler pack phase does not modify a hashtable, which is
1080 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1081 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1082 may be expanded. (reported by Alexander <malishev> on c.l.l)
1083 * bug fix: no more style-warnings when compiling code using LAST
1084 under high-SPEED low-SPACE settings. (reported by David Wallin
1086 * fixed some incompatibilities between SBCL's MOP and the MOP
1088 ** the METAOBJECT class is now implemented;
1089 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1090 STANDARD-OBJECT, as required;
1091 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1092 now compatible (as required by VALIDATE-SUPERCLASS); there
1093 remains a consistency requirement over the presence or absence
1094 of the FUNCTION class in the superclass list of finalized
1095 classes; see the manual for more details;
1096 ** the :AROUND method for COMPUTE-SLOTS on
1097 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1098 requested slot ordering.
1100 ** bug fix: parent thread now can be gc'ed even with a live
1102 ** bug fix: binding a special with PROGV to no value is not
1103 the same as that symbol not having been bound (thanks to
1105 ** bug fix: binding specials is thread safe (thanks to
1107 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1108 and SB-SPROF do not die with 'no handler for signal XX in
1109 interrupt_handle_now(..)' anymore
1110 ** bug fix: WITH-TIMEOUT works with multiple running threads
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1113 the :method-class keyword argument.
1115 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1116 * new port: the Solaris operating system on x86 processors is now
1117 mostly supported, though some rough edges in the environment
1118 remain. (thanks to Daisuke Homma)
1119 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1120 space for its heap. (thanks to Thiemo Seufer)
1121 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1122 closures. (thanks to Thiemo Seufer)
1123 * minor incompatible change: eof selects abort in the debugger.
1124 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1125 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1127 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1128 deprecated, and will go away in a future revision of SBCL.
1129 * minor incompatible change: GC-ON and GC-OFF are no longer
1130 implemented with a counter, it does not matter how many times gc
1131 is switched on or off
1132 * bug fix: discriminating functions for generic function classes
1133 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1134 make invalid assumptions about method precedence order. (reported
1136 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1137 platforms. (thanks to Thiemo Seufer)
1138 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1139 * bug fix: correct pathname printing with printer escaping is on.
1140 (thanks to Kevin Reid)
1141 * bug fix: complex VOP definitions in "user-space" no longer trigger
1142 package locks. (reported by Zach Beane)
1143 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1144 for funcallable-instance functions, and is no different from
1146 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1147 characters in the current readtable. (reported by Nicholas Neuss)
1148 * bug fix: SUBTYPEP on various CONS types returns more a more
1149 accurate acknowledgment of its certainty.
1150 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1151 lists in some cases. This partially fixes bug 384.
1152 * flush all standard streams before prompting in the REPL and the
1154 * bug fix: signal handling and triggering gc do not conflict
1155 directly anymore, in particular a high frequency sb-sprof does
1156 not prevent gc from running
1157 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1158 approximation for timezone and DST information between the
1159 universal time epoch and the smallest negative 32-bit time_t.
1160 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1161 year 1899 so long as the encoded time is non-negative.
1162 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1163 the correct answer for non-constant rotate amounts when performing
1164 an inline 32-bit rotation.
1166 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1167 there is only one thread in the session
1168 ** bug fix: memory leak for streams created in one thread and
1169 written to in another
1170 ** bug fix: lockup when compiled with gcc4
1171 ** bug fix: race that allows the gc to be triggered when gc is
1173 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1174 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1176 ** bug fix: finished threads are gc'ed properly
1177 * fixed some bugs revealed by Paul Dietz' test suite:
1178 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1179 has arguments with a different set of applicable methods from
1180 the orignal arguments.
1181 ** The type error thrown by MAP now has a correct expected-type
1183 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1184 name a compiled function.
1185 ** Three MISC tests where a large bignum was improperly coerced to
1186 a float that couldn't represent that bignum during type
1187 derivation were fixed.
1188 ** SUBTYPEP can now handle types involving the negation of a
1189 list-form FUNCTION type.
1190 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1191 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1192 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1194 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1195 * New feature: Experimental support for bivalent streams: streams
1196 opened with :element-type :default now allow character and binary
1197 (unsigned-byte 8) I/O
1198 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1199 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1201 * bug fix: no more highly sporadic "couldn't check whether ~S is
1202 readable" when reading a stream and an interrupt hits in the middle
1203 of a select system call
1204 * compiler better recognizes complex arrays (reported by Eduardo
1206 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1207 for variables on SBCL built with linkage-tables. (reported by Luis
1209 * various error reporting improvements.
1210 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1211 (Thanks to Thiemo Seufer)
1212 * bug fix: nested reader invokations work correctly
1213 * bug fix: it is possible to have simultaneous references to foreign
1214 code and foreign data with the same name.
1216 ** added x86-64 support
1217 ** incompatible change: the threading api now works with thread
1218 objects instead of thread ids
1219 ** bug fix: threads are protected from signals and interruption when
1220 starting up or going down
1221 ** bug fix: a race where an exiting thread could lose its stack to gc
1222 ** bug fix: don't halt on infinite error in threads if possible
1223 ** fixed numerous gc deadlocks introduced in the pthread merge
1224 ** bug fix: fixed thread safety issues in read and print
1225 ** bug fix: debugger doesn't hang on session lock if interrupted at
1226 an inappropriate moment
1227 ** bug fix: run-program is now thread safe(r)
1228 ** bug fix: inner with-recursive-lock no longer releases the mutex
1229 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1232 EXPECTED-TYPE slots filled.
1233 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1235 ** MAKE-CONDITION accepts classes as type-designators.
1236 ** COMPILE may never return NIL.
1237 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1238 range before calling Unix time functions
1240 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1241 * numerous signal handling fixes to increase stability
1242 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1243 * minor incompatible change: we now correctly canonize default
1244 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1245 as per the MOP, rather than the historical (INITARG INITFUNCTION
1246 INITFORM). (reported by Bruno Haible)
1247 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1248 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1249 for more information.
1250 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1251 pathname is a directory pathname.
1252 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1253 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1255 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1256 :METACLASS argument in addition to a class name. (reported by
1257 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1258 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1259 (reported by Bruno Haible; patch from Gerd Moellmann)
1260 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1262 * bug fix: more cleanups to the floating point exception handling on
1263 x86-64 (thanks to James Knight)
1264 * bug fix: the compiler does not try to constant fold calls of
1265 COERCE and BIT-* functions when they return freshly constructed
1266 objects. (reported by Nikodemus Siivola and Paul Dietz)
1267 * optimization: improved the MIPS versions of generic (in the
1268 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1269 * optimization: direct conversion of (unsigned-byte 32) to floats on
1270 the PowerPC platform.
1271 * optimization: structure instances with raw slots now use less
1272 memory, and probably show better memory locality. (thanks to
1274 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1275 are treated as requests for stack allocation on the x86-64,
1276 Alpha-32, PPC and SPARC platforms.
1277 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1278 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1279 (thanks to Tony Martinez)
1281 ** gcing a dead thread can no longer lead to lockups
1282 ** threads block signals until they are set up properly
1283 ** errno is no longer shared by threads
1284 ** interrupt-thread restores the eflags register on x86
1285 ** fixed some lockups due to gc/thread interaction
1286 * fixed some bugs revealed by Paul Dietz' test suite:
1287 ** invalid dotted lists no longer raise a read error when
1288 *READ-SUPPRESS* is T
1289 ** signal an error if a symbol that names a declaration is used
1290 as the name of a type, or vice versa
1291 ** allow using the (declare (typespec var*)) abbreviation for
1292 (declare (type typespec var*)) with all type specifiers
1293 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1294 ** always return NIL from PPRINT-POP when OBJECT is NIL
1295 ** don't signal errors when pretty-printing malformed LABELS,
1296 FLET or MACROLET forms
1297 ** declarations in a DOLIST body are also in scope for the
1299 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1301 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1304 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1305 * fixed cross-compiler leakages that prevented building a 32-bit
1306 target with a 64-bit host compiler.
1307 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1308 opened with :IF-EXISTS :OVERWRITE.
1309 * fixed bug 281: error for an invalid qualifier in a short-form method
1310 combination method is not signalled until the faulty method is called.
1311 * bug fix: iteration variable type inferrer failed to deal with open
1312 intervals. (reported by Alan Shields)
1313 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1314 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1315 or more runtime options were provided to the sbcl binary.
1316 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1317 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1318 * optimization: there's now a fast-path for fixnum arguments in the
1319 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1320 * optimization: the code generated on x86-64 is more compact thanks
1321 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1322 * bug fix: floating-point exception handling now partly works on
1323 x86-64. (Thanks to James Knight)
1324 * improvement to the MIPS backend from Thiemo Seufer: C-style
1325 64-bit long long arguments and return values to alien functions
1327 * fixed some bugs revealed by Paul Dietz' test suite:
1328 ** the type-error signalled from WARN has a filled-in DATUM slot.
1329 ** the type-error required when a stream is not associated with
1330 a file has the stream as its datum.
1331 ** type-errors on single-floats on x86-64 no longer have
1332 :INVALID-OBJECT as the datum
1333 ** the type-errors signalled for invalid function names now have
1334 a correct expected type
1335 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1336 for structures defined with a :TYPE. Documentation strings for
1337 typed structures are no longer immediately discarded
1338 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1339 broadcast streams and synonym streams. FILE-LENGTH now also works
1340 on broadcast streams.
1342 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1343 * incompatible change: the --noprogrammer option, deprecated since
1344 version 0.7.5, has been removed. Please use the equivalent
1345 --disable-debugger option instead.
1346 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1348 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1349 inoperational for a while now) has been completely removed.
1350 * null lexical environments are now printed as #<NULL-LEXENV>,
1351 significantly reducing the amount of clutter in typical backtraces.
1352 * documentation on weak pointers, finalization, and after GC hooks
1353 has been added to the manual.
1354 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1355 as other specialized array types, is much faster. SUBSEQ and
1356 COPY-SEQ on such arrays have also been sped up.
1357 * optimization: EQL is now more efficient when at least other argument
1358 is known to be of type (OR FIXNUM (NOT NUMBER)).
1359 * fixed bug: the runtime is now less vulnerable to changes in the
1360 size of the SBCL object on OS X, and virtual memory is reserved for
1361 all spaces that need to be at a fixed address.
1362 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1363 * fixed bug: finalizers and after GC hooks that cause consing are now
1365 * fixed bug: compiler error messages and summaries are now printed to
1366 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1367 * fixed inference of the upper bound of an iteration variable.
1368 (reported by Rajat Datta).
1369 * fixed bug 211e: calling local functions with duplicated constant
1370 keyword argument no longer causes a bogus style warning about an
1372 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1373 ftype declaration to be disregarded. (reported by Dave Roberts)
1374 * fixed bug 373: caused by erronous compilation of references to alien
1375 variables in the runtime on ppc/darwin.
1376 * fixed bug 376: CONJUGATE type deriver.
1377 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1378 (reported by Baughn on #lisp)
1379 * a cross-compiler bug on non-x86 platforms has been identified and
1380 fixed. (thanks to Bruno Haible)
1381 * improvements to the MIPS runtime code for increased stability.
1382 (thanks to Thiemo Seufer)
1383 * increased the maximimum compact environment size to allow
1384 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1385 * partial workaround for bug 135: don't name the function we're
1386 calling for hairy cases of EVAL, so as not to accumulate one environment
1387 entry per call to EVAL. (reported by Kevin Reid)
1388 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1389 * optimization: type testing for non-vector arrays should be faster.
1390 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1392 * bug fix: setting 31st element of a bit vector to zero did not work
1394 * bug fix: redefining a class definition which failed due to a
1395 previous accessor / function clash now works (but see BUGS entry
1396 #380 for more problems in this area). (thanks to Zach Beane)
1397 * the long form of DEFINE-METHOD-COMBINATION disables method group
1398 checking when given a single method group with pattern *. (thanks
1400 * on x86 compiler supports stack allocation of results of simple
1401 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1402 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1403 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1404 * fixed some bugs related to Unicode integration:
1405 ** the restarts for recovering from input and output encoding
1406 errors only appear when there is in fact such an error to
1408 * fixed some bugs revealed by Paul Dietz' test suite:
1409 ** MISC.549 and similar: late transformation of unsafe type
1410 assertions into derived types caused unexpected code
1412 ** SCALE-FLOAT type deriver is less wrong.
1413 ** type derivers for EXP, LOG and similar functions compute result
1414 types for complex arguments better.
1415 ** (MISC.563) CONJUGATE type deriver works for very restricted
1417 ** out-of-line type testers for character strings are available.
1418 ** EQUAL compiler transform understands specialness of objects
1420 ** accessing double-floats stored on the stack now works on x86-64.
1421 ** debugger internals could sometimes create invalid lispobjs,
1422 resulting in GC crashes.
1423 ** MISC.548: type check weakening can convert required type into
1425 ** initialization forms for bindings are not in scope of free special
1428 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1429 * incompatible change: thread support for non-NPTL systems has
1430 been removed - locking is buggy and unreliable. A threaded
1431 SBCL build will now warn at startup and refuse to create
1432 new threads, unless futex support is detected in the kernel
1433 * incompatible change: the top level REPL now has only an ABORT
1434 restart associated with it, not TOPLEVEL and ABORT as it used to.
1435 TOP and TOPLEVEL are now available as debugger commands for
1436 returning to the top level.
1437 * incompatible change: forms evaluated in the REPL now use the
1438 global optimization policy.
1439 * incompatible change: user- and system-initialization files are
1440 no longer processed with LOAD, but by READ and EVAL; hence the
1441 global optimization policy, startup package, readtable, etc,
1443 * The .fasl file format number has been incremented because of
1444 various incompatible changes.
1445 * internal entry point details and argument counts no longer appear
1446 in backtraces unless explicitly requested by setting
1447 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1448 * built-in and standard functions no longer have names like "top
1449 level local call to FOO".
1450 * fixed bug 32: functions defined in non-null lexical environments
1451 now have more legible printed representation
1452 * fixed bug 33: functions defined in non-null lexical environemnts
1453 are now more amenable to inspection by INSPECT.
1454 * workaround for bug 354: XEPs no longer appear in backtraces unless
1455 explicitly requested.
1456 * fixed bug: receiving the signal which results from attempting to
1457 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1458 SIGBUS on other *BSDs) on architectures where the C stack is also
1459 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1460 to debugger confusion or wild instability. (reported by Cheuksan
1462 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1463 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1464 (reported by Lutz Euler)
1465 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1466 compiled code. (reported by Paul Dietz)
1467 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1468 specialized parameter to a method to something that is not TYPEP
1469 the specializer is now possible.
1470 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1471 face of package deletion.
1472 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1473 pathname. (reported by David Tolpin on c.l.l)
1474 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1475 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1476 STRING-STREAM and FILE-STREAM, respectively.
1477 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1478 than previously; in addition, its test suite is now run on build.
1479 (thanks to Luke Gorrie)
1480 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1481 * optimization: added a immediate representation for single-floats
1483 * fixed some bugs revealed by Paul Dietz' test suite:
1484 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1485 ** attempting to create a package with a colliding nickname causes
1486 correctable errors to be signalled.
1487 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1488 ** several x86-64 backend bugs related to sign-extension of immediate
1491 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1492 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1493 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1495 * fixed loading of multiply forward-referenced layouts.
1496 (thanks to Cheuksan Wang)
1497 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1498 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1499 (see COMPILE-FILE documentation for details of :PRINT option)
1500 * fixed bugs 19 and 317: fixed-format floating point printing is
1501 more accurate. This also fixes a bug reported by Adam Warner
1502 related to the ~@F format directive.
1503 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1505 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1506 dispatch table if the from-char is a dispatch macro character.
1507 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1508 by the wordsize no longer ignore the last word. (reported by Lutz
1510 * fixed bug in type checking of dynamic-extent variables. (reported
1512 * optimization: sequence traversal functions use their freedom to
1513 coerce function designators to functions.
1514 * optimization: code with many calls to CLOS methods specialized on
1515 CLOS classes has had redundant type checks removed.
1516 * optimization: added declarations to speed up operations that access
1517 the internal character database (for example STRING-UPCASE)
1518 * optimization: comparison operations between floats and sufficiently small
1519 fixnums no longer create extra rationals
1520 * fixed some bugs related to Unicode integration:
1521 ** portions of multibyte characters at the end of buffers for
1522 character-based file input are correctly transferred to the
1523 start of the buffer at the next read.
1524 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1525 passing it through to OPEN.
1526 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1527 argument given to internal calls to OPEN.
1528 ** the built-in comment readers (introduced by character sequences
1529 ";" and "#|") are more forgiving to encoding errors; they will
1530 STYLE-WARN and then attempt to resync the stream at a character
1531 boundary. (thanks to Teemu Kalvas)
1532 * fixed some bugs in the x86-64 port:
1533 ** Negative short int return values from c-calls are sign-extended
1535 ** The stack is aligned to 16-bytes for c-calls, as required by
1536 the ABI. (thanks to Cheuksan Wang)
1537 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1538 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1539 for enums. (thanks to Vincent Arkesteijn)
1540 ** Multiple small optimizations and bugfixes for floating point
1542 * fixed some bugs revealed by Paul Dietz' test suite:
1543 ** Space, Tab, Linefeed, Return and Page have the invalid
1544 secondary constituent character trait.
1545 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1547 ** WITH-INPUT-FROM-STRING should only update the index place on
1549 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1550 works more reliably.
1551 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1552 with one-dimensional arrays with fill pointers.
1553 ** TYPE-OF failed on a complex with an integer realpart and a
1555 ** compiler failure during type inference for the code of form
1556 (IF (EQL X (THE ...)) ...) (MISC.535).
1558 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1559 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1560 platform. The port supports 61-bit fixnums, large memory spaces
1561 and reloading shared object files.
1562 * enhancement: saving cores with foreign code loaded is now
1563 supported on ppc/Darwin in addition to the previously supported
1565 * enhancement: the statistical profiler now walks deeper into the
1566 call stack for more meaningful call-graphs and accrued time
1567 reports (x86/x86-64 only). It also now reports time spent in
1569 * enhancement: it is now possible to trace most individual methods
1570 of a generic function in addition to tracing the generic function
1572 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1574 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1575 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1576 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1577 produces an error. (thanks to Vincent Arkesteijn)
1578 * bug fix: NAMESTRING on pathnames with :WILD components in their
1579 directories works correctly. (thanks to Artem V. Andreev)
1580 * fixed bug 125: compiler preserves identity of closures. (reported
1582 * bug fixed: functions with &REST arguments sometimes failed with
1583 "Undefined function" when compiled with (DEBUG 3). (reported by
1585 * bug fix: overflow during compiling of setting element of a bit
1586 vector with constant index and value. (reported by Timmy Douglas)
1587 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1588 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1589 * on x86 compiler partially supports stack allocation of dynamic-extent
1591 * GO and RETURN-FROM do not check the extent of their exit points
1592 when compiled with SAFETY 0.
1593 * fixed some bugs related to Unicode integration:
1594 ** encoding and decoding errors are now much more robustly
1595 handled; it should now be possible to recover even from invalid
1596 input or output to the terminal. (thanks to Teemu Kalvas)
1597 ** provided a first cut at implementing STRING-TO-OCTETS and
1598 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1599 ** altered the SB-MD5 contributed module slightly, changing the
1600 interface just enough for it to be supportable for builds where
1601 lisp characters are not eight bits.
1602 * fixed some bugs revealed by Paul Dietz' test suite:
1603 ** the FORMATTER-generated functions for ~V[ conditionals require
1604 the correct number of arguments.
1605 ** READ-FROM-STRING returns the mandated second value when applied
1606 to displaced strings.
1607 ** the #\Rubout and #\Backspace characters are treated as invalid
1608 constituent characters by the tokenizer.
1610 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1611 * new feature: reloading changed shared object files with
1612 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1613 * new feature: references to foreign variables and functions
1614 can now be compiled and loaded before the corresponding shared
1615 object file is loaded, as long as the foreign definitions are
1616 available at runtime.
1617 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1618 supported. (thanks to Dan Debertin)
1619 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1620 just keywords. (thanks to Vincent Arkesteijn)
1621 * fixed bug #331: structure-class instances corresponding to
1622 DEFSTRUCT forms are now created eagerly.
1623 * fixed bug #345: backtraces from calls to undefined functions work
1624 on x86 as well. Related bug #61 is now also partially fixed on x86
1625 (backtraces from throws to unknown catch tags.)
1626 * bug fix: lambda-list parsing is now stricter vrt. order and number
1627 of lambda-list keywords.
1628 * bug fix: as specified by AMOP, an error is signalled if a
1629 class-option appears multiple times in a DEFCLASS form. (reported
1631 * bug fix: RANDOM can be compiled when the compiler derives the type
1632 of its numeric argument as a disjoint set of small integers.
1633 (reported by Paul Dietz)
1634 * bug fix: starting a core saved with shared objects loaded when
1635 those objects are not available no longer causes threaded SBCL to
1636 hang. (reported by Sean Ross)
1637 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1639 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1640 argument, which must always be NIL. (reported by Kalle Niemitalo)
1641 * bug fix: printing 1.0d+23 no longer results in an error.
1642 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1643 * bug fix: structure slot setters preserve evaluation order. (thanks
1645 * fixed some bugs related to Unicode integration:
1646 ** RUN-PROGRAM can allow its child to take input from a Lisp
1647 stream. (reported by Stefan Scholl)
1648 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1649 locales. (reported by Ken Causey)
1650 * fixed some bugs revealed by Paul Dietz' test suite:
1651 ** INCF, DECF and REMF evaluate their place form as specified in
1653 ** FORMATTER expands ~{ ~} iteration directives with V or #
1654 parameters correctly.
1655 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1656 consequent uses no arguments correctly.
1657 ** the system has a partial understanding of the (COMPLEX RATIO)
1659 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1662 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1663 * new feature: a build-time option (controlled by the :SB-UNICODE
1664 keyword feature, enabled by default) for building the system with
1665 support for the entire 21-bit character space defined by the
1667 * new feature: the system now has rudimentary external-format
1668 support; the primary user-visible change at this time is that
1669 characters with the high bit set (such as lower-case-e-acute) will
1670 print correctly to a terminal in a UTF-8 environment.
1671 * minor incompatible change: BASE-CHAR no longer names a class;
1672 however, CHARACTER continues to do so, as required by ANSI.
1673 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1674 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1675 has been moved to the SB-EXT package (temporarily re-exported from
1677 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1678 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1680 * fixed bug #351: better error-handlind and reporting for malformed
1682 * fixed bug #350: bignum-printing is now more memory-efficient,
1683 allowing printing of very large bignums, eg. (expt 2 10000000).
1684 (reported by Bruno Haible)
1685 * fixed bug #302: better primitive-type selection for intersection
1687 * fixed bug #308: non-graphic characters now all have names, as
1688 required. (reported by Bruno Haible)
1689 * bug fix: redefining a class with different superclasses now correctly
1690 removes it from the direct-subclasses of its previous superclasses.
1691 (reported by David Morse)
1692 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1693 new value now works. (reported by Bruno Haible)
1694 * bug fix: correct canonicalization of multiple non-standard slot
1695 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1696 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1697 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1698 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1699 now exists, an signals an error.
1700 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1701 during macroexpansion for non-iteration variables as well. (reported
1702 by Bruno Haible for CMUCL)
1703 * bug fix: Cyclic structures and unprintable objects in compiler
1704 messages no longer cause errors. (reported by Bruno Haible)
1705 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1706 and READ-FROM-STRING all now return a primary value of NIL if
1707 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1708 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1709 (reported by Bruno Haible for CMUCL)
1710 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1711 do not designate a condition. (reported by Bruno Haible for
1713 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1714 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1715 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1716 permitting resolution in favour of any of the conflicting symbols.
1717 (reported by Bruno Haible for CMUCL)
1718 * bug fix: EQUAL compiler optimizations is less aggressive on
1719 strings which can potentially compare true despite having distinct
1720 specialized array element types.
1721 * bug fix: unit enumerations can be defined without dividing by
1722 zero. (thanks to Vincent Arkesteijn)
1723 * FORMAT compile-time argument count checking has been enhanced.
1724 (report from Bruno Haible for CMUCL)
1725 * a partial workaround for the bug 262: the compiler does not try to
1726 inline-expand a local function doing RETURN-FROM from a deleted
1727 BLOCK. (thanks to Peter Denno for the bug report and to David
1728 Wragg for the simple test case)
1729 * fixed some bugs revealed by Paul Dietz' test suite:
1730 ** RENAME-PACKAGE allows all package designators as new package
1732 ** constraint propagation and lambda variable substitution are
1733 more cautious in dealing with partially deleted code.
1734 ** compiler sometimes lost reoptimization passes.
1735 ** CERROR, given a condition as condition designator, passes its
1736 remaining arguments to the continue format control without
1738 ** Case-altering FORMAT directives work correctly on non-ASCII
1740 ** The REAL type specifier handles bounds outside the floating
1741 point ranges without signalling FLOATING-POINT-OVERFLOW.
1742 ** Functions with IR1-transformations can create intercomponent
1743 references to global functions.
1744 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1746 ** FORMAT ~R treats a nil value for its first parameter correctly.
1748 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1749 * enhancement: saving cores with foreign code loaded is now
1750 supported on x86/NetBSD and sparc/Linux in addition to the previously
1751 supported platforms.
1752 * bug fix: on some platforms repeated installations caused multiple
1753 copies of HTML documentation to be installed -- should not happen
1754 any more. (reported by Stefan Scholl)
1755 * bug fix: parsing self-recursive alien record types multiple times
1756 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1757 original patch by Helmut Eller for CMUCL)
1758 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1759 (thanks to Richard Kreuter)
1760 * bug fix: defining classes whose accessors are methods on existing
1761 generic functions in other (locked) packages no longer signals
1762 bogus package lock violations. (reported by François-René Rideau)
1763 * bug fix: special variables as DEFMETHOD parameters no longer have
1764 associated bogus type declarations. (reported by David Wragg and
1766 * bug fix: read-write consistency on streams of element-type
1767 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1768 * bug fix: redefiniton of the only method of a generic function with
1769 no DEFGENERIC no longer emits a full WARNING. In addition,
1770 redefinition of generic functions with no DEFGENERIC to an
1771 incompatible lambda list now signals an error. (thanks to Zach
1773 * bug fix: DEFGENERIC now works even when there's a function of the
1774 same name in an enclosing lexical environment. (thanks to Zach
1776 * fixed compiler failure, caused by instrumenting code during
1777 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1778 * optimization: added loop analysis and improved register allocation
1779 * fixed some bugs revealed by Paul Dietz' test suite:
1780 ** POSITION on displaced vectors with non-zero displacement
1781 returns the right answer.
1782 ** (SIMPLE-STRING) is a valid type specifier for sequence
1784 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1786 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1787 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1789 ** The justification version of the FORMAT ~< directive treats
1790 non-zero minpad parameter correctly.
1792 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1793 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1794 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1795 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1796 the supported interface.
1797 * new feature: Single-stepping of code compiled with DEBUG 2 or
1798 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1799 * new feature: saving cores with foreign code loaded is now
1800 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1801 Timothy Moore's work for CMUCL)
1802 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1803 parameters to * instead of NIL if no initform is supplied.
1804 (reported by Johan Bockgaard)
1805 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1806 correctly when FUNCALL appears as the car of the form. Note:
1807 despite this FUNCALL forms are not currently subject to
1808 compiler-macro expansion. (port of Raymond Toy's fix for the
1809 same from CMUCL, reported by Johan Bockgaard)
1810 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1811 (thanks for Teemu Kalvas)
1812 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1813 case instead of terminating the iteration (thanks for Julian
1814 Squires, Sean Champ and Raymond Toy)
1815 * bug fix: incorrect expansion of defgeneric that caused a style
1816 warning. (thanks for Zach Beane)
1817 * bug fix: slot accessor effective method computation works properly
1818 for classes with multiple non-standard applicable methods on
1819 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1820 * on x86 compiler supports stack allocation of results of LIST and
1821 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1822 CMUCL implementation by Gerd Moellmann)
1823 * fixed some bugs revealed by Paul Dietz' test suite:
1824 ** FORMAT strings with both the ~<~:;~> form of the justification
1825 directive and pretty-printing directives cause an error.
1827 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1828 * incompatible change: the internal functions
1829 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1830 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1831 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1832 instead of the old functions.
1833 * new feature: on platforms where "dladdr" is available foreign
1834 function names now appear in backtraces. (based on Helmut Eller's
1836 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1837 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1839 * documentation: ASDF manual is now distributed with SBCL.
1840 * bug fix: SBCL can now load its contributed modules with REQUIRE
1841 even if the system-provided entries have been removed from
1842 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1844 * bug fix: backtraces involving undefined functions or assembly
1845 routines are more informative. (thanks to Brian Downing)
1846 * bug fix: mutually referent alien structures now work correctly.
1847 (reported by Rick Taube)
1848 * bug fix: structures defined by WITH-ALIEN can be referred to
1849 within other definitions in the same WITH-ALIEN.
1850 * bug fix: division operators (MOD, TRUNCATE and the like) with
1851 constant zero divisors and integer dividends no longer generate
1853 * bug fix: provide default methods for INPUT-STREAM-P and
1854 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1855 * bug fix: improve the detection and resolution of MOP metacycles
1856 (where effective method computation depends on the generic function
1857 having its effective method computed). (reported by Bruno Haible)
1858 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1859 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1860 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1861 * bug fix: pathnames with relative directory components can be
1862 represented relative to default pathnames.
1863 * optimization: in taking the GCD of bignums, reduce the two bignums
1864 to approximately the same size (using Euclid's algorithm) before
1865 applying the more sophisticated binary GCD. (thanks to Juho
1867 * optimization: COUNT on bitvectors now operates word-at-a-time.
1868 * optimization: ASH with a positive, but not necessarily constant,
1869 (leftwards) shift, when in a modular context, is compiled to a
1871 * fixed some bugs revealed by Paul Dietz' test suite:
1872 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1873 if the corresponding argument is NIL.
1874 ** FORMAT directives accepting a minpad parameter treat negative
1876 ** literal commas as character parameters to FORMAT directives are
1878 ** literal spaces directly after ~<Newline> directives within a
1879 format-logical-block (~:< ~@:>) do not induce :FILL-style
1880 conditional newlines.
1881 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1882 as specified: it no longer includes conditional newlines).
1883 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1885 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1886 :PER-LINE-PREFIX argument does not evaluate to a string.
1888 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1889 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1890 the manual for details; for now, package locks can be disabled by
1891 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1892 if no major problems are found then it is likely that they will be
1893 compiled in unconditionally.
1894 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1895 unsupported operators on all platforms. To load a shared library
1896 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1897 object file, link it into a shared library outside of SBCL and
1898 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1900 * bug workaround: ROOM T and the GC were not working together
1901 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1902 reporting facility aren't preserved. That reporting has been
1903 disabled, so now until and unless someone figures out how to make it
1904 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1905 * minor incompatible change: as threatened around sbcl-0.8.0, the
1906 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1907 The symbols are also exported from SB-PCL for backwards
1908 compatibility, but more so than before SB-PCL should be treated as
1909 an implementation-internal package.
1910 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1912 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1913 It is known as of this release not to work on the Alpha, however.
1914 * fixed bug #167: errors signalled due to illegal syntax in method
1915 bodies are now more legible.
1916 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1917 designators and can hence be used with TYPEP.
1918 * fixed bug #333: CHECK-TYPE now ensures that the type error
1919 signalled, if any, has the right object to be accessed by
1920 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1921 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1922 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1924 * fixed bug #334: programmatic addition of slots using specialized
1925 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1926 and :ALLOCATION :CLASS slots.
1927 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1928 float exponents. (rereported by Peter Seibel)
1929 * fixed a bug: #\Space (and other whitespace characters) are no
1930 longer considered to be macro characters in standard syntax by
1931 GET-MACRO-CHARACTER.
1932 * fixed bug: initialization of condition class metaobjects no longer
1933 causes an instance of the condition to be created. (reported by
1935 * fixed bug: it is now possible to have slots such that
1936 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1937 description is neither :INSTANCE nor :CLASS.
1938 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1939 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1940 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1941 non-standard SLOT-DEFINITION-ALLOCATION values.
1942 * fixed bugs: various race conditions handling exiting threads.
1943 CL-PPCRE's thread torture test now passes.
1944 * fixed arguably-a-bug: GC time proportional to number of threads in
1945 system even when most of them are idle
1946 * optimization: improved performance of BIT and SBIT on bit-vectors.
1947 * .fasl file incompatibility: The fasl file version number has
1948 been incremented because of changes associated with package locks.
1950 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1951 * minor incompatible change: the system no longer provides
1952 optimization hints (conditions which are TYPEP
1953 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1954 mode; these hints are still emitted when higher SPEED optimization
1956 * new contrib module: a sampling profiler (profiling by statistical
1957 sampling, rather than by instrumenting functions) is available as
1958 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1959 * the behaviour of the standard function ED is now customizeable by
1960 third parties through a hook variable: see ED's documentation
1961 string for information on the protocol.
1962 * the compiler no longer emits efficiency notes for (FUNCALL X)
1963 when the type of X is uncertain under default optimization
1965 * fixed bug 276: mutating a binding of a specialized parameter to a
1966 method to something that is not TYPEP the specializer is now
1968 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1969 exhibit the required behaviour on the x86 platform. (thanks to
1970 Peter van Eynde, Eric Marsden and Bruno Haible)
1971 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1972 even for difficult arguments. (reported by Peter Graves)
1973 * fixed bug 141a: the backquote printer now descends quoted
1975 * fixed another bug in backquote printing: no more destructive
1976 modification of the form's list structure. (reported by Brian
1978 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1979 move between its address being taken and the call to
1980 interrupt_thread, fixing a crashing race condition.
1981 * the SB-POSIX contrib implementation has been adjusted so that it
1982 no longer exhibits ridiculously poor performance when constructing
1983 instances corresponding to C structs.
1985 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1986 * minor incompatible change: the sb-grovel contrib now treats C
1987 structures as alien (in the sense of SB-ALIEN) objects rather than
1988 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1989 has implications for memory management of client code
1990 (sb-grovel-returned objects must now be manually managed) and for
1991 type safety (alien objects now have full types).
1992 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1993 used to control emission of compiler diagnostics, rather than the
1994 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1995 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1996 quality should be considered deprecated.
1997 * install.sh now installs the user manual as well
1998 * (not quite a new documentable feature, but worth considering in
1999 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2000 beginnings of a semantically meaningful condition hierarchy is
2001 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2003 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2004 designator as the defaults argument.
2005 * fixed bug: Displaced arrays whose displaced-to array has become
2006 too small now cause ARRAY-DIMENSION to signal an error, providing
2007 for safer bounds-checking. (reported by Bruno Haible)
2008 * fixed bug: DEFCLASS slot definitions with identical :READER and
2009 :WRITER names now signal a reasonable error. (reported by Thomas
2011 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2013 * fixed bug: Invalid slot specification errors now print correctly.
2014 (thanks to Zach Beane)
2015 * fixed bug 320: Shared to local slot value transfers in class
2016 redefinitions now happen corrently from superclasses as
2017 well. (reported by Bruno Haible)
2018 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2020 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2021 improper lists correctly. (reported by Bruno Haible)
2022 * fixed bug 313: source-transform for <fun-name> was erroneously
2023 applied to a call of a value of a variable with name <fun-name>.
2024 (reported by Antonio Menezes Leitao)
2025 * fixed bug 307: The obsolete instance protocol ensures that
2026 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2027 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2028 remove dead unknown-values globs from the middle of the stack.
2029 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2030 easier to resume long computations after using *BREAK-ON-SIGNALS*
2031 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2032 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2033 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2034 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2035 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2036 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2038 * fixed a (fairly theoretical) bug in string printing: if
2039 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2040 does not have array-element-type equal to the most general string
2042 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2043 function name. (thanks to Nikodemus Siivola)
2044 * fixed bug: compile-time format string checker failed on
2045 non-closed ~{. (reported by Thomas F Burdick)
2046 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2047 don't warn on legal loop code involving more than one aggregate
2048 boolean. (thanks to Nikodemus Siivola)
2049 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2050 in keyword position to MAKE-PACKAGE was not being reported as
2052 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2053 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2055 * fixed bugs 280 and 312: the checking for multiple definitions in a
2056 file is less likely to become confused by uses of inline
2058 * fixed bug: the #S reader macro performs the keyword coercion
2059 specified for slot names. (reported by Kalle Niemitalo)
2060 * fixed bug: lambda lists may contain symbols whose names start with
2061 & but are not lambda-list-keywords; their occurrence triggers a
2063 * fixed bug 321: define-method-combination argument lambda lists do
2064 not cause invalid code to be generated when &optional or &aux
2065 variables have default values. (reported by Bruno Haible)
2066 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2067 have CLOS classes; this appears also to have fixed reported
2068 instability in stack exhaustion detection.
2069 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2070 MUFFLE-WARNING when no associated restart is present is now
2072 * optimization: rearranged the expansion of various defining macros
2073 so that each expands into only one top-level form in a
2074 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2076 * optimization: used a previously little-used slot in symbols to
2077 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2079 * fixed some bugs revealed by Paul Dietz' test suite:
2080 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2082 ** Fixed an optimization bug related to inheritance of initforms
2083 from local to shared slots.
2084 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2085 ** CAST optimizer forgot to flush argument derived type.
2086 ** print/read consistency on floats is now orders of magnitude
2087 more likely. (thanks also to Bruno Haible for a similar report
2089 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2090 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2091 for objects of type REAL. Make it so.
2092 ** SXHASH is specified (infelicitously) to respect similarity,
2093 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2094 it so. (thanks to Markus Ziegler)
2095 ** on the Alpha, the compiler succeeds in compiling functions
2096 returning a known number of arguments greater than 63.
2097 ** fixed handling of invalid NIL arguments in keyword position
2099 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2101 ** provided a readably-printable representation for RANDOM-STATE
2103 ** ensured that pathnames, where they have a namestring, always
2104 print using #P"..." syntax.
2106 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2107 * Support for the forthcoming 2.0 version of the NetBSD kernel
2108 running on x86 hardware has been added. (thanks to Perry
2109 E. Metzger most immediately, and others for their past work)
2110 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2111 current release version; SBCL's OpenBSD support had been broken
2112 since about the time of OpenBSD's switch to ELF binary format.)
2113 * [placeholder for DX summary]
2114 ** user code with &REST lists declared dynamic-extent, under high
2115 speed or space and low safety and debug optimization policy.
2116 * The manual has been converted to Texinfo format and the debugger
2117 chapter from the cmucl manual has been added.
2118 * A facility has been added to extract documentation strings from
2119 sbcl and store them as Texinfo-formatted snippets for inclusion in
2120 the manual (via Texinfo's @include directive)
2121 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2122 the test case to Dave Roberts)
2123 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2124 pointers. (reported by Sean Ross)
2125 * bug fix: PROFILE output is printed nicely even for large numerical
2126 values. (thanks to Zach Beane)
2127 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2128 greater than 32 handle EOF correctly.
2129 * bug fix: on X86 an immediate argument of the IMUL instruction is
2130 correctly printed in disassembly. (thanks to Lutz Euler)
2131 * bug fix: class slots in redefined classes preserve their old
2132 values. (thanks to Bruno Haible and Nikodemus Siivola)
2133 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2134 succeeds. (reported by Marco Baringer)
2135 * fixed some bugs revealed by Paul Dietz' test suite:
2136 ** READ-SEQUENCE now works on ECHO-STREAMs.
2137 ** RATIONALIZE works more according to its specification. (thanks
2140 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2141 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2142 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2143 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2144 should matter to you only if you rebind the printer control
2145 variables and then find you want different bindings in the
2146 debugger than in the ordinary execution of your program.)
2147 * The runtime build system has been tweaked to support building
2148 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2149 assembler and linker. (thanks to Nikodemus Siivola)
2150 * Unbound, undefined, undeclared variables now trigger full
2151 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2152 more often programmer error than deliberate exploitation of undefined
2154 * optimization: the hash algorithm for strings has changed to one
2155 that is less vulnerable to spurious collisions. (thanks to Juho
2157 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2158 needless bounds checking. (thanks to Juho Snellman)
2159 * optimization: implemented multiplication as a modular
2160 (UNSIGNED-BYTE 32) operation on the PPC backend.
2161 * fixed some bugs revealed by Paul Dietz' test suite:
2162 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2164 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2165 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2166 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2168 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2170 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2172 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2173 their output stream on EOF from read.
2174 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2175 have been read to end-of-file.
2176 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2178 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2179 description of determination of which consecutive characters
2181 ** Printing the "Space" character with escaping on now yields "#\\ ",
2182 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2183 ** Reading floating-point numbers with *READ-BASE* set to a number
2184 less than 10 works correctly.
2185 ** Reading floating-point numbers with *READ-BASE* set to a number
2186 more than 10 works correctly.
2187 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2188 the readtable currently in effect.
2190 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2191 * minor incompatible change: parsing of namestrings on a physical
2192 (Unix) host has changed; numbers after the final #\. in a
2193 namestring are no longer interpreted as a version field. This is
2194 intented to be largely invisible to the user, except that the
2195 meaning of the namestring "*.*.*" has changed: it now refers to a
2196 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2197 should usually be replaced by
2198 (make-pathname :name :wild :type :wild :version :wild)
2199 with the added benefit that this is more likely to be portable.
2200 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2201 option to OPEN now signals an error if the file being opened
2202 exists; this may have an impact on existing code.
2203 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2204 signals received, on the PowerPC platforms (both Linux and
2205 Darwin). (thanks to Pierre Mai for pointing out the location of
2207 * several fixes on OS X: The system now builds and runs cleanly on
2208 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2209 (thanks to Brian Mastenbrook)
2210 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2211 second-resolution: integer multiples of 1/3600 between -24 and 24.
2212 (thanks to Vincent Arkesteijn)
2213 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2214 type when called with 1 argument; PEEK-CHAR checked type of
2215 PEEK-TYPE only after having read first character from a
2216 stream. (reported by Peter Graves)
2217 * bug fix: the garbage collector now has much better locality
2218 behaviour, and in particular no longer treats all memory as being
2219 exhausted when many small objects point to each other in a deeply
2221 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2222 recognized as being TYPEP their class.
2223 * bug fix: the PUSHNEW documentation string has been corrected.
2224 (thanks to Vincent Arkesteijn)
2225 * bug fix: defaulting of the value for the last of an atypically
2226 large number of multiple values being bound was not being
2227 performed correctly on the Alpha or PPC platforms
2228 * optimization: implemented multiplication as a modular
2229 (UNSIGNED-BYTE 32) operation on the x86 backend.
2230 * optimization: SEARCH on simple-base-strings can now be open-coded.
2231 (see also contrib/compiler-extras.lisp for inspiration for
2232 teaching the compiler about the Boyer-Moore algorithm).
2233 * value, returned by MAX (and MIN) called with several EQUALP, but
2234 not EQL, arguments now does not depend on compiler settings.
2235 * fixed some bugs revealed by Paul Dietz' test suite:
2236 ** in stack analysis liveness information is propagated from
2237 non-local entry points.
2238 ** pathwise CAST removing failed when the CAST node did not start
2240 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2241 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2243 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2244 host is already defined.
2245 ** RENAME-FILE works on streams instead of signalling an internal
2247 ** PEEK-CHAR uses the current readtable when determining whether
2248 or not a character is whitespace.
2249 ** MERGE-PATHNAMES handles the case when the pathname does not
2250 specify a name while the default-pathname specifies a version.
2251 ** Pathnames now stand a better chance of respecting print/read
2253 ** Attempting to use standardized file system operators with a
2254 pathname with invalid :DIRECTORY components signals a
2256 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2258 ** DIRECTORY on logical pathnames is more correct.
2259 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2260 signal a TYPE-ERROR if their argument is not a stream.
2261 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2262 designator argument does not designate a stream.
2263 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2264 examining the synonym.
2265 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2267 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2268 element-type larger than ([UN]SIGNED-BYTE 32).
2270 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2271 * When built with the :SB-FUTEX feature, threaded builds now take
2272 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2273 for faster/more reliable mutex and condition variable support.
2274 * Incompatible change (but one you probably shouldn't have been using
2275 anyway): the interface and code for arbitrating between multiple
2276 threads in the same user session has been redesigned.
2277 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2278 error when called without an explicit environment argument.
2279 (thanks to Nikodemus Siivola)
2280 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2281 confused about their position. (thanks to Adam Warner and Gerd
2283 * bug fix: Pretty printing backquoted forms with unquotations in the
2284 argument list position of various code constructs such as LAMBDA
2285 now works correctly. (reported by Paul Dietz)
2286 * bug fix: Pretty printing unquotations no longer loses all
2287 stream position information.
2288 * optimization: performance of string output streams is now less
2289 poor for multiple small sequence writes.
2290 * optimization: performance of CSUBTYPEP in the presence of complex
2291 expressions involving CONS and NOT many times has been improved.
2292 (reported by Paul Dietz)
2293 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2295 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2296 streams for the REPL, for the debugger, and so on. (thanks to
2298 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2299 * fixed some bugs revealed by Paul Dietz' test suite:
2300 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2301 is ignored if it is too small, rather than propagating through
2302 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2303 ** extremely complex negations of CONS types were not being
2304 sufficiently canonicalized, leading to inconsistencies in
2306 ** VALUES tranformer lost derived type.
2308 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2309 * fixed a bootstrapping bug: the build process no longer assumes
2310 that the various BOOLE-related constants have the same value in
2311 host and target lisps. (noted by Paul Dietz' test suite on an
2312 SBCL binary built from CLISP)
2313 * The system can now be dynamically linked on the MIPS platform,
2314 which enables dynamic loading of foreign code from Lisp. (thanks
2315 to Ralf Baechle for discussions on the MIPS ABI)
2316 * The system now records debugging information for its own source
2317 files in a filesystem-position-independent manner, relative to
2318 the "SYS" logical host.
2319 * fixed a compiler bug: MV-LET convertion did not check references
2320 to the "max args" entry point. (reported by Brian Downing)
2321 * tweaked disassembly notes to be less confident about proclaiming
2322 some instruction as an LRA. (thanks to Brian Downing)
2323 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2324 now each have their own history, command character, and other
2325 characteristics. (thanks to David Lichteblau)
2326 * fixed some bugs revealed by Paul Dietz' test suite:
2327 ** compiler failure in compiling LOGAND expressions including a
2329 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2330 shift greater than 32.
2331 ** FLUSH-DEST did not mark blocks for type check regeneration.
2332 ** HANDLER-CASE failed to accept declarations in handler clauses
2333 in some circumstances.
2335 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2336 * New code in contrib/sb-introspect (still probably not entirely
2337 stable yet) provides some support for smart Lisp development
2338 environments like SLIME.
2339 * The conditions signalled for errors occurring when loading .fasl
2340 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2341 in a way which should help ASDF recover gracefully.
2342 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2343 stuff has been cleaned up. If you code contrib/ stuff, this might
2344 affect you, and you can look at contrib/README, contrib/STANDARDS,
2345 and/or the 0.8.4.27 diff to check.
2346 * In full calls the compiler now does not generate checks for declared
2347 argument types for all arguments.
2348 * various threading fixes
2349 ** and some experimental patches which didn't make it into
2350 the main tree for this release, but which are shipped in
2351 contrib/experimental-thread.patch as a possible fix for some
2352 failures (deadlock, spinning...) in GC-intensive multithreaded
2354 * fixed PPC build problem (source code incompatibility of different
2355 library versions): added offsetof() hackery which attempts to divine
2356 where glibc maintainers put uc_mcontext today
2357 * fixed bug 282: compiler does not trust type assertions while passing
2358 arguments to a full call.
2359 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2360 &OPTIONAL VALUES type parameter.
2361 * fix bug 214: algorithm for noting rejected templates is now more
2362 similar to that of template seletion. (also reported by rydis on
2364 * fixed bug 141b: printing backquoted information readably and prettily
2365 inserts a space where necessary.
2366 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2367 (CHARACTER 10) are now reported as errors, rather than propagated
2368 as unknown types. (reported by piso on #lisp)
2369 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2370 with values NIL and :ERROR. (thanks to Milan Zamazal)
2371 * fixed bug 191c: CLOS now does proper keyword argument checking as
2372 described in CLHS 7.6.5 and 7.6.5.1.
2373 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2374 longer raise an error; further, using a list as a for-as-arithmetic
2375 counter now raises a meaningful error.
2376 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2377 now understood by sequence creation functions such as MAKE-SEQUENCE
2379 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2380 a STRING-INPUT-STREAM.
2381 * compiler enhancement: SIGNUM is now better able to derive the type
2383 * type declarations inside WITH-SLOTS are checked. (reported by
2385 * fixed some bugs revealed by Paul Dietz' test suite:
2386 ** incorrect optimization of TRUNCATE for a positive first
2387 argument and negative second.
2388 ** compiler failure in let-convertion during flushing dead code.
2389 ** compiler failure while deriving type of TRUNCATE on an
2390 interval, containing 0.
2391 ** ASH of a negative bignum by a negative bignum count now returns
2393 ** intersection of CONS types now canonicalizes properly, fixing
2394 inconsistencies in SUBTYPEP.
2396 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2397 * incompatible change: The --disable-debugger command line
2398 option now clobbers the debugger at a more fundamental
2399 level, by redefining #'INVOKE-DEBUGGER instead of by
2400 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2401 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2402 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2403 enter the debugger REPL and then suffer recursive errors
2404 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2405 while under the new system, BREAK in --disable-debugger mode
2406 terminates the system just as an unhandled error would.
2407 * fixed compiler performance when processing loops with a step >1;
2408 * bug fix: DOCUMENTATION now retrieves generic function
2409 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2410 support has been systematized, and now supports the methods
2411 specified by ANSI, along with a default method and a method for
2412 slot documentation. (reported by Nathan Froyd)
2413 * bug fix: effective methods associated with a generic function are
2414 no longer cached over a change of that generic function's method
2415 combination. (reported by Andreas Fuchs)
2416 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2417 names a non-existent directory. (thanks to Andreas Fuchs)
2418 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2419 the PPC platform, lead to stack corruption; this has been fixed.
2420 (reported by Rainer Joswig)
2421 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2422 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2423 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2424 arithmetic in general, is now much more reliable on the PPC
2426 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2427 platform now returns the right answer.
2428 * optimization: restored some effective method precomputation in
2429 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2430 precomputation is now tunable.
2431 * optimization: compiler-internal data structure use has been
2432 reviewed, and changes have been made that should improve the
2433 performance of the compiler by about 20%.
2434 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2435 simple-base-strings and simple-bit-vectors is improved.
2436 * optimization: the optimization of 32-bit logical and arithmetic
2437 functions introduced in version 0.8.3 on the x86 has been
2438 implemented on the mips, ppc and sparc platforms; an
2439 implementation of the same facility, but for 64-bit arithmetic,
2440 has been added for the alpha.
2441 * microoptimization: the compiler is better able to make use of the
2442 x86 LEA instruction for multiplication by constants.
2443 * bug fix: in some situations compiler did not report usage of
2444 generic arithmetic in (SPEED 3) policy.
2445 * bug 145b fix: compiler used wrong type specifier while converting
2446 MEMBER-types to numeric.
2447 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2449 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2450 index argument now works properly on the Alpha platform.
2451 * bug fix: floating point exception treatment on the Alpha platform
2453 * bug fix: FILE-POSITION works much better on string input and
2454 output streams. (thanks to Nikodemus Siivola)
2455 * bug fix: many threading/garbage collection symptoms sorted.
2456 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2457 might be pseudo-atomic.
2458 * internal change: Stopping for GC is now done with signals not
2459 ptrace. GC is now done in whichever thread wanted it, instead of
2461 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2463 * bug fix: result form in DO is not contained in the implicit
2465 * incompatible change: ICR structure is changed; the value part of
2466 CONTINUATION is now called LVAR; corresponding functions are
2467 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2468 * added type deriver for ISQRT (thanks to Robert E. Brown).
2469 * bug fix: better support for loading from the command line when an
2470 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2472 * fixed some bugs revealed by Paul Dietz' test suite:
2473 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2474 ** ROUND and FROUND now give the right answer when given very
2475 small float arguments.
2476 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2478 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2479 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2480 ** LCM should return a non-negative integer.
2481 ** PARSE-INTEGER returned the index of a terminator instead of the
2482 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2483 ** PARSE-INTEGER returned an incorrect index being applied to a
2485 ** LCM with two arguments of 0 returns 0 rather than signalling
2487 ** unsigned addition of a 32-bit constant with the high bit set no
2488 longer causes an internal compiler error on the x86.
2489 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2490 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2491 with negative last argument.
2492 ** byte specifiers with zero size and position no longer cause
2493 an error during type derivation.
2494 ** bignum multiplication on the Alpha platform now returns the
2496 * porting: The system now builds on SuSE AMD64, although it still
2497 generates a 32-bit binary.
2498 * .fasl file incompatibility: The fasl file version number has
2499 been incremented (because of the changes to internal compiler
2500 data structures referred to above).
2502 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2503 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2504 more accurately, on the Darwin kernel running on PowerPC hardware.
2505 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2506 * Compiler code deletion notes now signal a condition of type
2507 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2508 an associated MUFFLE-WARNING restart.
2509 * The compiler now performs limited argument count validation of
2510 constant format strings in FORMAT, and where appropriate in ERROR,
2511 CERROR and WARN. (thanks to Gerd Moellmann)
2512 * New ASDF-INSTALL contrib can be used for automatic download and
2513 installation of third-party Lisp code from CCLAN or other sites
2515 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2516 which forces another thread to execute a function supplied by the
2518 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2519 now accept and act upon their :ELEMENT-TYPE keyword argument.
2520 (reported by Martin Atzmueller, Edi Weitz)
2521 * bug fix: FILE-POSITION now accepts position designators up to
2522 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2523 is the greater. (thanks to Patrik Nordebo)
2524 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2525 Kalle Olavi Niemitalo)
2526 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2527 printable. (reported by Eric Marsden)
2528 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2529 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2530 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2531 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2532 (thanks to Stig Sandoe)
2533 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2534 UNREAD-CHAR now correctly includes the unread character in the
2535 target sequence. (thanks to Gerd Moellmann)
2536 * bug fix in threaded builds: the system can now be suspended and
2537 resumed by shell job control with minimal disruption.
2538 * bug fixes in times and timezones >2038AD
2539 * better handling of "where is GNU make?" problem in build scripts
2540 (thanks to Nikodemus Siivola)
2541 * new optimization: inside a named function any reference to a
2542 function with the same name is considered to be a self-reference;
2543 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2544 optimization quality.
2545 * new optimization on x86: logical functions and + now have
2546 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2547 used when the result is truncated to 32 bits.
2548 * VALUES declaration is partially enabled.
2549 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2550 * fixed some bugs revealed by Paul Dietz' test suite:
2551 ** The system now obeys the constraint imposed by
2552 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2553 types form a lattice under type intersection.
2554 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2555 ** ASSOC now ignores NIL elements in an alist.
2556 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2557 and (1+ MOST-POSITIVE-FIXNUM) answers.
2558 ** The addition of a method with invalid qualifiers to a generic
2559 function does not cause an error to be signalled immediately;
2560 a warning is signalled, and the error is generated only on
2561 calling the generic function.
2562 * changed .fasl file version number, in part to add type codes for
2563 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2564 obscure ANSI requirements
2566 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2567 * fixed bug 148: failure to inline-expand a local function left
2568 garbage, confusing the compiler.
2569 * fixed bugs 3cd: structure slot readers perform type check if the
2570 slot can have an invalid value (i.e. it is either not initialized
2571 or can be written with a less specific slot writer).
2572 * bug fix: the compiler now traps array references to elements off
2573 the end of an array; previously, the bounds checking in some
2574 circumstances could go off-by-one.
2575 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2577 * type declarations for array element types now obey the description
2578 on the CLHS page "Declaration TYPE", as per discussions on
2579 sbcl-help around 2003-05-08. This means that a declaration
2580 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2581 declaration, all references to BAR will be asserted or assumed
2582 (with THE, so dependent on compiler policy) to involve objects of
2583 type FOO. Note that no such declaration is implied in
2584 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2585 * declared types of functions from the "Conditions"
2586 chapter. (reported by Paul Dietz)
2587 * bug fix: CERROR accepts a function as its first argument.
2588 * bug fix: NTH an NTHCDR accept a bignum as index
2589 arguments. (reported by Adam Warner)
2590 * optimization: character compare routines now optimize comparing
2591 against a constant character. (reported by Gilbert Baumann)
2592 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2593 argument now works properly on the MIPS platform.
2594 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2595 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2596 declared to have a functional type, does not cause an error
2598 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2599 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2600 * fixed bug: if last continuation of a deleted block has a
2601 destination, this destination should be deleted too. (reported by
2603 * fixed a bug in the bootstrap process: the host compiler's values
2604 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2605 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2606 test case from Patrik Nordebo)
2607 * improved the ability of the disassembler on the PPC platform to
2608 provide helpful disassembly notes.
2609 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2610 the class in more cases than previously.
2611 * bug fix: FILE-POSITION now understands :START and :END for
2612 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2613 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2614 8) data. (thanks to David Lichteblau)
2615 * bug fix: it is possible to add a method to a generic function
2616 without lambda list.
2617 * bug fix: reader failed to signal END-OF-FILE inside an
2618 object representation. (reported by Nikodemus Siivola)
2619 * fixed some bugs revealed by Paul Dietz' test suite:
2620 ** LAST and [N]BUTLAST should accept a bignum.
2621 ** condition slot accessors are methods.
2622 ** (VECTOR NIL) is a subtype of STRING.
2624 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2625 * minor incompatible change: some nonsensical specialized lambda
2626 lists (used in DEFMETHOD) which were previously ignored now signal
2628 * minor incompatible change: the system is now aware of the types of
2629 variables in the COMMON-LISP package, and will signal errors for
2630 most violations of these type constraints (where previously they
2631 were silently accepted).
2632 * minor incompatible change: COMPILE-FILE now uses the freedom
2633 afforded (ANSI 3.2.2.3) to use derived function types for
2634 functions defined in the same file. This also permits the system
2635 to warn on static type mismatches and function
2636 redefinition. (Currently it does not work with high DEBUG level.)
2637 * minor incompatible change: VALUES declaration is disabled.
2638 * When issuing notes, the compiler now signals a condition of type
2639 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2640 restart for use in user handlers. It is expected that the
2641 COMPILER-NOTE condition will eventually become a condition
2642 supertype to a hierarchy of note types, which will then be
2643 handleable in a similar fashion. However, at the moment, no such
2644 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2645 but it's an implementation detail, not a classification for the
2647 * Changes in type checking closed the following bugs:
2648 ** type checking of unused values (192b, 194d, 203);
2649 ** template selection based on unsafe type assertions (192c, 236);
2650 ** type checking in branches (194bc).
2651 * A short form of VALUES type specifier has ANSI meaning (it has
2652 increased the number of situations when SBCL cannot perform type
2654 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2655 keywords or constants is permissible.
2656 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2657 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2658 classes. (thanks to Antonio Martinez)
2659 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2660 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2661 argument to a pretty-print stream. Instead, it leaves any such
2662 operations to DESCRIBE-OBJECT methods.
2663 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2665 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2666 lambda lists. (thanks to David Lichteblau)
2667 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2668 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2669 (thanks to Gerd Moellmann)
2670 * fixed bug 246: increased compilation speed of long
2671 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2673 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2675 * DEFCONSTANT now throws a condition of type
2676 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2677 constant to a non-EQL value; CONTINUE and ABORT restarts
2678 respectively change and preserve the value.
2679 * fixed bug 63: The code walker, part of the implementation of CLOS,
2680 is now better at handling symbol macros.
2681 * bug fix: There is no longer an internal implementation type named
2682 CL:LENGTH. (reported by Raymond Toy)
2683 * bug fix: In macro-like defining macros/special operators the
2684 implicit block does not enclose the lambda list.
2685 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2686 enclosed in parentheses) are not suitable as type specifiers, and
2687 their use properly signals an error now.
2688 * bug fix: An argument count mismatch for a type specifier in code
2689 being compiled no longer causes an unhandled error at compile
2690 time, but signals a compile-time warning.
2691 * fixed simple vector readable printing
2692 * bug fix: DESCRIBE takes more care over whether the class
2693 precedence list slot of a class is bound before accessing it.
2694 (reported by Markus Krummenacker)
2695 * bug fix: FORMATTER can successfully compile pretty-printer format
2696 strings which use variants of the ~* directive inside.
2697 * bug fix: SEARCH now applies its TEST predicate to the elements of
2698 the arguments in the correct order. (thanks to Wolfhard Buss)
2699 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2700 in type inference. (thanks to Robert E. Brown)
2701 * bug fix: Reading in symbols with an explicit package name of ""
2702 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2703 * fixed some bugs revealed by Paul Dietz' test suite:
2704 ** NIL is now allowed as a structure slot name.
2705 ** Arbitrary numbers, not just REALs, are allowed in certain
2706 circumstances in LOOP for-as-arithmetic clauses.
2707 ** Multiple class redefinitions before slot access no longer
2708 causes a type error.
2709 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2710 association between the name and a class.
2711 ** Generic functions with non-standard method-combination and over
2712 five methods all of which return constants no longer return NIL
2713 after the first few invocations. (thanks to Gerd Moellmann)
2714 ** CALL-NEXT-METHOD with no arguments now passes the original
2715 values of the arguments, even in the presence of assignment.
2716 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2718 ** DISASSEMBLE works with closures and funcallable instances.
2719 ** ADD-METHOD now returns the generic function, not the new method.
2720 ** FIND-METHOD signals an error if the lengths of the specializers
2721 is incompatible with the generic function, even if the ERRORP
2723 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2724 which its argument is a member.
2725 ** DEFCLASS only redefines the class named by its class-name
2726 argument if that name is the proper name of the class;
2727 otherwise, it creates a new class.
2728 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2729 of the UNBOUND-SLOT condition to the name of the slot.
2730 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2731 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2732 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2733 SLOT-MAKUNBOUND in the specified fashion.
2735 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2736 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2737 cross-compilation host. As a consequence, we can now bootstrap our
2738 way up to SBCL starting with a bare gcc toolchain and human-readable
2739 source code (first the source to CLISP, then the source to SBCL).
2740 * A contributed module containing a partial implementation of the
2741 simple-streams interface has been included. (thanks to Rudi
2743 * A contributed module implementing the RFC1321 Message Digest
2744 Algorithm, known as MD5, has been included.
2745 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2746 no longer has any effect, as the code controlled by this feature
2747 has been deleted. (As far as we know, no-one has ever built using
2748 this feature, and its semantics were confused in any case).
2749 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2750 work on conditions (as required by the ANSI specification),
2751 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2752 expected behaviour on conditions. Users should note, however,
2753 that such behaviour is not required by the ANSI specification,
2754 and so use of this behaviour may render their code unportable.
2755 * fixed some bugs revealed by Paul Dietz' test suite:
2756 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2758 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2759 ** SXHASH on condition objects no longer returns NIL.
2760 ** :ALLOCATION :CLASS slots are better treated; their values are
2761 updated on class redefinition, and initforms inherited from
2762 superclasses are applied.
2763 ** REMOVE-METHOD returns its generic function argument even when
2764 no method was removed.
2765 ** SHARED-INITIALIZE now initializes the values of the requested
2766 slots, including those with :ALLOCATION :CLASS.
2767 ** ALLOCATE-INSTANCE now works on structure classes defined via
2768 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2770 ** SLOT-EXISTS-P now works on conditions, as well as structures
2772 ** MAKE-LOAD-FORM now has the required methods on
2773 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2774 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2775 keyword, and now implements the SLOT-NAMES argument.
2776 ** methods with &OPTIONAL arguments no longer allow too many
2777 arguments to be passed in the call without error.
2778 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2779 option is consistent with the required arguments of the generic
2780 function lambda list.
2781 * bug fix: REQUIRE accepts a string designator. (Thanks to
2783 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2784 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2785 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2786 * bug fix: The compiler now checks for duplicated variables in macro
2788 * bug fix: SETQ on globals returns the correct value.
2789 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2790 not a COMPILER-ERROR (followed by some other strange error on
2791 choosing the CONTINUE restart).
2792 * bug fix: make.sh and friends are now more consistent in the way that
2793 they look for GNU "make".
2795 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2796 * experimental native threads support (on x86 Linux >=2.4 only).
2797 This is not compiled in by default: you need to add :SB-THREAD to
2798 the target features. See the "Beyond ANSI" chapter of the manual
2800 * fix for longstanding nonANSIism: The old distinction between
2801 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2802 The return value from CL:FIND-CLASS is now a CLOS class, and
2803 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2804 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2805 * An interface to the MetaObject Protocol, as described in Kiczales,
2806 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2807 MIT Press, 1991, is available from the SB-MOP package.
2808 * incompatible change: the SB-PCL package should now be considered
2809 a private implementation detail, and no longer a semi-private MOP
2811 * minor incompatible change: due to rearrangement for threads, the
2812 control stack and binding stack are now allocated at arbitrary
2813 addresses instead of being hardcoded per-port. Users affected by
2814 this probably have to be doing advanced things with shared
2815 libraries, and will know who they are.
2816 * minor incompatible change: Previously, all --eval forms used were
2817 processed with READ before any of them were processed with EVAL.
2818 Now each --eval form is processed with both READ and EVAL before
2819 the next --eval form is processed. (Thus package operations like
2820 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2821 work as the user might reasonably expect.)
2822 * minor incompatible change: *STANDARD-INPUT* is now only an
2823 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2825 * minor incompatible change: Y-OR-N-P is now character-oriented,
2826 not line oriented. Also, YES-OR-NO-P now works without errors.
2827 (thanks to Antonio Martinez)
2828 * sb-aclrepl module improvements: an integrated inspector, added
2829 repl features, and a bug fix to :trace command.
2830 * Known functions, which cannot be open coded by the backend, are
2831 considered to be able to check types of their arguments. (fixing
2832 a bug report by Nathan J. Froyd)
2833 * fixed a bug in computing method discriminating functions: It is
2834 now possible to define methods specialized on classes which have
2835 forward-referenced superclasses. (thanks to Gerd Moellmann)
2836 * fixed evaluation order in optional entries (reported by Gilbert
2838 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2839 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2840 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2841 required-by-AMOP NAME argument, as well as CLASS and
2842 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2843 * fixed bug 20: DEFMETHOD can define methods using names that are
2844 not the proper names of classes to designate class specializers.
2845 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2846 Unix file descriptors, instead of blowing up. (thanks to Antonio
2848 * Garbage collection refactoring: user-visible change is that a
2849 call to the GC function during WITHOUT-GCING will not do garbage
2850 collection until the end of the WITHOUT-GCING. If you were doing
2851 this you were probably losing anyway.
2852 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2853 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2854 TYPEP the latter but not the former.
2855 * The compiler issues a full WARNING for calls to undefined functions
2856 with names from the CL package.
2857 * MAP-INTO for a vector destination is open coded. (reported by
2858 Brian Downing on c.l.l)
2859 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2860 documentation string.
2861 * fixed some bugs revealed by Paul Dietz' test suite:
2862 ** COPY-ALIST now signals an error if its argument is a dotted
2864 ** Condition slots are now accessed more correctly in the presence
2865 of multiple initargs for a given slot.
2866 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2867 exclude restarts of the same name associated with a different
2869 ** DEFCLASS of forward-referenced classes with another
2870 forward-referenced class in the superclasses list no longer
2872 ** Condition slots are now initialized once each, not multiple
2873 times. (thanks to Gerd Moellmann)
2874 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2876 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2877 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2878 arguments contain duplicated elements.
2879 ** RESTART-CASE understands local macros.
2880 ** RESTART-CASE associates exactly its own restarts with a condition.
2881 ** ENDP in safe mode checks its argument to be of type LIST.
2882 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2883 in question is unbound.
2884 ** Optimized MAKE-INSTANCE functions no longer cause internal
2885 assertion failures in the presence of duplicate initargs.
2886 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2887 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2888 propagated even in the ctor optimized implementation of
2890 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2892 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2893 * a better implementation of SXHASH on (simple) bit vectors,
2894 measured both in execution speed and in distribution of results
2895 over the positive fixnums, has been installed. Likewise, a better
2896 implementation of EQUAL for simple bit vectors is now available.
2897 * fixed CEILING optimization for a divisor of form 2^k.
2898 * fixed bug 240 (emitting extra style warnings "using the lexical
2899 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2900 by Antonio Martinez)
2901 * fixed SXHASH, giving different results for NIL depending on type
2902 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2903 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2904 variable. (found by Rolf Wester)
2905 * SBCL does not ignore type declarations for special
2906 variables. (reported by rif on c.l.l 2003-03-05)
2907 * some bug fixes in contrib/sb-aclrepl/
2908 * fixed some bugs revealed by Paul Dietz' test suite:
2909 ** a bug in the CONS type specifier, whereby the CAR and CDR
2910 types got intertwined, has been fixed;
2911 ** the type system is now able to reason about the interaction
2912 between INTEGER and RATIO types more completely;
2913 ** APPEND, [N]REVERSE and NRECONC check that those their
2914 arguments, which must be proper lists, are really so;
2915 ** An array specialized to be unable to hold elements has been
2916 implemented, as required -- yes, really -- by ANSI;
2917 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2918 on malformed property lists;
2920 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2921 * incompatible packaging change: in line with Unix convention,
2922 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2923 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2924 * REQUIRE and PROVIDE are now optionally capable of doing something
2925 useful. See the documentation string for REQUIRE.
2926 * infrastructure for a managed SBCL contrib system: contributed
2927 modules in this release include:
2928 ** the ASDF system definition facility;
2929 ** an interface to the BSD Sockets API;
2930 ** an ACL-like convenience interface to the repl;
2931 (thanks to Kevin Rosenberg)
2932 ** an implementation of ROTATE-BYTE, with efficient implementation
2934 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2935 gives a closer approximation to the right answer than previously.
2936 (thanks to Raymond Toy)
2937 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2938 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2939 optional environment arguments, as required by ANSI.
2940 * fixed bugs in other functions taking environment objects, allowing
2941 calls with an explicit NIL environment argument to be compiled
2943 * fixed bug 228: primary return values from
2944 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2945 COMPILE or FUNCTION.
2946 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2947 :TYPE will no longer signal errors on innocuous objects.
2948 * fixed bug 231b: SETQ is better at respecting type declarations in
2949 the lexical environment.
2950 * fixed a bug in DEFCLASS: classes named by symbols with no or
2951 unprintable packages can now be defined.
2952 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2953 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2954 * fixed compiler failure related to checking types of functions.
2955 (reported by Robert E. Brown)
2956 * the compiler is now much more consistent in its error-checking
2957 treatment of bounding index arguments to sequence functions: in
2958 (SAFETY 3) code, errors will be signalled in almost all cases if
2959 invalid sequence bounding indices are passed to functions defined
2960 by ANSI to operate on sequences.
2961 * fixed a bug in the build procedure: documentation of SBCL-specific
2962 packages is now preserved and available in the final Lisp image.
2963 * lifted FDEFINITION lookup out of loops in the implementation of
2964 many list operations. (thanks to Robert E. Brown)
2965 * fixed a bug in the reader: the #n# reader macro now works for
2966 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2967 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2968 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2969 E. Brown and Tony Martinez respectively)
2970 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2971 SLOT-BOUNDP: in particular, it is now able to optimize them much
2972 better, and is now not vulnerable to having packages renamed.
2973 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2974 always called when a slot is not present in an instance. (thanks
2976 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2977 longer signal an error on unreading a character following EOF on
2978 the previous constituent stream. (thanks to Tony Martinez)
2979 * fixed some bugs revealed by Paul Dietz' test suite:
2980 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2981 not just nonnegative fixnums;
2982 ** the logical bit-array operators such as BIT-AND now accept an
2983 explicit NIL for their "opt-arg" argument (to indicate a
2984 freshly-consed result bit-array);
2985 ** ELT now signals an error on an invalid sequence index in safe
2987 ** the type system is now cleverer about negations of numeric
2988 types, and consequently understands the BIGNUM and RATIO types
2990 ** the type system is now cleverer about the interaction between
2991 INTEGER and RATIO types: while bugs still remain, many more
2992 cases are accurately computed;
2993 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2994 if it is in the last clause;
2995 ** CONSTANTLY now correctly returns a side-effect-free function in
2997 ** DECLARE is no longer treated as a special-operator; in
2998 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2999 * incremented fasl file version number due to the change in the
3000 DEFSTRUCT-SLOT-DESCRIPTION structure.
3002 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3003 * minor incompatible change: code processed by the "interpreter" or
3004 EVAL now has a compilation optimization policy of (DEBUG 2)
3005 (changed from (DEBUG 1)) to improve debuggability of interactive
3006 development, and to allow the use of the debug RETURN command in
3008 * an experimental implementation of the RETURN command for the
3009 debugger has been included. (thanks to Frederik Kuivinen)
3010 * fixed bug 62: constraints were not propagated into a loop.
3011 * fixed bug in embedded calls of SORT (reported and investigated by
3013 * fixed some bugs revealed by Paul Dietz' test suite:
3014 ** printing and reading of arrays with some dimensions having
3015 length 0 (thanks to Gerd Moellmann);
3016 ** BOA constructor with &AUX argument without a default value does
3017 not cause a type error;
3018 ** CONSTANTP now returns true for all self-evaluating objects.
3020 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3021 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3022 accessors that are related by inheritance, as specified in the
3023 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3024 Valtteri Vuorikoski)
3025 * The compiler is now able to inline functions that were defined in
3026 a complex lexical environment (e.g. inside a MACROLET).
3027 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3029 * Support for the upcoming FreeBSD-5.0 release has been included.
3030 (thanks to Dag-Erling Smorgrav)
3031 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3032 effect when it is not in a toplevel context.
3033 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3034 stand a better chance of being correct. (thanks to Gerd
3036 * fixed bug in COERCE, which now signals an error on coercing a
3037 rational to a bounded real type which excludes the expected
3039 * The compiler is now able to derive types more accurately from the
3040 COERCE and COMPILE functions.
3041 * fixed bug 223: functional binding is considered to be constant
3042 only for symbols in the CL package.
3043 * fixed bug 231: SETQ did not check the type of a variable being set
3044 (reported by Robert E. Brown)
3045 * A new optimization for MAKE-INSTANCE has been included, fixing
3046 various bugs (including relating to :ALLOCATION :CLASS slots and
3047 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3049 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3050 ** As required by ANSI, LOOP now disallows anonymous collection
3051 clauses such as COLLECT I in conjunction with aggregate boolean
3052 clauses such as THEREIS (= I 1);
3053 ** LOOP now signals an error when any variable is reused in the
3054 same loop (including the potentially useful construct analogous
3055 to WITH A = 1 WITH A = (1+ A);
3056 ** IT is only a special loop symbol within the first clause of a
3057 conditional loop clause;
3058 ** LOOP with a typed iteration variable over a hashtable now
3059 signals a type error iff it should.
3060 * fixed some other bugs revealed by Paul Dietz' test suite:
3061 ** FILE-STREAM now names the class previously known as FD-STREAM;
3062 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3063 argument) no longer signals an error;
3064 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3065 of the slot symbol, rather than using the current package
3066 ((:CONC-NAME "") continues to intern the slot's name in the
3068 * incremented fasl file version number, because of the incompatible
3069 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3070 of the new implementation of DEFINE-COMPILER-MACRO.
3072 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3073 * Support for building SBCL for MIPS platforms running in
3074 little-endian mode has now been checked in, and basic
3075 functionality on said platforms verified.
3076 * minor incompatible change: PCL now records the pathname of a file
3077 in which methods and the like are defined, rather than its
3079 * minor incompatible change: TRUENAME now considers the truename of
3080 a file naming a directory to be the pathname with :DIRECTORY
3081 component indicating that directory.
3082 * minor incompatible change: a NAMED clause in the extended form of
3083 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3084 reason for the previous behaviour is unclear.
3085 * more systematization and improvement of CLOS and MOP conformance
3086 in PCL (thanks to Gerd Moellman and Pierre Mai):
3087 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3089 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3090 primary methods with no specializers;
3091 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3093 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3094 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3095 FORWARD-REFERENCED-CLASSes; error reporting on
3096 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3098 ** SXHASH on CLOS instances now uses a slot internal to the
3099 instance to return different numbers on distinct instances,
3100 while preserving the same return value through invocations of
3102 ** DEFMETHOD signals errors when methods with longer incongruent
3103 lambda lists are added to generic functions;
3104 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3105 CLASS, as specified in AMOP;
3106 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3107 on the order returned by the primary method for classes of
3108 class STANDARD-CLASS;
3109 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3110 * fixed some bugs shown by Paul Dietz' test suite:
3111 ** DOLIST puts its body in TAGBODY;
3112 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3114 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3115 value producing form;
3116 ** if more variables are given to PROGV than values, extra
3117 variables are bound and made to have no value;
3118 ** NSUBSTITUTE on list arguments gets the right answer with
3120 ** ELT signals an error of type TYPE-ERROR when the index argument
3121 is not a valid sequence index;
3122 ** LOOP signals (at macroexpansion time) an error of type
3123 PROGRAM-ERROR when duplicate variable names are found;
3124 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3125 ** FUNCALL of special-operators now cause an error of type
3127 ** PSETQ now works as required in the presence of side-effecting
3128 symbol-macro places;
3129 ** NCONC accepts any object as its last argument;
3130 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3132 ** loop-for-as-package does not require a package to be explicitely
3134 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3136 * fixed bug 166: compiler preserves "there is a way to go"
3137 invariant when deleting code.
3138 * fixed bug 172: macro lambda lists with required arguments after
3139 &REST arguments now cause an error to be signalled. (thanks to
3141 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3142 bug. (thanks to Matthew Danish)
3143 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3145 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3146 arguments are lexically rebound. (thanks to Gerd Moellmann and
3148 * fixed bug 194: error messages are now more informative when there
3149 is no primary method applicable in a call to a generic
3150 function. (thanks to Gerd Moellmann)
3151 * fixed bug in command line argument checking (thanks to Julian
3153 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3154 (thanks to Matthew Danish)
3155 * incremented fasl file version number, because of the
3156 SXHASH-related changes in the layout of CLOS data structures
3158 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3159 * minor incompatible change: The runtime (the Unix executable named
3160 "sbcl") is now much pickier about the .core files it will load.
3161 Essentially it now requires .core files to descend from the same
3162 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3163 as the runtime does. (The intent is to prevent the crashes which
3164 can occur, and which can even be reported as mysterious failures,
3165 when people patch the sources or change the build parameters
3166 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3167 sbcl and .core files.)
3168 * fixed bug: VALUES-LIST is no longer optimized away.
3169 * fixed bug 142: The FFI conversion of C string values to Lisp
3170 string values no longer conses excessively. (thanks to Nathan
3171 Froyd porting Raymond Toy's fix to CMU CL)
3172 * began to systematize and improve MOP conformance in PCL (thanks to
3173 Nathan Froyd, Gerd Moellman and Pierre Mai):
3174 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3176 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3177 ** FINALIZE-INHERITANCE is now called on class finalization;
3178 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3179 argument precedence order.
3180 * fixed bug 202: The compiler no longer fails on functions whose
3181 derived types contradict their declared type.
3182 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3183 so it can be non-toplevel.
3184 * The fasl file version number has changed (because of the new
3185 implementation of DEFMACRO).
3186 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3187 safe code, that any length requirement by their type-specifier
3188 argument is valid. The exceptions to this are described in bug
3190 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3191 functions now check the types of their inputs as required by ANSI.
3192 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3193 introduced symbol is DECLAREd to be SPECIAL.
3194 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3195 * fixed bug: :COUNT argument to sequence functions may be negative
3196 * fixed bug: body of DO-SYMBOLS may contain declarations
3197 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3198 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3199 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3200 symbol macro only once
3201 * fixed printing of call frame when argument list is unavailable
3202 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3203 * compiler no longer signals WARNING on unknown keyword
3206 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3207 * A beta-quality port to the mips architecture running Linux,
3208 based on the old CMUCL backend, has been made. It has been tested
3209 on a big-endian kernel, and works sufficiently well to be able to
3210 rebuild itself; it has not been tested in little-endian mode.
3211 * fixed an inconsistency between gencgc.c and purify.c which made
3212 dumping/loading .core files unreliable
3213 * fixed bug 120a: The compiler now deals correctly with IFs where
3214 the consequent is the same as the alternative, instead of
3215 misderiving the return type. (thanks to Alexey Dejneka)
3216 * fixed bug 113: Logical pathnames are now dumpable (the logical
3217 host is resolved at load-time, throwing an error if it is not
3219 * fixed bug 174: FORMAT's error message is slightly clearer when a
3220 non-printing character is used in a format directive.
3221 * fixed several bugs in compiler checking of type declarations, i.e.
3222 violations of the Python "declarations are assertions" principle
3223 (thanks to Alexey Dejneka)
3224 * fixed several bugs in PCL's error checking (thanks to Gerd
3226 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3228 * fixed bug in compilation of functions as first class values
3229 (thanks to Antonio Martinez-Shotton)
3230 * The compiler's handling TYPE-ERRORs which it can prove will
3231 inevitably happen at runtime has been cleaned up and corrected
3232 in several ways. (thanks to Alexey Dejneka)
3233 * improved argument type checking for various basic arithmetic
3234 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3235 lavished on them in the past that they can be compiled in many
3236 ways in different special cases
3237 * fixed bug 181: compiler checks validity of user supplied type
3239 * cleaned up code flushing in optimization: Function calls which
3240 should signal errors for safety purposes (e.g. which ANSI says
3241 should signal errors when their arguments are of incorrect type)
3242 are no longer optimized away.
3243 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3244 * incremented fasl file version number, because changes in the
3245 implementation of sequence functions like COERCE caused
3246 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3247 in old inline expansions) to become undefined. (Actually these
3248 changes were later undone, so we might very well be binary
3249 compatible with 0.7.7 after all, but leaving the version number
3250 incremented seemed like the simplest and most conservative
3253 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3254 * An alpha-quality port to the parisc architecture running Linux,
3255 based on the old CMUCL backend, has been made. This, even more so
3256 than the other backends, should be considered still a work in
3257 progress; known problems include that the Linux kernel in 64-bit
3258 mode does not propagate the correct sigcontext structure to
3259 userspace, and consequently SBCL on a parisc64 kernel will not
3261 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3262 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3263 suppressed.) Also now that the compiler is looking at declarations
3264 in the environment, it checks optimization declarations as well,
3265 and suppresses inlining when (> DEBUG SPEED).
3266 * More fixes have been made to treatment of floating point exception
3267 treatment and other Unix signals. In particular, floating point
3268 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3269 * The detection and handling of control stack exhaustion (infinite
3270 or very deeply nested recursion) has changed. Stack exhaustion
3271 detection is now done by write-protecting pages at the OS level
3272 and applies at all optimization settings; when found, a
3273 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3274 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3275 be caught using IGNORE-ERRORS.
3276 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3277 correctly when transforming recursive LABELS functions to LETs.
3278 (thanks to Alexey Dejneka)
3279 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3280 that are names of constants or global variables.
3281 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3282 alien routines with docstrings.
3283 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3284 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3286 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3287 (thanks to Eric Marsden)
3288 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3289 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3290 * Bug fix: The SPARC backend can now compile functions involving
3291 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3292 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3293 to a functions expecting a PATHNAME-DESIGNATOR.
3294 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3295 lambda lists. (thanks to Alexey Dejneka)
3296 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3297 functions so that EXPORTing the name of the class doesn't cause
3298 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3299 to Antonio Martinez for reporting this)
3300 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3301 constant-folded without failing an assertion. (thanks to Einar
3302 Floystad Dorum for reporting this)
3303 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3304 types are now dealt with more correctly by the compiler.
3305 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3306 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3307 behaviour analogous to RENAME-FILE. This puts its behaviour more
3308 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3310 * The fasl file version number has changed again. (because of the
3311 bug fix involving the names of PCL MAKE-INSTANCE functions)
3313 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3314 * bug fix: Floating point exceptions are treated much more
3315 consistently on the x86/Linux and PPC/Linux platforms.
3316 * Array initialization with :INITIAL-ELEMENT is now much faster for
3317 cases when the compiler cannot open code the array creation, but
3318 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3319 array accesses have also seen a speed increase.
3320 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3321 specified. (This was at the root of some bad interactions between
3322 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3324 * bug fix: Internal error arguments for undefined functions are now
3325 computed correctly on the PPC/Linux platform.
3326 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3327 Raymond Toy's patch for CMU CL)
3328 * Support for the Solaris 9 operating environment has been included
3329 (thanks to Daniel Merritt)
3330 * A very ugly but hopefully complete draft of the missing FFI chapter
3331 of the manual has been created by reformatting the corresponding
3332 CMU CL manual chapter into (currently very ugly and incoherent)
3333 DocBook and bringing it up to date for SBCL behavior. Thus, the
3334 manual is now essentially complete, at least by my extreme
3335 once-and-only-once standards, whereby it's acceptable to refer to
3336 the doc strings of SB-EXT functions as the primary documentation.
3337 * The fasl file version number has changed again, due to cleanup of
3338 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3339 is no longer a static symbol.)
3341 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3342 * SBCL now builds with OpenMCL (version 0.12) as the
3343 cross-compilation host; also, more progress has been made toward
3344 bootstrapping under CLISP.
3345 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3347 * bug 158 fixed: The compiler can now deal with integer loop
3348 increments different from 1; fixing this turned out also to fix
3350 * bug 169 fixed: no more bogus warnings about using lexical bindings
3351 despite the presence of perfectly good SPECIAL declarations (thanks
3352 to David Lichteblau)
3353 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3354 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3355 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3357 * bug 180 fixed: Method combination specifications no longer ignore
3358 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3359 * bug fix: Structure type predicate functions now check their argument
3360 count as they should.
3361 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3362 correctly. (thanks to Pierre Mai)
3363 * minor incompatible change: The --noprogrammer option is deprecated
3364 in favor of the new --disable-debugger option, which is very similar.
3365 (The major difference is that it takes effect at a slightly different
3366 time at startup, causing handling of errors in --sysinit and
3367 --userinit files will be affected differently.) The
3368 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3369 been added to allow this functionality to be controlled from ordinary
3370 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3371 maintainers, who might want to run non-interactive scripts to
3372 build SBCL cores which will later be used interactively.)
3373 * minor incompatible change: The LOAD function no longer, when given
3374 a wild pathname to load, loads all files matching that pathname.
3375 Instead, an error of type FILE-ERROR is signalled.
3377 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3378 * bug 147 fixed: The compiler preserves its block link/count
3379 invariants more correctly now so that it doesn't crash. (thanks
3381 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3383 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3384 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3385 no-value-for-this-character as NIL (as specified by ANSI).
3386 * HOST-NAMESTRING on physical pathnames now returns a string that is
3387 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3388 (thanks to Christophe Rhodes)
3389 * The Alpha port handles icache flushing more correctly. (thanks to
3391 * More progress has been made toward bootstrapping under CLISP. (thanks
3392 to Christophe Rhodes)
3393 * The fasl file format has changed again, because dynamic loading
3394 on OpenBSD (which has non-ELF object files) motivated some cleanups
3395 in the way that foreign symbols are transformed and passed around.
3396 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3397 is no longer treated as whitespace by the reader, but instead as
3398 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3399 |A
\7fB|, instead of A as it used to.
3401 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3402 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3403 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3404 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3405 this as of 0.7.1.45, but was left out of the previous news section
3406 (thanks to Dan Barlow)
3407 * SBCL now runs on the Solaris operating system on SPARC architectures
3408 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3409 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3410 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3411 * SPARC backend cleanups, allowing builds of cores optimized for V8
3412 and V9 SPARCS, and also emission of code targeted to a particular
3413 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3415 * SBCL is closer to bootstrapping under CLISP, thanks to various
3416 fixes by Christophe Rhodes.
3417 * The fasl file format has changed again, to allow the compiler's
3418 INFO database to support symbol macros.
3419 * The user manual (in doc/) is formatted into HTML more nicely.
3420 (thanks to coreythomas)
3421 * The system is smarter about SUBTYPEP relationships, especially
3422 those involving NOT types (including types such as ATOM which are
3423 represented internally using NOT types). Thus SUBTYPEP is less
3424 likely to return (VALUES NIL NIL) in general, and in particular
3425 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3427 * The fasl file format has changed again, because the internal
3428 representation of types now includes a new slot to support the new
3429 SUBTYPEP-of-NOT-types logic.
3430 * (not a change in the main branch of SBCL, but a related prototype
3431 which can hopefully be merged into the main branch of SBCL in the
3432 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3433 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3434 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3435 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3436 dispatch macro character combinations. (thanks to Alexey Dejneka)
3437 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3438 that superficially look like logical namestrings correctly.
3439 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3441 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3442 default supplied-p) for &optional and &key arguments. (thanks to
3445 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3446 * incompatible change: The compiler is now less aggressive about
3447 tail call optimization, doing it only when (> SPACE DEBUG) or
3448 (> SPEED DEBUG). (This is an incompatible change because there are
3449 programs which relied on the old CMU-CL-style behavior to optimize
3450 away their unbounded recursion which will now die of stack overflow.)
3451 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3452 for non-GENCGC systems has been increased to 20M (since that
3453 seems much closer to the likely performance optimum for modern
3454 systems than the old 4M value was)
3455 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3456 and *DEBUG-PRINT-LEVEL*
3457 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3458 of CMU CL's support for SPARC, and various endianness and other
3459 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3460 * new syntactic sugar for the Unix command line: --load foo.bar is now
3461 an alternate notation for --eval '(load "foo.bar")'.
3463 ** The system now detects stack overflow and handles it gracefully,
3464 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3465 optimization settings. (This is a good thing in general, and
3466 its introduction in this version should be particularly timely
3467 for anyone whose code fails because of suppression of tail
3469 ** The system now hunts for the C variable "environ" in a more
3470 devious way, to avoid segfaults when the C library version differs
3471 between compile time and run time. (thanks to Christophe Rhodes)
3472 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3473 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3474 ** The compiler no longer issues bogus style warnings for undefined
3475 classes in the same source file as the DEFCLASSes which defined
3476 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3478 ** fixes in CONDITION class precedence list for undefined function
3479 errors (thanks to Alexei Dejneka)
3480 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3481 correctly. (thanks to Dan Barlow)
3482 ** portability fixes aiming at bootstrapping under CLISP (thanks
3483 to Dave McDonald and Christophe Rhodes)
3484 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3485 ** fixes in type translation and and type inference (thanks to
3487 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3488 ** various fixes in the new ports (thanks to Dan Barlow)
3489 * several changes related to debugging:
3490 ** suppression of tail recursion, as noted above
3491 ** stack overflow detection, as noted above
3492 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3493 is now the default. (For some time encapsulation has been more
3494 reliable than the breakpoint-based :ENCAPSULATE NIL
3495 implementation, at least on X86 systems; and I just noticed that
3496 encapsulation also seems closer to the spirit of the ANSI
3499 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3501 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3502 up properly again. (There was a packaging bug in 0.7.0 which
3503 left their definitions in SB-SYS::LOAD-FOREIGN and
3504 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3505 vital for most things which interface to C-level interfaces,
3506 like extensions working with sockets or databases or
3507 Perl-compatible regexes or whatever, and the need to fix
3508 this bug is the main reason that 0.7.1 was released so
3510 ** DEFGENERIC is now choosier about the methods it redefines, so that
3511 reLOADing a previously-LOADed file containing DEFGENERICs does
3512 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3513 cycle now works as it should. (thanks to Alexey Dejneka)
3514 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3515 fixed by Christophe Rhodes. (He actually submitted this patch
3516 months ago, and I delayed until after 0.7.0.)
3517 ** Bug 111 (internal compiler confusion about runtime checks on
3518 FUNCTION types) was fixed by Alexey Dejneka.
3519 * Some internal cleanups (getting rid of variables which aren't
3520 needed now that the byte interpreter is gone) caused the fasl
3521 file format number to change again.
3523 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3524 * major incompatible change: The default fasl file extension, i.e. the
3525 default extension for files produced by COMPILE-FILE, has changed
3526 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3528 ** There are many changes in the implementation of the compiler.
3529 SBCL is now essentially a compiler-only implementation of ANSI
3530 Common Lisp. EVAL still "interprets" a few special cases, but
3531 almost all the interesting cases are handled by creating
3532 a LAMBDA expression, calling COMPILE on it, then calling
3533 FUNCALL on the result.
3534 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3535 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3536 Since the code is newer, there might still be some new bugs
3537 (though not as many as before Martin Atzmueller's fixes:-). But
3538 the new code is substantially simpler and clearer, and hopefully
3539 any remaining bugs will be simpler, less fundamental, and more
3540 fixable then the bugs in the old code.
3541 ** The revised compiler is still a little unsteady on its feet.
3543 *** The debugging information it produces (particularly the names
3544 of FUNCTION objects) is sometimes much less useful than what
3545 the old compiler produced.
3546 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3547 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3548 within a MACROLET) has been temporarily weakened.
3549 ** There are new compiler optimizations for various functions:
3550 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3551 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3552 *** the math functions TRUNCATE, FLOOR, and CEILING
3553 *** the function-of-all-trades COERCE
3554 Mostly these should be transparent, but there's one
3555 potentially-annoying problem (bug 117): when the compiler
3556 inline-expands a function and does type analysis on the result,
3557 it can create control paths which have type mismatches, and
3558 when it can't prove that those control paths aren't taken,
3559 it will issue WARNINGs about the type mismatches. This is
3560 a particular problem in practice for the new sequence functions.
3561 It's not clear how this should be fixed, and for now, a
3562 workaround is given in the entry for 117 in the BUGS file.
3563 ** (Because of the interaction between the two previous items --
3564 occasional inlining problems and new inline expansions -- some
3565 of the new sequence function optimizations won't really kick in
3566 completely until debugging information, and then inlining, are
3567 straightened out in some future version.)
3568 * minor incompatible changes:
3569 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3570 DIRECTORY no longer implicitly promotes NIL slots of its
3571 pathname argument to :WILD. In particular, when you ask for the
3572 contents of a directory (which you used to be able to do without
3573 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3574 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3575 ** changes in behavior that ANSI explicitly defines to be
3576 implementation dependent:
3577 *** The new compiler-only implementation still conforms with ANSI,
3578 but acts a little different than before. Besides the obvious
3579 changes in performance tradeoffs (that the cost per form passed
3580 to EVAL has gone up, and the cost per form executed by EVAL
3581 has gone down), the behavior of the system changes a little
3582 because there are no longer any interpreted function objects.
3583 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3584 e.g. doing COMPILE on the output of interactive DEFUN is
3586 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3588 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3589 has changed from (:CL) to NIL.
3590 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3591 e.g. #\Newline and #\Tab, don't have names specified in the
3592 ANSI Common Lisp standard, is now based on their ASCII symbolic
3593 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3594 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3595 are no longer used for output.
3596 ** changes in internal implementation constants:
3597 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3598 4 million. (If your application spends a lot of time GCing and
3599 you have a lot of RAM, you might want to experiment with
3600 increasing it even more.)
3601 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3602 However, almost all old code should still continue to work without
3603 immediate update, as SB-C-CALL is now a (deprecated) nickname
3605 ** Old operator names in the style DEF-FOO are now deprecated in
3606 favor of new corresponding names DEFINE-FOO, for consistency with
3607 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3608 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3609 internal symbols, but a few supported extensions like
3610 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3611 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3612 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3613 etc. as you get deeper into recursive calls to the debugger
3614 command loop, instead of the old "5]", "5]]", "5]]]"
3615 sequence. (I was motivated to do this when squabbles between
3616 ILISP and SBCL left me very deeply nested in the debugger. In the
3617 short term, this change will probably provoke more ILISP/SBCL
3618 squabbles, but hopefully it will be an improvement in the long run.)
3619 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3620 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3621 function, since declaiming return types involving aliens is
3622 (1) annoyingly messy to do by hand and (2) vital to efficient
3623 compilation of code which calls such functions.
3624 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3625 longer reexported by the SB-EXT package. They're solely useful
3626 for alien code, so it seems more logical that you should get
3627 them from the SB-ALIEN package, not in SB-EXT.
3628 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3629 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3630 features. Instead, the code that they used to control is always
3631 built into the system.
3632 * many other bug fixes
3633 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3634 advantage of the new EVAL-WHEN stuff and to clean them up in
3635 general, and they are now more ANSI-compliant in a number of
3636 ways. Martin Atzmueller is responsible for a lot of this.
3637 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3639 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3640 *** correct ERROR type for various file operations
3641 *** some fixes for Lisp streams
3642 *** DEFMETHOD syntax checking
3643 *** changing old weird representation of debug information as
3644 strings (which, among their other deficiencies, don't transform
3645 correctly when you rename packages, and don't change their
3646 print representation when you change things like *PACKAGE*
3647 and *PRINT-LENGTH*) to symbols and lists of symbols
3648 He also made several improvements and fixed several bugs in DESCRIBE.
3649 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3651 *** misbehavior of WRITE-STRING/WRITE-LINE
3652 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3653 and several other LOOP problems as well
3654 *** DIRECTORY when similar filenames are present
3655 *** DEFGENERIC with :METHOD options
3656 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3657 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3658 *** argument ordering in FIND with :TEST option
3659 *** mishandled package designator argument in APROPOS-LIST
3660 *** various problems in the backquote readmacro
3661 *** a bug in APROPOS
3662 *** probably some others that I'm not describing very well here,
3663 since the CVS log documents them by reference to sbcl-devel
3664 messages, and the SourceForge archives aren't working well.:-(
3665 ** Dan Barlow improved the Alpha port (and is making progress on the
3666 PPC port, for those of you who think different).
3667 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3668 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3669 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3670 continues to work on the SPARC port (for those of you in a position
3671 to look down upon our little PC-compatible boxes from a great height).
3672 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3673 and uses that copy, rather than the current dynamic value, when
3674 it's trying to decide whether to truncate output. Thus e.g.
3675 (let ((*print-lines* 50))
3676 (pprint-logical-block (stream nil)
3678 (let ((*print-lines* 8))
3679 (print (aref possiblybigthings i) stream)))))
3680 should now truncate the logical block only at 50 lines, instead of
3681 often truncating it at 8 lines, as it did before.
3682 * The doc/cmucl/ directory, containing old CMU CL documentation
3683 from the time of the fork, is no longer part of the base system.
3684 SourceForge has shut down its anonymous FTP service, and with it
3685 my original plan for distributing the old CMU CL documentation
3686 there. For now, if you need these files you can download an old
3687 SBCL source release and extract them from it.
3688 * The fasl file version number changed again, for dozens of reasons,
3689 some of which are apparent above.
3691 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3692 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3693 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3694 patch, so that DISASSEMBLE output is much nicer.
3695 * The code in the SB-PROFILE package now seems reasonably stable.
3696 I still haven't decided what the final interface should look like
3697 (I'd like PROFILE to interact cleanly with TRACE, since both
3698 facilities use function encapsulation) but if you have a need
3699 for profiling now, you can probably use it successfully with
3700 the current CMU-CL-style interface.
3701 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3702 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3703 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3704 link now returns the dangling link itself, and for similar
3705 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3706 link itself. (In these cases the old code signalled an error and
3707 looped endlessly, respectively.) Thus, DIRECTORY now works even
3708 in the presence of dangling and cyclic symbolic links.
3709 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3710 is now a supported extension again, since the consensus on
3711 sbcl-devel was that it can be useful for ordinary development
3712 work, not just for debugging SBCL itself.
3713 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3714 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3715 that currently-defined functions might be redefined later with
3716 different return types.
3717 * Hash tables can be printed readably, as inspired by CMU CL code
3718 of Eric Marsden and SBCL code of Martin Atzmueller.
3719 * better error handling in CLOS method combination, thanks to
3720 Martin Atzmueller porting Pierre Mai's CMU CL patches
3721 * more overflow fixes for >16Mbyte I/O buffers
3722 * A bug in READ has been fixed, so that now a single Ctrl-D
3723 character suffices to cause end-of-file on character streams.
3724 In particular, now you only need one Ctrl-D at the command
3725 line (not two) to exit SBCL.
3726 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3728 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3729 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3730 does the right thing.
3731 * The implementation of some type tests, especially for CONDITION
3732 types, is now tidier and maybe faster, due to CMU CL code
3733 originally by Douglas Crosher, ported by Martin Atzmueller.
3734 * Some math functions have been fixed, and there are new
3735 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3736 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3737 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3738 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3739 likely to be built into sbcl-0.7.0.)
3740 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3741 CL, ported to SBCL by Martin Atzmueller)
3742 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3743 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3744 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3745 * There's a new slam.sh hack to shorten the edit/compile/debug
3746 cycle for low-level changes to SBCL itself, and a new
3747 :SB-AFTER-XC-CORE target feature to control the generation of
3748 the after-xc.core file needed by slam.sh.
3749 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3750 COMPILE-FILE is no longer supported, so that now every function
3751 gets an entry point, so that block compilation looks a little
3752 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3753 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3754 returns the number of bytes consed since the system started,
3755 rather than the number consed since the first time the function
3756 was called. (The new definition parallels ANSI functions like
3757 CL:GET-INTERNAL-RUN-TIME.)
3758 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3759 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3760 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3761 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3762 * Fasl file version numbers are now independent of the target CPU,
3763 since historically most system changes which required version
3764 number changes have affected all CPUs equally. Similarly,
3765 the byte fasl file version is now equal to the ordinary
3768 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3769 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3770 is no longer recognized. I apologize for this, because it was
3771 listed in SB-EXT as a supported extension, but I found that
3772 its existing behavior was poorly specified, as well as incorrectly
3773 specified, and it looked like too much of a mess to straighten it
3774 out. I have enough on my hands trying to get ANSI stuff to work..
3775 * many patches ported from CMU CL by Martin Atzmueller, with
3776 half a dozen bug fixes in pretty-printing and the debugger, and
3777 half a dozen others elsewhere
3778 * fixed bug 13: Floating point infinities are now supported again.
3779 They might still be a little bit flaky, but thanks to bug reports
3780 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3781 as flaky as they were.
3782 * The --noprogrammer command line option is now supported. (Its
3783 behavior is slightly different in detail from what the old man
3784 page claimed it would do, but it's still appropriate under the
3785 same circumstances that the man page talks about.)
3786 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3787 are now supported, and enabled by default. Thus, the compiler can
3788 handle many floating point and complex operations much less
3789 inefficiently. (Thus e.g. you can implement a complex FFT
3791 * The compiler now detects type mismatches between DECLAIM FTYPE
3792 and DEFUN better, and implements CHECK-TYPE more correctly, and
3793 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3794 thanks to patches from Martin Atzmueller.
3795 * various fixes to make the cross-compiler more portable to
3796 ANSI-conforming-but-different cross-compilation hosts (notably
3797 Lispworks for Windows, following bug reports from Arthur Lemmens)
3798 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3799 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3800 fixed thanks to Pierre Mai's CMU CL patches.
3801 * fixes to keep the system from overflowing internal counters when
3802 it tries to use i/o buffers larger than 16M bytes
3803 * fixed bug 45a: Various internal functions required to support
3804 complex special functions have been merged from CMU CL sources.
3805 (When I was first setting up SBCL, I misunderstood a compile-time
3806 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3807 * improved support for type intersection and union, fixing bug 12
3808 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3809 more obscure bugs as well
3810 * some steps toward byte-compiling non-performance-critical
3811 parts of the system, courtesy of patches from Martin Atzmueller
3812 * Christophe Rhodes has made some debian packages of sbcl at
3813 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3814 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3815 stable, but are nonetheless usable. When he's ready, I'd be happy
3816 to add them to the SourceForge "File Releases" section. (And if
3817 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3818 * new fasl file format version number (because of changes in
3819 internal representation of (OR ..) types to accommodate the new
3820 support for (AND ..) types, among other things)
3822 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3823 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3825 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3827 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3828 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3829 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3830 * DESCRIBE now works on structure objects again.
3831 * Most function call argument type mismatches are now handled as
3832 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3833 know whether the function will be redefined before the call is
3834 executed. (The compiler could flag local calls with full WARNINGs,
3835 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3836 it doesn't keep track of enough information to know whether calls
3837 are local in this sense.)
3838 * Compiler output is now more verbose, with messages truncated
3839 later than before. (There should be some supported way for users
3840 to override the default verbosity, but I haven't decided how to
3841 provide it yet, so this behavior is still controlled by the internal
3842 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3843 src/compiler/ir1util.lisp.)
3844 * Fasl file format version numbers have increased again, because
3845 support for the Gray streams extension changes the layout of the
3846 system's STREAM objects.
3847 * The Gray subclassable streams extension now works, thanks to a
3848 patch from Martin Atzmueller.
3849 * The full LOAD-FOREIGN extension (not just the primitive
3850 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3851 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3852 but like most other programs, it defaults to copying the Unix
3853 environment from the original process instead of starting the
3854 new process in an empty environment.
3855 * Extensions which manipulate the Unix environment now support
3856 an :ENVIRONMENT keyword option which doesn't smash case or
3857 do other bad things. The CMU-CL-style :ENV option is retained
3858 for porting convenience.
3859 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3860 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3862 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3864 * A patch from Martin Atzmueller seems to have solved the SIGINT
3865 problem, and as far as we know, signal-handling now works cleanly.
3866 (If you find any new bugs, please report them!)
3867 * The system no longer defaults Lisp source file names to types
3868 ".l", ".cl", or ".lsp", but only to ".lisp".
3869 * The compiler no longer uses special default file extensions for
3870 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3871 seems to expect a single default extension for all compiled code,
3872 and there's no compelling reason to try to stretch the standard
3873 to allow two different extensions.) Instead, byte-compiled files
3874 default to the same extension as native-compiled files.
3875 * Fasl file format version numbers have increased again, because
3876 a rearrangement of internal implementation packages made some
3877 dumped symbols in old fasl files unreadable in new cores.
3878 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3879 many fewer weird special cases.
3880 * Bug #17 (differing COMPILE-FILE behavior between logical and
3881 physical pathnames) has been fixed, and some related misbehavior too,
3882 thanks to a patch from Martin Atzmueller.
3883 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3884 by Tim Moore, ported to SBCL by Martin Atzmueller.
3885 * Martin Atzmueller fixed several filesystem-related problems,
3886 including bug #36, in part by porting CMU CL patches, which were
3887 written in part by Paul Werkowski.
3888 * More compiler warnings in src/runtime/ are gone, thanks to
3889 more patches from Martin Atzmueller.
3890 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3893 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3895 * DESCRIBE now works on CONDITION objects.
3896 * The debugger now handles errors which arise when trying to print
3897 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3899 * The build system now uses an additional file, customize-target-features.lisp,
3900 to allow local modifications to the target *FEATURES* list. (The point of
3901 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3902 debugging features enabled, without having to worry about propagating it
3903 into everyone's system when I do a "cvs update".) When no
3904 customize-target-features.lisp file exists, the target *FEATURES* list
3905 should be constructed the same way as before.
3906 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3907 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3908 warn about behavior which is undefined under the ANSI spec. Note: This
3909 is specified by ANSI, but it's not very popular with programmers.
3910 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3911 macro in the SBCL sources for an example of a workaround which you
3912 might use to make portable ANSI-standard code which does what you want.
3913 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3914 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3915 when it's used as a non-toplevel form. (This is required in order
3916 to implement the DEFCONSTANT-EQX macro.)
3917 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3918 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3919 funny things, due to bugs in the implementation of EVAL-WHEN
3920 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3921 non-toplevel DEFCONSTANTs are uncommon.)
3922 * The core file version number and fasl file version number have been
3923 incremented, because the old noncompliant DEFCONSTANT behavior involved
3924 calling functions which no longer exist, and because I also took the
3925 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3926 * fixed bug 1 (error handling before read-eval-print loop starts), and
3927 redid debugger restarts and related debugger commands somewhat while
3929 ** The QUIT debugger command is gone, since it did something
3930 rather different than the SB-EXT:QUIT command, and since it never
3931 worked properly outside the main toplevel read/eval/print loop.
3932 Invoking the new TOPLEVEL restart provides the same functionality.
3933 ** The GO debugger command is also gone, since you can just invoke
3934 the CONTINUE restart directly instead.
3935 ** The TOP debugger command is also gone, since it's redundant with the
3936 FRAME 0 command, and since it interfered with abbreviations for the
3938 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3940 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3941 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3942 some porting work by Martin Atzmueller)
3943 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3944 with a boatload of patches which he ported to SBCL. Now that those
3946 ** The system tries to make sure that its low-priority messages
3947 are prefixed by semicolons, to help people who like to use
3948 syntax highlighting in their ILISP buffer. (This patch
3949 was originally due to Raymond Toy.)
3950 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3951 patches originally written by Raymond Toy.
3952 ** The compiler understands coercion between single-value and
3953 multiple-VALUES type expressions better, getting rid of some very
3954 weird behavior, thanks to patches originally by Robert MacLachlan
3955 and Douglas Crosher.
3956 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3957 lambda lists, thanks to a patch originally by Pierre Mai.
3958 ** The system no longer bogusly warns about "abbreviated type
3960 ** The compiler gets less confused by inlining and RETURN-FROM,
3961 thanks to some patches originally by Tim Moore.
3962 ** The system no longer hangs when dumping circular lists to fasl
3963 files, thanks to a patch originally from Douglas Crosher.
3964 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3965 undefined function error.
3966 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3967 in structure slot initforms) for now, documented workaround instead:-|
3968 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3969 * fixed bug 5: added stubs for various Gray stream functions called
3970 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3971 installed, at least appropriate type errors are generated
3972 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3973 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3974 consistently in DEFMETHOD forms.
3975 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3976 it doesn't seem to affect SBCL after all
3977 * The C runtime system now builds with better optimization and many
3978 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3980 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3982 * The system is now under CVS at SourceForge (instead of the
3983 CVS repository on my home machine).
3984 * The new signal handling code has been tweaked to treat register
3985 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3986 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3987 downstream, e.g. in debug-int.lisp, has implicit dependencies
3988 on the unsignedness of integer representation of machine words,
3989 and that caused the system to bomb out with infinite regress
3990 when trying to recover from type errors involving signed values,
3991 e.g. (BUTLAST '(1 2 3) -1).)
3992 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3993 (This was one of the bugs Peter Van Eynde reported back in July.)
3994 * The system now uses code inspired by Colin Walters' O(N)
3995 implementation of MAP (from the cmucl-imp@cons.org mailing
3996 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3997 inline the MAP operation, and there is more than one
3998 sequence argument to the MAP call (so that it can't just
3999 do ETYPECASE once and for all based on the type of the
4000 single sequence argument). (The old non-inline implementation
4001 of the general M-argument sequence-of-length-N case required
4002 O(M*N*N) time when any of the sequence arguments were LISTs.)
4003 * The QUIT :UNIX-CODE keyword argument has been renamed to
4004 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4005 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4007 * Raymond Wiker's patches to port dynamic loading from Linux to
4008 FreeBSD have been added.
4009 * The BUGS file is now more nearly up to date, thanks in large part
4010 to Martin Atzmueller's review of it.
4011 * The debugger now flushes standard output streams before it begins
4012 its output ("debugger invoked" and so forth).
4013 * The core version number and fasl file version number have both
4014 been incremented, because of incompatible changes in the layout
4016 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4017 used to not be called for a saved Lisp image.)
4018 * Martin Atzmueller's version of a patch to fix a compiler crash,
4019 as posted on sbcl-devel 13 September 2000, has been installed.
4020 * Instead of installing Martin Atzmueller's patch for the
4021 compiler transform for SUBSEQ, I deleted the compiler transform,
4022 and transforms for some similar consing operations.
4023 * A bug in signal handling which kept TRACE from working on OpenBSD
4025 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4026 away by constant folding
4027 * The system now defines its address space constants in one place
4028 (in the Lisp sources), and propagates them automatically elsewhere
4029 (through GENESIS and the sbcl.h file). Therefore, patching the
4030 address map is less unnecessarily tedious and error-prone. The
4031 Lisp names of address space constants have also been systematized.
4032 * CVS tags like dollar-Header-dollar have been removed from
4033 the sources, because they have never saved me trouble and
4034 they've been source of trouble working with patches and other
4035 diff-related operations.
4036 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4037 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4039 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4041 * The system has been ported to OpenBSD.
4042 * The system now compiles with a simple "sh make.sh" on the systems
4043 that it's supported on. I.e., now you no longer need to tweak
4044 text in the target-features.lisp-expr and symlinks in src/runtime/
4045 by hand, the make.sh takes care of it for you.
4046 * The system is no longer so grossly inefficient when compiling code
4047 involving vectors implemented as general (not simple) vectors (VECTOR T),
4048 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4049 which dares to use the various sequence functions on non-simple
4050 vectors, takes less of a performance hit.
4051 * There is now a primitive type predicate VECTOR-T-P
4052 to test for the (VECTOR T) type, so that e.g.
4053 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4054 can now be compiled with some semblance of efficiency. (The old code
4055 turned the type declaration into a full call to %TYPEP at runtime!)
4056 * AREF on (VECTOR T) is still not fast, since it's still compiled
4057 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4058 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4059 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4060 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4061 without making full calls to SUBTYPEP at runtime.
4062 (Some analogous efficiency issues for non-simple vectors specialized to
4063 element types other than T, or for non-simple multidimensional arrays,
4064 have not been addressed. They could almost certainly be handled the
4065 same way if anyone is motivated to do so.)
4066 * The changes in array handling break binary compatibility, so
4067 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4068 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4069 instead of (VALUES T T).
4070 * By following the instructions that Dan Barlow posted to sbcl-devel
4071 on 2 July 2000, I was able to enable primitive dynamic object
4072 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4073 functionality is not implemented (since it calls ld to resolve
4074 library references automatically, requiring RUN-PROGRAM for its
4075 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4076 try to resolve library references) is now supported.
4077 * The system now flushes the standard output streams when it terminates,
4078 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4079 them at several other probably-convenient times, e.g. in each pass of
4080 the toplevel read-eval-print loop, and after evaluating a form given
4081 as an "--eval" command-line option. (These changes were motivated by a
4082 discussion of stream flushing issues on cmucl-imp in August 2000.)
4083 * The source transform for TYPEP of array types no longer assumes
4084 that an array whose element type is a not-yet-defined type
4085 is implemented as an array of T, but instead punts, so that the
4086 type will be interpreted at runtime.
4087 * There is now some support for cross-compiling in make.sh: each of
4088 the phases of make.sh has its own script. (This should be transparent
4089 to people doing ordinary, non-cross-compile builds.)
4090 * Since my laptop doesn't have hundreds of megabytes of memory like
4091 my desktop machine, I became more motivated to do some items on
4092 my to-do list in order to reduce the size of the system a little:
4093 ** Arrange for various needed-only-at-cold-init things to be
4094 uninterned after cold init. To support this, those things have
4095 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4096 symbols with such names are now uninterned after cold init).
4097 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4098 fasl files for cold load.
4099 ** Remove the old compiler structure pooling code (which used to
4100 be conditional on the target feature :SB-ALLOC) completely.
4101 ** Redo the representation of some data in cold init to be more compact.
4102 (I also looked into supporting byte compiled code at bootstrap time,
4103 which would probably reduce the size of the system a lot, but that
4104 looked too complicated, so I punted for now.)
4105 * The maximum signal nesting depth in the src/runtime/ support code has
4106 been reduced from 4096 to 256. (I don't know any reason for the very
4107 large old value. If the new smaller value turns out to break something,
4108 I'll probably just bump it back up.)
4109 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4111 * Many, many bugs reported by Peter Van Eynde have been added to
4112 the BUGS list; some have even been fixed.
4113 * While enabling dynamic object file loading, I tried to make the
4114 code easier to understand, renaming various functions and variables
4115 with less ambiguous names, and changing some function calling
4116 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4117 * While trying to figure out how to do the OpenBSD port, I tried to
4118 clean up some of the code in src/runtime/. In particular, I dropped
4119 support for non-POSIX signal handling, added various comments,
4120 tweaked the code to reduce the number of compilation warnings, and
4121 renamed some files to increase consistency.
4122 * To support the new automatic configuration functionality in make.sh,
4123 the source file target-features.lisp-expr has been replaced with the
4124 source file base-target-features.lisp-expr and the machine-generated
4125 file local-target-features.lisp-expr.
4126 * fixed a stupid quoting error in make.sh so that using CMU CL
4127 "lisp -batch" as cross-compilation host works again
4129 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4131 * DESCRIBE no longer tries to call itself recursively to describe
4132 bound/fbound values, so that it no longer fails on symbols which are
4133 bound to themselves (like keywords, T, and NIL).
4134 * DESCRIBE now works on generic functions.
4135 * The printer now prints less-screwed-up representations of closures
4136 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4137 * A private symbol is used instead of the :EMPTY keyword previously
4138 used to mark empty slots in hash tables. Thus
4139 (DEFVAR *HT* (MAKE-HASH-TABLE))
4140 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4141 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4142 now does what ANSI says that it should. (You can still get
4143 similar noncompliant behavior if bang on the hash table
4144 implementation with all the symbols you get back from
4145 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4146 This breaks binary compatibility, since tests for equality to
4147 :EMPTY are wired into things like the macroexpansion of
4148 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4150 * There's now a minimal placeholder implementation for CL:STEP,
4151 as required by ANSI.
4152 * An obscure bug in the interaction of the normal compiler, the byte
4153 compiler, inlining, and structure predicates has been patched
4154 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4155 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4157 * Missing ordinary arguments in a macro call are now detected even
4158 when the macro lambda list contains &KEY or &REST.
4159 * The debugger no longer complains about encountering the top of the
4160 stack when you type "FRAME 0" to explicitly instruct it to go to
4161 the top of the stack. And it now prints the frame you request even
4162 if it's the current frame (instead of saying "You are here.").
4163 * As specified by ANSI, the system now always prints keywords
4164 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4166 * The default initial SIZE of HASH-TABLEs is now smaller.
4167 * Type information from CLOS class dispatch is now propagated
4168 into DEFMETHOD bodies, so that e.g.
4169 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4171 is now basically equivalent to
4172 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4173 (DECLARE (TYPE SINGLE-FLOAT X))
4175 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4176 operation, without having to do run-time type dispatch.
4177 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4178 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4179 * Since I don't seem to be making big file reorganizations very often
4180 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4181 is overflowing my ability to conveniently back them up), I've finally
4182 checked the system into CVS. (The CVS repository is on my home system,
4183 not at SourceForge -- putting it on SourceForge might come later.)
4184 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4185 high-level GC-NOTIFY-FOO functions send their output. (There's
4186 still very little control of where low-level verbose GC functions
4187 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4188 less than it used to -- the GC-NOTIFY-FOO functions are now under
4189 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4190 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4191 in only one place in the source code, and propagates it automatically
4192 everywhere that it's needed. Thus e.g. when I bump the version from
4193 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4194 * The C source files now include boilerplate legalese and documentation
4195 at the head of each file (just as the Lisp source files already did).
4196 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4197 to his page will be replaced with a link to his new CLiki service.
4199 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4201 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4202 * The build process now looks for GNU make under the default name "gmake",
4203 instead of "make" as it used to. If GNU make is not available as "gmake"
4204 on your system, you can change this default behavior by setting the
4205 GNUMAKE environment variable.
4206 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4207 can build without error under CMU CL.
4209 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4211 * There is now a partial SBCL user manual (with some new text and some
4212 text cribbed from the CMU CL manual).
4213 * The beginnings of a profiler have been added (starting with the
4214 CMU CL profiler and simplifying and cleaning up). Eventually the
4215 main interface should be through the TRACE macro, but for now,
4216 it's still accessed through vaguely CMU-CL-style functions and macros
4217 exported from the package SB-PROFILE.
4218 * Some problems left over from porting CMU CL to the new
4219 cross-compilation bootstrap process have been cleaned up:
4220 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4221 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4222 part when getting rid of the compiler *BACKEND* stuff.)
4223 ** The value of *NULL-TYPE* was screwed up, because it was
4224 being initialized before the type system knew the final
4225 definition of the 'NULL type. This screwed up several key
4226 optimizations in the compiler, causing inefficiency in all sorts
4227 of places. (I found it because I wanted to understand why
4228 GET-INTERNAL-RUN-TIME was consing.)
4229 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4230 PROCLAIM FTYPE information. Unfortunately this broke binary
4231 compatibility again, since now the forms output by DEFGENERIC
4232 to refer to functions which didn't exist in 0.6.3.
4233 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4234 can use the new (as of 0.6.3) transform for SOME into MAP into
4236 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4237 values to (INTEGER 0 1000000), so that the time code will no longer
4238 occasionally get blown up by Linux returning 1000000 microseconds
4239 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4240 its output conform to the ANSI spec. (Alas, this makes its output
4241 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4242 * A full call to MAP NIL with a single sequence argument no longer conses.
4243 * fixes to problems pointed out by Martin Atzmueller:
4244 * The manual page no longer talks about multiprocessing as though
4245 it were currently supported.
4246 * The ILISP support patches have been removed from the distribution,
4247 because as of version 5.10.1, ILISP now supports SBCL without us
4248 having to maintain patches.
4249 * added a modified version of Raymond Toy's recent CMU CL patch for
4250 EQUALP comparison of HASH-TABLE
4252 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4254 * The system still can't cross-compile itself with
4255 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4256 entails), but at least it can compile more of itself that way
4257 than it used to be able to, and various buglets which were uncovered
4258 by trying to cross-compile itself that way have now been fixed.
4259 * This release breaks binary compatibility again. This time
4260 at least I've incremented the FASL file format version to 2, so that the
4261 problem can be detected reliably instead of just causing weird errors.
4262 * various new style warnings:
4263 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4264 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4265 ** using the lexical binding of a variable named in the *FOO* style
4266 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4267 as specified by ANSI.
4268 * *RANDOM-STATE* is no longer automatically initialized from
4269 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4270 default behavior of the system is to repeat its behavior every time
4271 it's run. If you'd like to change this behavior, you can always
4272 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4273 old convention there was no comparably easy way to get the system to
4274 repeat its behavior every time it was run.
4275 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4276 FTYPE declarations has been removed, in favor of their ANSI
4277 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4278 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4279 the types of their sequence arguments can be determined at compile time.
4280 This is done through a new open code expansion for MAP which eliminates
4281 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4282 argument types can be determined at compile time.
4283 * The optimizer now transforms COERCE into an identity operation when it
4284 can prove that the coerced object is already of the correct type. (This
4285 can be a win for machine generated code, including the output of other
4286 optimization transforms, such as the MAP transform above.)
4287 * Credit information has been moved from source file headers into CREDITS.
4288 * Source file headers have been made more standard.
4289 * The CASE macro now compiles without complaining even when it has
4292 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4294 * (Note that the way that the PCL macroexpansions were rewritten
4295 to accommodate the change in DEFGENERIC below breaks binary
4296 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4297 not run under sbcl-0.6.2. Once we get out of alpha releases,
4298 i.e. hit release 1.0.0, we'll probably try to maintain binary
4299 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4300 and sbcl-1.4.4. Until then, however, it might be fairly common
4301 for maintenance releases to break binary compatibility.)
4302 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4304 * The reporting of unhandled signals has been changed to print some
4305 explanatory text as well as the report form. (Previously only
4306 the report form was printed.)
4307 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4308 it defines, so that the compiler no longer issues undefined function
4309 warnings for compiled-but-not-yet-loaded generic functions.
4310 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4311 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4312 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4313 section of the ANSI spec are supported.
4314 * The "" nickname for the "KEYWORD" package has been removed.
4315 The reader still handles symbol tokens which begin with a package marker
4316 as keywords, but it doesn't expose its mechanism for doing so in the
4317 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4318 * The system now issues STYLE-WARNINGs for contradictory TYPE
4319 proclamations. (Warnings for contradictory FTYPE proclamations would
4320 be nice too, but those can't be done usefully unless the type system
4321 is made smarter about FUNCTION types.)
4322 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4323 systematized, so that "*target-*.lisp is supposed to exist only on the
4324 target and imply that there's a related file which exists on the
4325 host, and *host-*.lisp is supposed to exist only on the host and imply
4326 that there's a related file which exists on the target. This involves a
4327 lot of renaming. Hopefully the acute confusion caused by the renaming
4328 will be justified by the reduction in chronic confusion..
4329 ** runtime-type.lisp -> early-target-type.lisp
4330 ** target-type.lisp -> late-target-type.lisp
4331 ** early-host-format.lisp -> early-format.lisp
4332 ** late-host-format.lisp -> late-format.lisp
4333 ** host-error.lisp -> misc-error.lisp
4334 ** early-error.lisp -> early-target-error.lisp
4335 ** late-error.lisp -> late-target-error.lisp
4336 ** host-defboot.lisp -> early-defboot.lisp
4337 ** code/misc.lisp -> code/target-misc.lisp
4338 ** code/host-misc.lisp -> code/misc.lisp
4339 ** code/numbers.lisp -> code/target-numbers.lisp
4340 ** code/early-numbers.lisp -> numbers.lisp
4341 ** early-host-type.lisp -> early-type.lisp
4342 ** late-host-type.lisp -> late-type.lisp
4343 ** host-typep.lisp -> typep.lisp
4344 ** load.lisp -> target-load.lisp
4345 ** host-load.lisp -> load.lisp
4346 ** host-disassem.lisp -> disassem.lisp
4347 ** host-insts.lisp -> insts.lisp
4348 ** byte-comp.lisp -> target-byte-comp.lisp
4349 ** host-byte-comp.lisp -> byte-comp.lisp
4350 ** host-signal.lisp -> signal.lisp
4351 ** host-defstruct.lisp -> defstruct.lisp
4352 ** late-target-type.lisp -> deftypes-for-target.lisp
4353 Furthermore, several other previously target-only files foo.lisp (e.g.
4354 hash-table.lisp and random.lisp) have been split into a target-and-host
4355 foo.lisp file and a target-only target-foo.lisp file, with their key type
4356 definitions in the target-and-host part, so that the cross-compiler will
4357 know more about target types.
4358 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4359 gone away. In their place are various *BACKEND-FOO* variables
4360 corresponding to the slots of the old structure.
4361 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4362 into the target SBCL has been fixed.
4363 * The chill.lisp system for loading cold code into a running SBCL
4365 * Support for the CMU CL "scavenger hook" extension has been removed.
4366 (It was undocumented and unused in the CMU CL sources that SBCL was
4367 derived from, and stale in sbcl-0.6.1.)
4368 * Various errors in the cross-compiler type system were detected
4369 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4370 (enabling various consistency checks). Many of them were fixed,
4371 but some hard problems remain, so the compiler is back to
4372 running without *TYPE-SYSTEM-INITIALIZED* for now.
4373 * As part of the cross-compiler type system cleanup, I implemented
4374 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4375 * I have started adding UNCROSS calls throughout the type system
4376 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4377 on cross-compiler input files will be able to go away, and various
4379 * CONSTANTP now returns true for quoted forms (as explicitly required
4382 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4384 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4385 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4386 (MAKE-LIST -1). In the longer term, it ought to become true
4387 that declarations are assertions even at SAFETY 1. For now, it's not
4388 quite true even at SAFETY 3, but it's at least more nearly true..
4389 (Note that this change seems to increases the size of the system by
4390 O(5%) and to decrease the speed of the compiler by 20% or more.)
4391 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4392 for the problem of printing dozens of lines of distracting information
4393 about low-level system machinery as part of the top stack frame
4394 on entry to the debugger when an undefined function was called.
4395 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4396 is not reset to COMMON-LISP-USER.
4397 * Compilation of stuff related to dyncount.lisp has been made conditional
4398 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4399 smaller. The various dyncount-related symbols have been moved into
4400 a new "SB-DYNCOUNT" package.
4401 * tty-inspect.lisp has been renamed to inspect.lisp.
4402 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4403 feature has gone away. (When we eventually port to other flavors of
4404 libc and/or Unix, we'll try to make the differences between flavors
4405 invisible at the user level.)
4406 * Various other *FEATURES* tags, and/or their associated conditionals,
4407 have been removed if obsolescent, or given better documentation, or
4408 sometimes given more-mnemonic names.
4410 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4412 * tidied up "make.sh" script
4413 * tidied up system directory structure
4414 * better "clean.sh" behavior
4415 * added doc/FOR-CMUCL-DEVELOPERS
4416 * many many small tweaks to output format, e.g. removing possibly-confusing
4417 trailing #\. character in DESCRIBE-INSTANCE
4418 * (EQUALP #\A 'A) no longer signals an error.
4419 * new hashing code, including EQUALP hashing
4420 * tidied up Lisp initialization and toplevel
4421 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4422 * command line argument processing
4423 * added POSIX-GETENV function to deal with Unix-ish environment variables
4424 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4425 terminating SBCL on EOF
4426 * non-verbose GC by default
4427 * There is no more "sbcl" shell script; the sbcl file is now the C
4428 runtime executable (just like CMU CL).
4429 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4431 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4432 symbol and package information as Lisp native symbol and package objects
4433 instead of strings naming symbols and strings naming packages. This way,
4434 whenever packages are renamed (as in warm init), debug information is
4435 transformed along with everything else.
4436 * tweaked the optimization policy declarations which control the building
4437 of SBCL itself. Now, among other things, the system no longer saves
4438 source location debugging information. (This helps two problems at once
4439 by reducing SBCL size and by keeping SBCL from trying to look for its
4440 sources -- which may not exist -- when reporting errors.)
4441 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4442 debugging and testing purposes
4443 * cleaned up printing, making the printer call PRINT-OBJECT for
4444 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4445 methods, giving nearly-ANSI behavior
4446 * converted almost all special variables to use *FOO* naming convention
4447 * deleted PARSE-TIME functionality, since it can be done portably
4448 * moved some files out of cold init into warm init
4449 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4451 * regularized formatting of source files
4452 * added an install.sh script
4453 * fixed ridiculous memory usage of cross-compiler by making
4454 compiler/alloc.lisp not try to do pooling unless it can hook
4455 itself into the GC of the cross-compilation host. Now the system
4456 builds nicely on my old laptop.
4457 * added :SB-ALLOC in target-features.lisp-expr
4458 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4459 was not implemented)
4460 * re-did condition handling and note reporting in the compiler. Notes
4461 are no longer handled by signalling conditions. Style warnings
4462 and warnings are handled more correctly and reported in such a way
4463 that it's easy to find one or the other in your output (so that you
4464 can e.g. figure out which of many problems caused COMPILE-FILE to
4466 * changed the severity of several compiler warnings from full WARNING
4467 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4468 compiler note reporting so that it doesn't use the condition system
4469 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4470 COMPILE-FILE command)
4471 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4472 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4473 no longer supported, since their implementation was deeply intertwingled
4474 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4475 * removed bogus "support" for compiler macros named (SETF FOO), and
4476 removed the compiler macro for SETF INFO (but only after making a fool
4477 of myself on the cmucl-imp mailing list by posting a bogus patch for
4478 DEFINE-COMPILER-MACRO..)
4479 * Compiled files containing forms which have side effects on the Lisp
4480 reader (such as DEFPACKAGE forms) are now handled more correctly.
4481 (Compiler queuing of top level lambdas has been suppressed by setting
4482 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4483 * deleted various currently-unused source files, e.g. gengc.lisp. They
4484 may be added back at some point e.g. when porting to other architectures,
4485 but until they are it's distracting to distribute them and to try to
4487 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4488 there were so many of them they're just distractions, and UNCROSS is
4489 known to be able to handle the current sources
4490 * moved PROFILE functionality into TRACE, so that it will be clear
4491 how the wrapping and unwrapping of functions when you profile them
4492 interacts with the wrapping and unwrapping of functions when you
4493 trace them. (Actually, the functionality isn't there yet, but at least
4494 the interface specification is there. Hopefully, the functionality will
4495 arrive with some future maintenance release.)
4496 * removed host-oops.lisp
4497 * changed signature of QUIT function to allow UNIX-CODE argument
4498 * fixed READ-SEQUENCE bug
4499 * tweaked verbose GC output so that it looks more like the progress
4500 output that ANSI specifies for functions like LOAD
4501 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4502 * added <http://sbcl.sourceforge.com> to the banner information printed by