1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
3 * incompatible change: hash-table accessor functions are no longer
4 automatically protected by locks. Concurrent accesses on the same hash-table
5 from multiple threads can give inconsistent results or even corrupt the
6 hash-table completely. Multi-threaded applications should do their own
7 locking at the correct granularity. In the current implementation it is
8 still safe to have multiple readers access the same table, but it's not
9 guaranteed that this property will be maintained in future releases.
10 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
11 and will signal an error at runtime.
12 * enhancement: SB-THREAD package now exports a semaphore interface.
13 * enhancement: CONS can now stack-allocate on x86 and
14 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
16 * enhancement: nested lists can now be stack allocated on
17 platforms providing stack allocation support.
18 * enhancement: dynamic-extent support has been extended to support
19 cases where there are multiple possible sources for the stack
21 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
22 if the mutex is uncontested on Linux.
23 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
24 as specified by the CLHS. (thanks to Tobias Rittweiler)
25 * bug fix: NaN comparison now works on x86-64.
26 * bug fix: CLOSE :ABORT T on a stream with pending output now
28 * bug fix: instances of non-standard metaclasses using standard
29 instance structure protocol sometimes missed the slot type checks
32 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
33 * minor incompatible change: the MSI installer on Windows no longer
34 associates .lisp and .fasl files with the installed SBCL.
35 * minor incompatible change: :UNIX is no longer present in *FEATURES*
36 on Windows. (thanks to Luis Oliviera)
37 * new feature: SB-CLTL2 contrib module now implements
38 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
39 * optimization: scavenging weak pointers is now more efficient,
40 requiring O(1) instead of O(N) per weak pointer to identify
41 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
42 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
43 method bodies using constant slot-names when the first argument is
44 a specializer parameter for the method.
45 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
46 names, when the class of the instance is a direct instance
47 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
48 as slow as the constant slot-name case.)
49 * optimization: member type construction is now O(N) instead
51 * optimization: UNION and NUNION are now O(N+M) for large
52 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
54 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
55 now more readable in environments like Slime which display it.
56 (thanks to Tobias C. Rittweiler)
57 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
58 was able to infer the correct argument type for the object on which
59 the CAS operation was being performed.
60 * bug fix: copy propagation interfered with parallel assignment
61 semantics in local calls. (reported by Paul Khuong)
62 * bug fix: the signed modular fixnum shift compiled to wrong code on
63 x86 and x86-64. (spotted by a slight modification to some of
66 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
67 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
68 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
69 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
71 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
72 and generic function now print the object identity for anonymous
73 instances. (thanks to Attila Lendvai)
74 * enhancement: as an extension to MOP, SBCL now supports SETF of
75 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
76 (thanks to Attila Lendvai)
77 * enhancement: sb-cover has an alternate annotation mode which puts
78 the form annotations at the car of the form, instead of annotating
79 it completely. The benefit of this mode is that it shows better
80 which forms were instrumented by the compiler.
81 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
82 by printer control variables. (reported by Dan Corkill)
83 * bug fix: system leaked memory when delayed output was performed by
84 the OS in smaller chunks then expected. (thanks to David Smith)
85 * bug fix: system leaked memory when file streams were not closed
87 * bug fix: large objects written to slow streams that were modified
88 after the write could end up with the modified state written to
89 the underlying file descriptor.
90 * bug fix: multiple threads operating in parallel on the same stream
91 could cause buffer-overflows.
92 * bug fix: source location information is stored correctly
93 (broken since 1.0.6). This bug would generally show up as the
94 Slime debugger higlighting the wrong form when the sldb-show-source
96 * bug fix: Branch forms are again annotated as branches in the sb-cover
98 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
99 on x86-64. (reported by Gregory Vanuxem)
100 * bug fix: EQUALP could return wrong results for structures with raw
101 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
102 word). (reported by Vjacheslav Fyodorov)
103 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
106 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
107 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
108 atomic compare-and-swap operations on threaded platforms.
109 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
110 allows assining a global minimum value to optimization qualities
111 (overriding proclamations and declarations).
112 * enhancement: closed over variables can be stack-allocated on x86
114 * performance bug fix: GETHASH and (SETF GETHASH) are once again
116 * optimization: slot definition lookup is now O(1). This speeds up
117 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
118 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
119 * optimization: ASSOC and MEMBER can now be open-coded for all
120 combinations of keyword arguments when second argument is constant
121 and SPEED >= SPACE. In other cases a specialized version is
123 * bug fix: using obsoleted structure instances with TYPEP and
124 generic functions now signals a sensible error.
125 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
126 (reported by Kristoffer Kvello)
127 * bug fix: backtrace construction is now more careful when making
128 lisp-objects from pointers on the stack, to avoid creating bogus
129 objects that can be seen by the GC.
130 * bug fix: defaulting of values in contexts expecting more than 7
131 variables now works on x86-64. (reported by Christopher Laux)
132 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
134 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
135 as the property-list of a symbol.
136 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
137 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
138 situations requested, are once again file-compileable. (reported
141 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
142 * MOP improvement: support for user-defined subclasses of
143 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
144 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
145 the expansion of DEFMETHOD, is responsible for generating a form
146 which creates a list of specializers when evaluated. Additional
147 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
148 debugging and introspective support.
149 * minor incompatible change: the (unsupported) spinlock interface
150 has changed: free spinlock now has the value NIL, and a held spinlock
151 has the owning thread as its value.
152 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
153 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
155 * enhancement: name of a socket-stream is now "a socket" instead of
157 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
158 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
159 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
160 for generic functions. (Thanks to Tobias C. Rittweiler)
161 * optimization: bignum printing speed has been improved by 20-40%
162 (depending on the bignum size.)
163 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
165 * bug fix: the cache used by the CLOS to store precomputed effective
166 methods, slot offsets, and constant return values is now thread and
168 * bug fix: generic function dispatch function updating is now thread
169 and interrupt safe (in the sense that the known issues have been
171 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
172 * bug fix: interning EQL-specializers is now thread and interrupt safe.
173 * bug fix: asdf systems with dependencies to the SB-POSIX or
174 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
175 * bug fix: interrupt safety of applicable method computation has been
178 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
179 * new contrib: sb-cover, an experimental code coverage tool, is included
181 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
182 significantly faster.
183 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
184 produce more efficient code on x86-64 if the compiler can determine
185 that (+ INDEX <constant>) does not require a bounds check and FOO
186 has an element type at least 8 bits wide.
187 * enhancement: a new, experimental synchronous timeout facility is
188 provided. Refer to SB-SYS:WITH-DEADLINE for details.
189 * enhancement: when a symbol name conflict error arises, the
190 conflicting symbols are always printed with a package prefix.
191 (thanks to Kevin Reid)
192 * enhancement: stepping is now once again supported on the SPARC and
193 MIPS platforms. (It is also now more likely to work on CheneyGC
195 * enhancement: sb-sprof can now also track and report accurate call
197 * bug fixes: the treatment of non-standard subclasses of
198 SB-MOP:SPECIALIZER is more correct.
199 * incompatible change: PURIFY no longer copies the data from the
200 dynamic space into the static and read-only spaces on platforms
201 that use the generational garbage collector
202 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
204 * bug fix: GC race condition occasionally resulting in crashes with
205 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
207 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
209 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
210 a STYLE-WARNING to be signalled (regression from 1.0.4.)
211 * bug fix: an asynchronous interrupt could previously leave the
212 system running with GC inhibited.
213 * bug fix: a DECLARE form evaluated at top-level now causes an error
214 rather than silently (or verbosely) returning NIL.
215 * bug fix: trying to dissassemble functions compiled at high DEBUG could
216 result in a error being signalled due to source form lookup errors.
217 (reported by Peter Graves)
219 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
220 * incompatible change: removed writer methods for host-ent-name,
221 host-ent-addresses -- changing the values did not update the DNS
223 * minor incompatible change: changed experimental JOIN-THREAD interface
224 * documentation: the manual now lists reader and writer methods
225 in class slot documentation sections. (thanks to Richard M Kreuter)
226 * documentation: unwinding from asyncronous events has been
227 documented as unsafe.
228 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
229 in multithreaded application code.
230 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
231 platforms. (thanks to James Anderson for the optimization hint)
232 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
234 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
235 variants no longer cons.
236 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
237 their NOT- variants no longer cons.
238 * optimization: EQUAL hash tables no longer use SXHASH for objects
239 of all data types, but instead use an EQL hash for types for which
240 EQUAL is the same as EQL.
241 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
242 are significantly faster.
243 * optimization: new STRING-OUTPUT-STREAM implementation is much
244 faster for large outputs and conses 30% less on average.
245 * enhancement: XREF information is now collected for references made
246 to global variables using SYMBOL-VALUE and a constant argument.
247 * enhancement: SIGINT now causes a specific condition
248 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
249 * bug fix: FILE-POSITION used to signal an error in some cases where
250 ANSI requires it to return NIL.
251 * bug fix: ADJUST-ARRAY is now interrupt-safe.
252 * bug fix: adding and removing fd-handlers is now interrupt-safe.
253 * bug fix: inlined calls to C now ensure 16byte stack alignment on
255 * bug fix: bad type declaration in the CLOS implementation has
256 been fixed. (reported by James Anderson)
257 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
258 (thanks to Lutz Euler)
259 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
260 required by the CLHS. (reported by Samium Gromoff)
261 * bug fix: dead unbound variable references now signal an error.
262 (reported by Marco Monteiro)
263 * bug fix: / with an unused value was being deleted in safe code.
264 (thanks to Marco Monteiro and Kevin Reid)
265 * bug fix: number of characters that can be written onto a single
266 line in a file is unlimited.
267 * bug fix: some GC deadlocks caused by asynchronous interrupts have
268 been fixed by inhibiting interrupts for when GC is disbled.
269 * bug fix: some interrupt safety issues with GETHASH, (SETF
270 GETHASH), CLRHASH and REMHASH have been fixed.
271 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
272 specifier no longer causes infinite recursion.
273 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
274 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
275 * bug fix: an &environment argument with the correct variable information
276 is passed to macros that are expanded during byte compilation.
277 (reported by Samium Gromoff)
278 * bug fix: a logic error was causing the hash function for bignums to
279 have excessive amounts of collisions. (reported by Faré Rideau)
280 * bug fix: modifying the contents of an array could change the return
281 value of SXHASH on that array, which is only allowed for strings
282 and bit vectors. (bug introduced in 0.9.16)
283 * improvement: the x86-64/darwin port now passes all tests (except
284 for the debugger tests) but should still be considered
285 experimental until this is fixed.
286 * improvement: a style-warning is signaled for CASE (etc) clauses with
287 duplicate keys. (thanks to Kevin Reid)
288 * improvement: macos/x86 and macos/x86-64 now use mach exception
289 handlers for dealing with illegal instructions (for trapping and
290 error handling) and memory protection violations (for GC).
292 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
293 * new platform: experimental support for x86-64/darwin (MacOS).
294 * incompatible change: the thread-safe (on most platforms) getaddrinfo
295 and getnameinfo sockets functions are used instead of gethostbyaddr
296 and gethostbyname, on platforms where the newer functions are available.
297 As a result, the ALIASES field of HOST-ENT will always be NIL on these
299 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
300 don't print the startup banner, but behave as if --noinform was passed
301 as a command line argument. (thanks to Kevin Reid)
302 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
303 * optimization: code using alien values with undeclared types is much faster.
304 * optimization: the compiler is now able to open code SEARCH in more cases.
305 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
306 * bug fix: using standardized COMMON-LISP special variables as loop
307 variables no longer signals bogus package lock violations. (reported
309 * bug fix: declaring local loop variables to be of a range-limited type
310 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
311 (reported by Andras Simon)
312 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
313 bugs remain on x86-64.)
314 * bug fix: the #= and ## reader macros now interact reasonably with
315 funcallable instances.
316 * bug fix: type-checks for function arguments were compiled using the
317 compiler policy settings of the wrong lexical environment (in 1.0.2
319 * bug fix: SHADOW accepts characters as string designators, as required
320 by the spec (thanks to Eric Marsden)
321 * bug fix: fixed GC safety issues when foreign functions are called with
322 non-base strings as arguments
323 * bug fix: more consistent error output for fatal-compiler-warnings, like
325 * bug fix: functions with proper names were showing up as NIL in
326 backtraces (reported by Edi Weitz, regression in 1.0.2)
328 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
329 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
330 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
331 produce more efficient code on the x86 if the compiler can determine
332 that (+ INDEX <constant>) does not require a bounds check and FOO
333 has an element type at least 8 bits wide.
334 * bug fix: references to undefined variables are handled the same way
335 in toplevel forms as in the normal compiler
336 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
338 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
339 (thanks to Jon Buller)
340 * improvement: faster compilation times for complex functions
341 * improvement: added readlink support to SB-POSIX (thanks to Richard
344 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
345 * improvement: experimental support for mach exception handling on
346 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
348 * improvement: support for GBK external format.
349 (thanks to Chun Tian (binghe))
350 * improvement: the debugger now displays variables that have been closed
351 over, in code compiled with (DEBUG 2) or higher
352 * improvement: support for executable cores on NetBSD (thanks to
354 * new feature: added a RESTART-FRAME debugger command
355 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
356 be used to provide an implementation for FILE-POSITION on Gray streams
357 (thanks to Eric Marsden)
358 * new feature: add syslog, openlog and closelog support to SB-POSIX
359 (thanks to Richard Kreuter)
360 * optimization: the function call overhead in code compiled with
361 a high DEBUG optimization setting is significantly
362 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
363 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
364 (thanks to Tony Martinez)
365 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
366 evaluated once (reported by Kevin Reid)
367 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
368 works again (thanks to Kilian Sprotte)
369 * bug fix: an error is signaled for tagbodies with duplicate tags
370 (thanks to Stephen Wilson)
371 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
372 * bug fix: Win32 port can now handle foreign code unwinding Lisp
373 stack frames from alien callbacks.
374 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
375 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
376 65k pages (thanks to David Woodhouse)
377 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
379 changes in sbcl-1.0.1 relative to sbcl-1.0:
380 * new platform: FreeBSD/x86-64, including support for threading.
381 * new feature: the compiler stores cross-referencing information
382 abount function calls (who-calls), macroexpansion (who-macroexpands)
383 and special variables (who-binds, who-sets, who-references) for code
384 compiled with (< SPACE 3). This information is available through the
385 sb-introspect contrib.
386 * new feature: users may subclass SEQUENCE, and have instances of
387 these classes interoperate with standard Common Lisp functions if
388 a number of methods are defined. (This feature is experimental
389 and the interface subject to change based on feedback from SBCL
390 users and the general community)
391 * improvement: sb-sprof traces call stacks to an arbitrary depth on
392 x86/x86-64, rather than the previous fixed depth of 8
393 * improvement: another pthread back-end of mutex "pthread-futex".
394 * bug fix: non-ascii command-line arguments are processed correctly
395 (thanks to Yaroslav Kavenchuk)
396 * bug fix: non-required arguments were not passed correctly when a method
397 defined using DEFMETHOD was called from a mop-generated method using
398 CALL-NEXT-METHOD (reported by Pascal Costanza)
399 * bug fix: recursion is now permitted in accessors through
400 SLOT-UNBOUND. (reported by Pascal Costanza)
401 * bug fix: an error was signaled at startup if the HOME environment
402 variable was defined, but had an empty value (reported by Peter Van Eynde)
403 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
404 signal a PROGRAM-ERROR, not a BUG.
405 * bug fix: SB-POSIX:READDIR works when built with large file support.
406 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
408 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
409 for better automatic indentation support. (thanks to Matt Pillsbury)
410 * optimization: loading generic functions no longer takes O(n^2) time,
411 proportional to the amount of methods in the generic function
412 (reported by Todd Sabin and Jeremy Brown)
413 * optimization: the FIND and POSITION family of sequence functions
414 are significantly faster on arrays whose element types have been
416 * improvements to the Windows port:
417 ** Intermittent heap corruption problems have been fixed. (thanks
418 to Alastair Bridgewater)
419 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
421 ** Lisp is able to unwind foreign exception frames from alien
422 callbacks. (thanks to Alastair Bridgewater)
424 changes in sbcl-1.0 relative to sbcl-0.9.18:
425 * improvement: experimental support for threading on FreeBSD/x86.
426 (thanks to NIIMI Satoshi)
427 * improvement: runtime option --dynamic-space-size can be used
428 to set the size of the dynamic space reserved on startup.
429 * improvement: floating point modes in effect are now saved in
430 core, and restored on startup.
431 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
432 startup, not time since first call to GET-INTERNAL-REAL-TIME.
433 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
434 threads are not running after *SAVE-HOOKS* have run.
435 * improvement: writes to CLOS instance slots are type-checked in code
436 compiled with (SAFETY 3)
437 * improvement: floating-point exception handling on FreeBSD (thanks to
439 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
440 (thanks to Zach Beane)
441 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
443 * improvement: added support for the Shift-JIS external format.
444 (contributed by NIIMI Satoshi)
445 * improvement: callbacks are supported on Linux/PPC. (thanks to
447 * bug fix: compiler bug triggered by a (non-standard) VALUES
448 declaration in a LET* was fixed. (reported by Kaersten Poeck)
449 * bug fix: file compiler no longer confuses validated and already
450 dumped structurres. (reported by Kaersten Poeck)
451 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
452 fill-pointer signals a type-error as required. (thanks to
454 * bug fix: disassemly of funcallable instances works.
455 * bug fix: single stepping on PPC.
456 * bug fix: fix thread-safety problems in the type system (generally
457 manifesting as nonsensical errors like "STRING is a bad type specifier
458 for sequences" or "The value 1 is not of type FIXNUM")
459 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
460 (reported by Josip Gracin)
461 * bug fix: an error is signaled for attempts to displace arrays with
462 incompatible element types (thanks to Mario Mommer)
463 * bug fix: more correct handling of wide characters in debug info
464 (bug reported by Attila Lendvai and fixed by Juho Snellman)
465 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
466 and don't cause extra consing
467 * optimization: MAP and MAP-INTO are significantly faster on vectors
468 whose elements types have been declared.
469 * Improvements to SB-SPROF:
470 ** Support for allocation profiling
471 ** Reduced profiling overhead, especially for long profiling runs
472 * Improvements to the Windows port:
473 ** floating point exceptions are now reported correctly.
474 ** stack exhaustion detection works partially.
475 ** more accurate GET-INTERNAL-REAL-TIME.
476 ** better RUN-PROGRAM behaviour in case of errors in the child
478 ** PROBE-FILE now simplifies pathnames correctly.
479 ** DIRECTORY now works correctly with :WILD-INFERIORS.
480 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
481 1st 1970 (although time timezone data will be unreliable).
482 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
484 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
485 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
486 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
488 * bug fix: two potential GC deadlocks affecting threaded builds.
489 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
490 returns T (reported by Anton Kazennikov)
491 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
492 with non-variable places
493 * bug fix: remove a race condition in the setting of
494 funcallable-instance functions, this should make threaded CLOS
495 code more stable against memory faults.
496 * bug fix: corruption of specials when unbinding is interrupted by an
497 asynchronous unwind (reported by Hannu Koivisto)
498 * improvement: the debugger will now also display local variables that
499 are only used once, for code compiled with a DEBUG optimization quality
502 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
503 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
504 * incompatible change: External-format support for FFI calls. The
505 SB-ALIEN:C-STRING no longer implies an ASCII
506 external-format. Instead the string is subject to external-format
507 conversion. Additionally return values of type SB-ALIEN:C-STRING
508 are by default of type (SIMPLE-ARRAY CHARACTER), not
509 SIMPLE-BASE-STRING. For an alien type that behaves like the old
510 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
511 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
512 * incompatible change: SB-EXT package no longer contains the
513 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
514 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
516 * incompatible change: the single-stepper is no longer available
517 on Alpha, Mips and Sparc platforms.
518 * minor incompatible change: the direct superclasses of
519 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
520 not (STANDARD-OBJECT FUNCTION). This makes the
521 class-precedence-lists of GENERIC-FUNCTION and
522 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
524 * minor incompatible change: the default stream external format on
525 non-unicode SBCL is no longer detected from the locale, but is always
527 * new feature: Add a version of evaluator that uses an interpreter instead
528 of the compiler. EVAL still uses the compiler by default, to switch it
529 to use the interpreter, set the value of the variable
530 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
531 * minor incompatible change: the single-stepper REPL has been merged
532 with the normal debugger (see the "Stepping" heading of the debugger help
533 for more details). The debugger command STEP will no longer switch
534 to the single-stepper REPL.
535 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
536 for a type now works.
537 * bug fix: loading of default sysinit file works. (thanks to Leonid
539 * bug fix: better detection of circularities in the file-compiler.
540 (reported by Marco Monteiro)
541 * bug fix: the CL pathname functions now work with files that have
542 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
543 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
544 systems with Unix98 pty semantics.
545 * bug fix: ASDF-INSTALL will now work with bsd tar.
546 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
548 * bug fix: timers expiring in dead threads no longer cause a
549 type-error (reported by Paul "Nonny Mouse").
550 * bug fix: thanks to more lightweight single-stepper instrumentation,
551 code compiled with (DEBUG 3) will compile and execute significantly faster,
552 and will have more accurate type-inferencing than before
553 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
554 whose bindings are modified
555 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
556 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
557 ** replace dummy implementations of CL:MACHINE-INSTANCE and
558 CL:SOFTWARE-VERSION with proper version
560 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
561 * feature: implemented the READER-METHOD-CLASS and
562 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
563 as specified by AMOP.
564 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
566 * optimization: faster LOGCOUNT implementation on x86 and x86-64
567 (thanks to Lutz Euler)
568 * optimization: hashing of general arrays and vectors has been
569 improved. (reported by Any Fingerhut)
570 * enhancement: SB-INTROSPECT is now able to find definitions of
571 profiled functions. (thanks to Troels Henriksen)
572 * enhancement: compiler-macro expansion applies now to FUNCALL forms
573 as well, allowing compiler-macros for SETF-functions to expand.
574 * enhancement: step-instrumentation no longer wraps known
575 single-value functions with multiple-value context, allowing
576 better type inference.
577 * fixed bug #337: use of MAKE-METHOD in method combination now works
578 even in the presence of user-defined method classes. (reported by
579 Bruno Haible and Pascal Costanza)
580 * fixed bug #339(c): if there are applicable methods not part of any
581 long-form method-combination group, call INVALID-METHOD-ERROR.
582 (reported by Bruno Haible)
583 * fixed bug #361: the :FUNCTION initarg in the protocol for
584 initialization of methods can now be used to override
585 internally-produced optimized functions. (reported by Bruno
587 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
588 system-provided lambda expression no longer cause warnings about
589 unbound #:|pv-table| symbols.
590 * bug fix: improved the handling of type declarations and the
591 detection of violations for keyword arguments with non-constant
593 * bug fix: potentially erronous calls to PATHNAME and
594 MERGE-PATHNAMES were being flushed in some cases.
595 (reported by Richard Kreuter)
596 * bug fix: compiled calls to TYPEP were mishandling obsolete
597 instances. (reported by James Bielman and Attila Lendvai)
598 * bug fix: format strings with ~> without matching ~< no longer
599 trigger an AVER, but signal an understandable error instead.
600 (reported by Antonio Martinez)
601 * bug fix: specifying an output-file without a pathname-type for
602 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
603 of input-file instead of "fasl". (reported by Robert Dodier)
604 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
605 (reported by James Y Knight).
606 * bug fix: compiler-macros-function did not consider the environment
607 argument for shadowing by local functions.
608 * bug fix: compiler-macros expansion was inhibited by local INLINE
610 * bug fix: inline expansions of known functions were subject to
611 step-instrumentation in high debug policies, leading to problems
613 * bug fix: compiler failed to differentiate between different CONS
615 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
616 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
617 * bug fix: #368: incorrect use of expressed vs. upgraded array
619 * bug fix: #306a: more precise unions of array types.
620 * thread-safety improvements:
621 ** CONDITION-WAIT could return early on Linux, if the thread was
622 interrupted and subsequently continued with SIGCONT.
623 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
625 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
626 * added support for the ucs-2 external format. (contributed by Ivan
628 * minor incompatible change: pretty printing of objects of type
629 (cons symbol) is, in the default pprint-dispatch-table, now
630 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
632 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
633 called later in a class's lifetime, possibly as late as when the
634 first instance of the class is created. Previously,
635 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
636 class became finalizeable.
637 * fixed bug: FILE-POSITION sometimes returned inconsistent results
638 for multibyte external-format streams. (thanks to "vbzoli")
639 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
640 with :ALLOCATION :CLASS inherited from superclasses of the
642 * fixed bug: anonymous classes can now be created using the :NAME
643 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
644 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
645 * fixed bug: core-files saved with :EXECUTABLE T can again be
646 executed when SBCL_HOME isn't set. (reported by James Knight)
647 * fixed bug: toplevel LOCALLY forms with declarations could
648 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
649 * fixed bug: printing from several different threads using different
650 values of *print-case* could cause invalid output, due to
651 some internal special variables of the printer not being bound
652 thread-locally (reported by Max Mikhanosha)
653 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
654 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
655 (reported by Pascal Costanza)
656 * fixed bug: FINALIZE-INHERITANCE is called from
657 REINITIALIZE-INSTANCE on classes when the class has previously
658 been finalized, as required by AMOP.
659 * minor code generation optimizations:
660 ** better register allocation in CLOS dispatching functions
661 ** overflow detection when coercing signed bytes to fixnums on x86-64
662 ** is now implemented with one IMUL instruction instead of three shifts
663 ** more efficient bit-vector access on x86 and x86-64
664 ** more efficient access to raw structure slots on x86 and x86-64
665 * fixed some bugs revealed by Paul Dietz' test suite:
666 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
669 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
670 * feature: thread support on Solaris/x86, and experimental thread support
672 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
674 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
675 default method for SLOT-UNBOUND.
676 * minor incompatible change: prevent the user from specializing the
677 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
678 somewhat counter to the intent of the protocol, I (CSR) think, and
679 additionally it just doesn't work in SBCL as currently
680 implemented, thanks to optimizations (that are always valid for
681 the other three SLOT-VALUEish functions, but not for the setter).
682 * minor incompatible change: the :SB-LDB feature is now enabled by
683 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
684 the low-level debugger.
685 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
686 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
687 down to an SB-SYS:FD-STREAM, allowing direct communication in
688 more cases. Temporary files and pipes are still used as a fallback
690 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
691 on the non-reentrant C functions localtime(3) and gmtime(3).
692 * bug fix: global optimization policy was not visible in LOCALLY and
694 * bug fix: class objects can be used as specializers in methods.
695 (reported by Pascal Costanza)
696 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
697 with a type error. (reported by blitz_ on #lisp)
698 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
699 failed with a type error. (reported by Pascal Bourguignon)
700 * bug fix: merging pathnames against defaults with :DIRECTORY
701 starting with '(:RELATIVE :BACK) should preserve the :BACK.
702 (reported by James Y Knight)
703 * bug fix: saving large (>2GB) cores on x86-64 now works
704 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
705 constant index and a (simple-array (signed-byte 32)) array
706 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
707 error (patch by Robert J. Macomber)
708 * feature: TIME now displays the amount of run-time spent in GC
709 * bug fix: The debugger now does a better job of respecting
710 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
711 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
712 workaround for bug 403.)
713 * bug fix: floating point exception handling now works on FreeBSD
714 (thanks to NIIMI Satoshi)
715 * fixed some bugs revealed by Paul Dietz' test suite:
716 ** MISC.641: LET-conversion were not supposed to work in late
718 * improvements to the Win32/x86 port:
719 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
720 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
722 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
723 * new feature: source path information is generated for macro-expansion
724 errors for use in IDE's like Slime (thanks to Helmut Eller)
725 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
726 * bug fix: compiler does not loop forever on an invalid type in
728 * improvement: compilation of most CLOS applications is significantly
730 * optimization: added a limited bytecode compiler for simple toplevel
731 forms, speeding up compilation and FASL loading
732 * bug fix: the statistical profiler now properly distinguishes anonymous
735 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
736 * minor incompatible change: in sbcl-0.9.11 (but not earlier
737 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
738 into subclasses of the finalized class. Now user calls to
739 FINALIZE-INHERITANCE finalize just the one class, and calls by the
740 system return before any subclasses are finalized.
741 * minor incompatible change: The reader no longer ignores errors
742 regarding non-existent packages in #+ and #- feature tests.
743 * new feature: command line options --no-sysinit, --no-userinit to
744 inhibit loading the corresponding init files
745 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
746 for sending data through UDP sockets (thanks to François-René Rideau)
747 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
748 error is signalled instead (thanks to François-René Rideau)
749 * minor incompatible change: Error signalling behaviour of lexical
750 operations violating package locks has changed slightly. Refer to
751 documentation on package locks for details.
752 * enhancement: EVAL can process IF-expressions without resorting to the
754 * bug fix: Some operations involving SIN, COS, or TAN weren't being
755 constant-folded properly due to missing out-of-line functions.
756 (reported by Mika Pihlajamäki)
757 * bug fix: LISTEN sometimes returned T even in cases where no data was
758 immediately available from the stream
759 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
760 were reversed. (reported by Levente Mészáros)
761 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
762 when given a vector displaced to an adjustable array. (reported
764 * bug fix: derivation of float boundaries from numbers outside the
765 appropriate float range (reported by John Wiseman)
766 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
768 * bug fix: precomputing cache entries for generic functions with
769 some subclasses of specializers as yet invalid does not attempt to
770 fill a cache line with a negative offset. (reported by Levente
772 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
773 allow condition class objects as arguments to DOCUMENTATION and
774 (SETF DOCUMENTATION); only find and set documentation for
775 structure names for the STRUCTURE doc-type. (suggested by Gary
777 * improvements to the Win32/x86 port:
778 ** (user-homedir-pathname) and default initialization file
779 locations now know about the user's "Documents and Settings"
780 directory (thanks to Yaroslav Kavenchuk)
781 ** run-program is implemented (thanks to Mike Thomas)
782 ** sockets support (thanks to Timothy Ritchey)
783 ** better backtrace support (thanks to Alastair Bridgewater)
784 ** sb-grovel supported
785 ** asdf-install and sb-posix work somewhat
786 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
787 * improvements to the Solaris/x86 port:
788 ** works on Solaris 11/Solaris Express
789 ** floating-point exception handling support
790 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
791 * fixed some bugs revealed by Paul Dietz' test suite:
792 ** REMOVE-METHOD returns its generic function argument even when
793 the method is not one of the generic functions' methods.
794 ** objects declared as MEMBER types can be admissible arguments to
796 ** printing characters should simply be printed by the FORMAT ~:C
798 ** compiler failure when compiling functions with hairy constant
799 defaults for optional parameters.
800 ** compiler produces wrong code when MAYBE-INLINE-expanding a
801 function, which is already optimized.
803 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
804 * new platform: experimental support for SBCL x86/Darwin, including
805 MacOS X 10.4.5 on Intel.
806 * new feature: Unicode character names are now known to the system
807 (through CHAR-NAME and NAME-CHAR).
808 * new feature: the filesystem location of SBCL's core file is
809 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
810 * minor incompatible change: the contrib modules SB-POSIX and
811 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
812 this change is to make it easier to distribute
813 location-independent binaries.
814 * bug fix: as implied by AMOP, standardized classes no longer have
815 slots named by external symbols of public packages. (reported by
817 * optimization: calling functions via constant symbols -- as in
818 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
819 via the function object as in (FUNCALL #'FOO).
820 * optimization: on x86 and x86-64, the calling convention for the
821 case of dealing with an unknown number of values has been altered
822 to be friendlier to the prediction heuristics implemented,
823 particularly with reference to CALL and RET pairing. (thanks to
824 Alastair Bridgewater)
825 * optimization: on x86, the code for access to thread-local symbol
826 values and binding/unbinding of thread-local symbols is smaller.
827 (thanks to Alastair Bridgewater)
828 * enhancement: CONSTANTP is now able to determine constantness of
829 more complex forms, including calls to constant-foldable standardized
830 functions and some special forms beyond QUOTE.
831 * fixed bug: occasional GC crashes on Solaris/x86
832 * optimization: x86-64 supports stack allocation of results of simple
833 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
834 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
835 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
836 (thanks to James Knight)
837 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
838 by the standard (thanks to Stephen Compall)
840 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
841 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
842 be used for bundling the runtime and the core file into one
843 executable binary. This feature is not currently supported on all SBCL
844 platforms. (thanks to James Bielman and NIIMI Satoshi)
845 * new feature: a generational or ephemeral garbage collector is now
846 the default on the PowerPC platform (both Linux and Darwin). The
847 old Cheney (stop and copy) collector is a build-time option.
848 (thanks to Cyrus Harmon, after Raymond Toy)
849 * minor incompatible change: the method by which SBCL finds its
850 contributed modules has changed; it no longer relies on symbolic
851 links from an $SBCL_HOME/systems directory, but searches directly
852 in the subdirectories of $SBCL_HOME.
853 * enhancement: the dynamic heap size on the Linux/PPC platform is
854 markedly larger, even using the older Cheney garbage collector.
855 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
856 certain complicated string types. (reported by Gary King)
857 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
858 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
859 * fixed bug: DOCUMENTATION on structures no longer signals an
860 error if no structure type of the right name exists. (reported by
862 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
863 applicable non-standard primary or :AROUND method on
864 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
865 no longer cause unbound variable errors. (reported by Kevin Reid)
866 * fixed bug: metacircle resolution in cases where methods have slots
867 added before the slots from STANDARD-METHOD. (reported by Jean
869 * fixed bug: the Power PC assembly code for calling into lisp
870 sometimes computed its return address wrongly (depending on the
871 15th bit of the address, fixed at link time). This used to
872 manifest itself in a segmentation violation while building PCL.
873 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
874 many others over the years)
875 * fixed bug: ROOM no longer reports silly numbers for stack usage if
876 the stack pointer (treated as a signed integer) is negative.
877 (thanks to Peter van Eynde)
879 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
880 * new platform: experimental support for the Windows operating
881 system has been added. (thanks to Alastair Bridgewater)
882 * fixed several bugs in and robustified the PPC FFI (including
883 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
884 * bug fix: allow non-simple string symbol names (reported by Paul
886 * bug fix: interrupt handling on NetBSD (thanks to Richard M
888 * bug fix: saving a core corrupted callbacks on x86/x86-64
889 * bug fix: closed a loophole in metacircularity detection and
890 grounding in the PCL implementation of CLOS.
891 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
892 (reported by tomppa on #lisp)
893 * bug fix: clear the direction flag on Lisp -> C transitions, as
894 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
895 (reported by Andrej Grozin and Hendrik Maryns)
896 * optimization: major improvements to GC efficiency on GENCGC platforms
897 * optimization: faster implementation of EQUAL
898 * optimization: emit more efficient opcodes for some common
899 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
900 * optimization: several other minor code-generation improvements on x86-64
901 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
903 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
904 * minor incompatible change: (SETF CLASS-NAME) and (SETF
905 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
906 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
907 by ANSI as a generic function, and for consistency (SETF
908 GENERIC-FUNCTION-NAME) is treated likewise.
909 * fixed bug #233.b: make constraint propagation notice when a variable
910 value is changed after it is referenced but before it is used
911 * fixed bug #296: no more arbitrary behaviour from filenames with
912 odd characters as --load arguments. (reported by Adam Warner)
913 * fixed bug #390: :CHARACTER-SET pathname components now work as
914 expected. (reported by Tim Daly Jr)
915 * fixed bug #391: complicated :TYPE intersections in slot
916 definitions no longer cause an error in PCL internals.
917 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
918 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
919 returning the number of octets which would be written to the
920 file-stream. (thanks to Robert J. Macomber)
921 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
922 arguments to the internal stream functions. (thanks to David
924 * bug fix: stack exhaustion is now handled on the x86/SunOS
925 platform. (thanks to Daisuke Homma)
926 * bug fix: more accurate ROOM results on GENCGC platforms
927 * optimization: improved type inference for arithmetic-for
928 index variables in LOOP
929 * optimization: faster floating-point SQRT on x86-64
930 * fixed some bugs revealed by Paul Dietz' test suite:
931 ** DOCUMENTATION returns NIL instead of "" for method combinations
932 that don't have a docstring
934 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
935 * minor incompatible change: (SETF CLASS-NAME) and (SETF
936 GENERIC-FUNCTION-NAME) are no longer generic functions, and
937 therefore are not customizeable by user code (as seems to be at
938 least permitted and maybe required by AMOP). As a consolation,
939 however, the SBCL implementation of these functions now calls
940 REINITIALIZE-INSTANCE as specified by AMOP.
941 * bug fix: slot-definition documentation is propagated as per ANSI
942 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
943 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
944 argument, as do eslotd initialization methods. (from Pascal
945 Costanza's "Closer" project)
946 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
947 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
949 * bug fix: it is now possible to have more than one subclass of
950 STANDARD-GENERIC-FUNCTION without causing stack overflow.
951 (reported by Bruno Haible, Pascal Costanza and others)
952 * bug fix: the dependent update protocol now works for generic
953 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
955 * bug fix: condition-class instances corresponding to
956 DEFINE-CONDITION forms are now created eagerly. (reported by
957 Kalle Olavi Niemitalo on comp.lang.lisp)
958 * bug fix: floating point printing is more accurate in some
959 circumstances. (thanks to Simon Alexander)
960 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
961 merged with *DEFAULT-PATHNAME-DEFAULTS*.
962 * bug fix: callbacks on OS X now preserve stack-alignment.
963 (thanks to Cyrus Harmon)
964 * enhancement: the x86-64 disassembler is much better at
965 disassembling SSE instructions. (thanks to Lutz Euler)
966 * enhancement: CHECK-TYPE on a variable now allows type-inference
967 (thanks to Vincent Arkesteijn)
968 * optimization: improved performance of EUC-JP external format.
969 (thanks to NIIMI Satoshi)
970 * optimization: performance improvements to IO on file streams of
971 :ELEMENT-TYPE CHARACTER
972 * optimization: much faster memory allocation on x86-64
973 * optimization: faster garbage collections (latency and throughput)
975 * optimization: faster bignum subtraction and division on x86 and x86-64
976 (thanks to Lutz Euler)
977 * optimization: more accurate type-inference for loops that use a
978 floating point index variable or a negative step.
980 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
981 * bug fix: add a workaround to SBCL looping infinitely at startup on
982 Linux kernels with apparently buggy implementations of personality().
983 (thanks to Svein Ove Aas)
984 * bug fix: Unicode symbols are correctly printed in LDB backtraces
985 (thanks to David Lichteblau)
986 * bug fix: local bindings shadow global symbol macros for
987 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
988 * optimization: non-open coded uses of numeric comparison operators
989 (e.g. >) no longer cons when called with more than one parameter
990 on platforms supporting dynamic-extent allocation.
991 * enhancement: saving cores with foreign code loaded is now supported
992 on MIPS/Linux in addition to the previously supported platforms.
993 * bug fix: FIND-RESTART now tests for activity, not applicability when given
994 a restart object as identifier. (reported by Helmut Eller for CMUCL)
995 * bug fix: division by zero in sb-sprof when no samples were collected
996 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
997 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
998 (thanks to Alexander Kjeldaas)
999 * new feature: ignoring the return values of destructive functions
1000 like NREVERSE signals a compile-time style-warning.
1001 (thanks to Kevin Reid)
1002 * new feature: the alignment of alien structure fields can be
1003 explicitly specified. (thanks to Cyrus Harmon)
1004 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1005 and dump core on SIGQUIT
1007 ** incompatible change: threads do not inherit values of specials
1008 from their parents (see manual)
1009 ** bug fix: threads stacks belonging to dead threads are freed by the
1010 next exiting thread, no need to gc to collect thread stacks anymore
1011 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1012 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1013 ** bug fix: dynamic variable and thread start related gc lossage
1014 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1016 * fixed some bugs revealed by Paul Dietz' test suite:
1017 ** SUBTYPEP is slightly more accurate on heinously complicated
1018 CONS types where some of the members have uncertain (in the
1019 NIL, NIL sense) type relationships to each other.
1021 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1022 no longer purified unless :PURIFY T is explicitly specified.
1023 ** Non-purified cores are significantly smaller than before
1025 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1026 * new feature: timers based on Zach Beane's excellent timer package
1027 * added support for the following external formats: koi8-u,
1028 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1029 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1030 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1031 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1032 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1033 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1034 * incompatible change: a threaded SBCL will no longer revert to
1035 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1036 * bug fix: interrupts are disabled until startup is complete; no
1037 more sigsegvs when receiving a signal to soon
1038 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1040 * bug fix: add a workaround for the memory randomization features in
1041 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1042 workaround will only be in effect on systems with the proc filesystem
1044 * bug fix: printing objects of type HASH-TABLE signals a
1045 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1047 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1048 have been running for over 50 days. (reported by Gilbert Baumann)
1049 * bug fix: the logic for getting names of functions gets less
1050 confused when confronded with alternate-metaclass
1051 funcallable-instances. (reported by Cyrus Harmon)
1052 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1053 consistent, even on internal alternate-metaclass objects.
1054 * bug fix: flush closure information collected by physical
1055 environment analysis prepass before the main pass. (bug reported
1057 * bug fix: compiler pack phase does not modify a hashtable, which is
1058 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1059 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1060 may be expanded. (reported by Alexander <malishev> on c.l.l)
1061 * bug fix: no more style-warnings when compiling code using LAST
1062 under high-SPEED low-SPACE settings. (reported by David Wallin
1064 * fixed some incompatibilities between SBCL's MOP and the MOP
1066 ** the METAOBJECT class is now implemented;
1067 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1068 STANDARD-OBJECT, as required;
1069 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1070 now compatible (as required by VALIDATE-SUPERCLASS); there
1071 remains a consistency requirement over the presence or absence
1072 of the FUNCTION class in the superclass list of finalized
1073 classes; see the manual for more details;
1074 ** the :AROUND method for COMPUTE-SLOTS on
1075 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1076 requested slot ordering.
1078 ** bug fix: parent thread now can be gc'ed even with a live
1080 ** bug fix: binding a special with PROGV to no value is not
1081 the same as that symbol not having been bound (thanks to
1083 ** bug fix: binding specials is thread safe (thanks to
1085 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1086 and SB-SPROF do not die with 'no handler for signal XX in
1087 interrupt_handle_now(..)' anymore
1088 ** bug fix: WITH-TIMEOUT works with multiple running threads
1089 * fixed some bugs revealed by Paul Dietz' test suite:
1090 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1091 the :method-class keyword argument.
1093 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1094 * new port: the Solaris operating system on x86 processors is now
1095 mostly supported, though some rough edges in the environment
1096 remain. (thanks to Daisuke Homma)
1097 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1098 space for its heap. (thanks to Thiemo Seufer)
1099 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1100 closures. (thanks to Thiemo Seufer)
1101 * minor incompatible change: eof selects abort in the debugger.
1102 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1103 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1105 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1106 deprecated, and will go away in a future revision of SBCL.
1107 * minor incompatible change: GC-ON and GC-OFF are no longer
1108 implemented with a counter, it does not matter how many times gc
1109 is switched on or off
1110 * bug fix: discriminating functions for generic function classes
1111 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1112 make invalid assumptions about method precedence order. (reported
1114 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1115 platforms. (thanks to Thiemo Seufer)
1116 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1117 * bug fix: correct pathname printing with printer escaping is on.
1118 (thanks to Kevin Reid)
1119 * bug fix: complex VOP definitions in "user-space" no longer trigger
1120 package locks. (reported by Zach Beane)
1121 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1122 for funcallable-instance functions, and is no different from
1124 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1125 characters in the current readtable. (reported by Nicholas Neuss)
1126 * bug fix: SUBTYPEP on various CONS types returns more a more
1127 accurate acknowledgment of its certainty.
1128 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1129 lists in some cases. This partially fixes bug 384.
1130 * flush all standard streams before prompting in the REPL and the
1132 * bug fix: signal handling and triggering gc do not conflict
1133 directly anymore, in particular a high frequency sb-sprof does
1134 not prevent gc from running
1135 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1136 approximation for timezone and DST information between the
1137 universal time epoch and the smallest negative 32-bit time_t.
1138 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1139 year 1899 so long as the encoded time is non-negative.
1140 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1141 the correct answer for non-constant rotate amounts when performing
1142 an inline 32-bit rotation.
1144 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1145 there is only one thread in the session
1146 ** bug fix: memory leak for streams created in one thread and
1147 written to in another
1148 ** bug fix: lockup when compiled with gcc4
1149 ** bug fix: race that allows the gc to be triggered when gc is
1151 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1152 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1154 ** bug fix: finished threads are gc'ed properly
1155 * fixed some bugs revealed by Paul Dietz' test suite:
1156 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1157 has arguments with a different set of applicable methods from
1158 the orignal arguments.
1159 ** The type error thrown by MAP now has a correct expected-type
1161 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1162 name a compiled function.
1163 ** Three MISC tests where a large bignum was improperly coerced to
1164 a float that couldn't represent that bignum during type
1165 derivation were fixed.
1166 ** SUBTYPEP can now handle types involving the negation of a
1167 list-form FUNCTION type.
1168 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1169 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1170 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1172 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1173 * New feature: Experimental support for bivalent streams: streams
1174 opened with :element-type :default now allow character and binary
1175 (unsigned-byte 8) I/O
1176 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1177 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1179 * bug fix: no more highly sporadic "couldn't check whether ~S is
1180 readable" when reading a stream and an interrupt hits in the middle
1181 of a select system call
1182 * compiler better recognizes complex arrays (reported by Eduardo
1184 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1185 for variables on SBCL built with linkage-tables. (reported by Luis
1187 * various error reporting improvements.
1188 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1189 (Thanks to Thiemo Seufer)
1190 * bug fix: nested reader invokations work correctly
1191 * bug fix: it is possible to have simultaneous references to foreign
1192 code and foreign data with the same name.
1194 ** added x86-64 support
1195 ** incompatible change: the threading api now works with thread
1196 objects instead of thread ids
1197 ** bug fix: threads are protected from signals and interruption when
1198 starting up or going down
1199 ** bug fix: a race where an exiting thread could lose its stack to gc
1200 ** bug fix: don't halt on infinite error in threads if possible
1201 ** fixed numerous gc deadlocks introduced in the pthread merge
1202 ** bug fix: fixed thread safety issues in read and print
1203 ** bug fix: debugger doesn't hang on session lock if interrupted at
1204 an inappropriate moment
1205 ** bug fix: run-program is now thread safe(r)
1206 ** bug fix: inner with-recursive-lock no longer releases the mutex
1207 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1208 * fixed some bugs revealed by Paul Dietz' test suite:
1209 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1210 EXPECTED-TYPE slots filled.
1211 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1213 ** MAKE-CONDITION accepts classes as type-designators.
1214 ** COMPILE may never return NIL.
1215 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1216 range before calling Unix time functions
1218 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1219 * numerous signal handling fixes to increase stability
1220 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1221 * minor incompatible change: we now correctly canonize default
1222 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1223 as per the MOP, rather than the historical (INITARG INITFUNCTION
1224 INITFORM). (reported by Bruno Haible)
1225 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1226 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1227 for more information.
1228 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1229 pathname is a directory pathname.
1230 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1231 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1233 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1234 :METACLASS argument in addition to a class name. (reported by
1235 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1236 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1237 (reported by Bruno Haible; patch from Gerd Moellmann)
1238 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1240 * bug fix: more cleanups to the floating point exception handling on
1241 x86-64 (thanks to James Knight)
1242 * bug fix: the compiler does not try to constant fold calls of
1243 COERCE and BIT-* functions when they return freshly constructed
1244 objects. (reported by Nikodemus Siivola and Paul Dietz)
1245 * optimization: improved the MIPS versions of generic (in the
1246 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1247 * optimization: direct conversion of (unsigned-byte 32) to floats on
1248 the PowerPC platform.
1249 * optimization: structure instances with raw slots now use less
1250 memory, and probably show better memory locality. (thanks to
1252 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1253 are treated as requests for stack allocation on the x86-64,
1254 Alpha-32, PPC and SPARC platforms.
1255 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1256 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1257 (thanks to Tony Martinez)
1259 ** gcing a dead thread can no longer lead to lockups
1260 ** threads block signals until they are set up properly
1261 ** errno is no longer shared by threads
1262 ** interrupt-thread restores the eflags register on x86
1263 ** fixed some lockups due to gc/thread interaction
1264 * fixed some bugs revealed by Paul Dietz' test suite:
1265 ** invalid dotted lists no longer raise a read error when
1266 *READ-SUPPRESS* is T
1267 ** signal an error if a symbol that names a declaration is used
1268 as the name of a type, or vice versa
1269 ** allow using the (declare (typespec var*)) abbreviation for
1270 (declare (type typespec var*)) with all type specifiers
1271 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1272 ** always return NIL from PPRINT-POP when OBJECT is NIL
1273 ** don't signal errors when pretty-printing malformed LABELS,
1274 FLET or MACROLET forms
1275 ** declarations in a DOLIST body are also in scope for the
1277 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1279 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1282 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1283 * fixed cross-compiler leakages that prevented building a 32-bit
1284 target with a 64-bit host compiler.
1285 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1286 opened with :IF-EXISTS :OVERWRITE.
1287 * fixed bug 281: error for an invalid qualifier in a short-form method
1288 combination method is not signalled until the faulty method is called.
1289 * bug fix: iteration variable type inferrer failed to deal with open
1290 intervals. (reported by Alan Shields)
1291 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1292 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1293 or more runtime options were provided to the sbcl binary.
1294 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1295 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1296 * optimization: there's now a fast-path for fixnum arguments in the
1297 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1298 * optimization: the code generated on x86-64 is more compact thanks
1299 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1300 * bug fix: floating-point exception handling now partly works on
1301 x86-64. (Thanks to James Knight)
1302 * improvement to the MIPS backend from Thiemo Seufer: C-style
1303 64-bit long long arguments and return values to alien functions
1305 * fixed some bugs revealed by Paul Dietz' test suite:
1306 ** the type-error signalled from WARN has a filled-in DATUM slot.
1307 ** the type-error required when a stream is not associated with
1308 a file has the stream as its datum.
1309 ** type-errors on single-floats on x86-64 no longer have
1310 :INVALID-OBJECT as the datum
1311 ** the type-errors signalled for invalid function names now have
1312 a correct expected type
1313 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1314 for structures defined with a :TYPE. Documentation strings for
1315 typed structures are no longer immediately discarded
1316 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1317 broadcast streams and synonym streams. FILE-LENGTH now also works
1318 on broadcast streams.
1320 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1321 * incompatible change: the --noprogrammer option, deprecated since
1322 version 0.7.5, has been removed. Please use the equivalent
1323 --disable-debugger option instead.
1324 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1326 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1327 inoperational for a while now) has been completely removed.
1328 * null lexical environments are now printed as #<NULL-LEXENV>,
1329 significantly reducing the amount of clutter in typical backtraces.
1330 * documentation on weak pointers, finalization, and after GC hooks
1331 has been added to the manual.
1332 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1333 as other specialized array types, is much faster. SUBSEQ and
1334 COPY-SEQ on such arrays have also been sped up.
1335 * optimization: EQL is now more efficient when at least other argument
1336 is known to be of type (OR FIXNUM (NOT NUMBER)).
1337 * fixed bug: the runtime is now less vulnerable to changes in the
1338 size of the SBCL object on OS X, and virtual memory is reserved for
1339 all spaces that need to be at a fixed address.
1340 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1341 * fixed bug: finalizers and after GC hooks that cause consing are now
1343 * fixed bug: compiler error messages and summaries are now printed to
1344 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1345 * fixed inference of the upper bound of an iteration variable.
1346 (reported by Rajat Datta).
1347 * fixed bug 211e: calling local functions with duplicated constant
1348 keyword argument no longer causes a bogus style warning about an
1350 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1351 ftype declaration to be disregarded. (reported by Dave Roberts)
1352 * fixed bug 373: caused by erronous compilation of references to alien
1353 variables in the runtime on ppc/darwin.
1354 * fixed bug 376: CONJUGATE type deriver.
1355 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1356 (reported by Baughn on #lisp)
1357 * a cross-compiler bug on non-x86 platforms has been identified and
1358 fixed. (thanks to Bruno Haible)
1359 * improvements to the MIPS runtime code for increased stability.
1360 (thanks to Thiemo Seufer)
1361 * increased the maximimum compact environment size to allow
1362 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1363 * partial workaround for bug 135: don't name the function we're
1364 calling for hairy cases of EVAL, so as not to accumulate one environment
1365 entry per call to EVAL. (reported by Kevin Reid)
1366 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1367 * optimization: type testing for non-vector arrays should be faster.
1368 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1370 * bug fix: setting 31st element of a bit vector to zero did not work
1372 * bug fix: redefining a class definition which failed due to a
1373 previous accessor / function clash now works (but see BUGS entry
1374 #380 for more problems in this area). (thanks to Zach Beane)
1375 * the long form of DEFINE-METHOD-COMBINATION disables method group
1376 checking when given a single method group with pattern *. (thanks
1378 * on x86 compiler supports stack allocation of results of simple
1379 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1380 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1381 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1382 * fixed some bugs related to Unicode integration:
1383 ** the restarts for recovering from input and output encoding
1384 errors only appear when there is in fact such an error to
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** MISC.549 and similar: late transformation of unsafe type
1388 assertions into derived types caused unexpected code
1390 ** SCALE-FLOAT type deriver is less wrong.
1391 ** type derivers for EXP, LOG and similar functions compute result
1392 types for complex arguments better.
1393 ** (MISC.563) CONJUGATE type deriver works for very restricted
1395 ** out-of-line type testers for character strings are available.
1396 ** EQUAL compiler transform understands specialness of objects
1398 ** accessing double-floats stored on the stack now works on x86-64.
1399 ** debugger internals could sometimes create invalid lispobjs,
1400 resulting in GC crashes.
1401 ** MISC.548: type check weakening can convert required type into
1403 ** initialization forms for bindings are not in scope of free special
1406 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1407 * incompatible change: thread support for non-NPTL systems has
1408 been removed - locking is buggy and unreliable. A threaded
1409 SBCL build will now warn at startup and refuse to create
1410 new threads, unless futex support is detected in the kernel
1411 * incompatible change: the top level REPL now has only an ABORT
1412 restart associated with it, not TOPLEVEL and ABORT as it used to.
1413 TOP and TOPLEVEL are now available as debugger commands for
1414 returning to the top level.
1415 * incompatible change: forms evaluated in the REPL now use the
1416 global optimization policy.
1417 * incompatible change: user- and system-initialization files are
1418 no longer processed with LOAD, but by READ and EVAL; hence the
1419 global optimization policy, startup package, readtable, etc,
1421 * The .fasl file format number has been incremented because of
1422 various incompatible changes.
1423 * internal entry point details and argument counts no longer appear
1424 in backtraces unless explicitly requested by setting
1425 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1426 * built-in and standard functions no longer have names like "top
1427 level local call to FOO".
1428 * fixed bug 32: functions defined in non-null lexical environments
1429 now have more legible printed representation
1430 * fixed bug 33: functions defined in non-null lexical environemnts
1431 are now more amenable to inspection by INSPECT.
1432 * workaround for bug 354: XEPs no longer appear in backtraces unless
1433 explicitly requested.
1434 * fixed bug: receiving the signal which results from attempting to
1435 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1436 SIGBUS on other *BSDs) on architectures where the C stack is also
1437 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1438 to debugger confusion or wild instability. (reported by Cheuksan
1440 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1441 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1442 (reported by Lutz Euler)
1443 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1444 compiled code. (reported by Paul Dietz)
1445 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1446 specialized parameter to a method to something that is not TYPEP
1447 the specializer is now possible.
1448 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1449 face of package deletion.
1450 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1451 pathname. (reported by David Tolpin on c.l.l)
1452 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1453 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1454 STRING-STREAM and FILE-STREAM, respectively.
1455 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1456 than previously; in addition, its test suite is now run on build.
1457 (thanks to Luke Gorrie)
1458 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1459 * optimization: added a immediate representation for single-floats
1461 * fixed some bugs revealed by Paul Dietz' test suite:
1462 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1463 ** attempting to create a package with a colliding nickname causes
1464 correctable errors to be signalled.
1465 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1466 ** several x86-64 backend bugs related to sign-extension of immediate
1469 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1470 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1471 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1473 * fixed loading of multiply forward-referenced layouts.
1474 (thanks to Cheuksan Wang)
1475 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1476 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1477 (see COMPILE-FILE documentation for details of :PRINT option)
1478 * fixed bugs 19 and 317: fixed-format floating point printing is
1479 more accurate. This also fixes a bug reported by Adam Warner
1480 related to the ~@F format directive.
1481 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1483 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1484 dispatch table if the from-char is a dispatch macro character.
1485 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1486 by the wordsize no longer ignore the last word. (reported by Lutz
1488 * fixed bug in type checking of dynamic-extent variables. (reported
1490 * optimization: sequence traversal functions use their freedom to
1491 coerce function designators to functions.
1492 * optimization: code with many calls to CLOS methods specialized on
1493 CLOS classes has had redundant type checks removed.
1494 * optimization: added declarations to speed up operations that access
1495 the internal character database (for example STRING-UPCASE)
1496 * optimization: comparison operations between floats and sufficiently small
1497 fixnums no longer create extra rationals
1498 * fixed some bugs related to Unicode integration:
1499 ** portions of multibyte characters at the end of buffers for
1500 character-based file input are correctly transferred to the
1501 start of the buffer at the next read.
1502 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1503 passing it through to OPEN.
1504 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1505 argument given to internal calls to OPEN.
1506 ** the built-in comment readers (introduced by character sequences
1507 ";" and "#|") are more forgiving to encoding errors; they will
1508 STYLE-WARN and then attempt to resync the stream at a character
1509 boundary. (thanks to Teemu Kalvas)
1510 * fixed some bugs in the x86-64 port:
1511 ** Negative short int return values from c-calls are sign-extended
1513 ** The stack is aligned to 16-bytes for c-calls, as required by
1514 the ABI. (thanks to Cheuksan Wang)
1515 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1516 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1517 for enums. (thanks to Vincent Arkesteijn)
1518 ** Multiple small optimizations and bugfixes for floating point
1520 * fixed some bugs revealed by Paul Dietz' test suite:
1521 ** Space, Tab, Linefeed, Return and Page have the invalid
1522 secondary constituent character trait.
1523 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1525 ** WITH-INPUT-FROM-STRING should only update the index place on
1527 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1528 works more reliably.
1529 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1530 with one-dimensional arrays with fill pointers.
1531 ** TYPE-OF failed on a complex with an integer realpart and a
1533 ** compiler failure during type inference for the code of form
1534 (IF (EQL X (THE ...)) ...) (MISC.535).
1536 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1537 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1538 platform. The port supports 61-bit fixnums, large memory spaces
1539 and reloading shared object files.
1540 * enhancement: saving cores with foreign code loaded is now
1541 supported on ppc/Darwin in addition to the previously supported
1543 * enhancement: the statistical profiler now walks deeper into the
1544 call stack for more meaningful call-graphs and accrued time
1545 reports (x86/x86-64 only). It also now reports time spent in
1547 * enhancement: it is now possible to trace most individual methods
1548 of a generic function in addition to tracing the generic function
1550 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1552 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1553 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1554 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1555 produces an error. (thanks to Vincent Arkesteijn)
1556 * bug fix: NAMESTRING on pathnames with :WILD components in their
1557 directories works correctly. (thanks to Artem V. Andreev)
1558 * fixed bug 125: compiler preserves identity of closures. (reported
1560 * bug fixed: functions with &REST arguments sometimes failed with
1561 "Undefined function" when compiled with (DEBUG 3). (reported by
1563 * bug fix: overflow during compiling of setting element of a bit
1564 vector with constant index and value. (reported by Timmy Douglas)
1565 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1566 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1567 * on x86 compiler partially supports stack allocation of dynamic-extent
1569 * GO and RETURN-FROM do not check the extent of their exit points
1570 when compiled with SAFETY 0.
1571 * fixed some bugs related to Unicode integration:
1572 ** encoding and decoding errors are now much more robustly
1573 handled; it should now be possible to recover even from invalid
1574 input or output to the terminal. (thanks to Teemu Kalvas)
1575 ** provided a first cut at implementing STRING-TO-OCTETS and
1576 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1577 ** altered the SB-MD5 contributed module slightly, changing the
1578 interface just enough for it to be supportable for builds where
1579 lisp characters are not eight bits.
1580 * fixed some bugs revealed by Paul Dietz' test suite:
1581 ** the FORMATTER-generated functions for ~V[ conditionals require
1582 the correct number of arguments.
1583 ** READ-FROM-STRING returns the mandated second value when applied
1584 to displaced strings.
1585 ** the #\Rubout and #\Backspace characters are treated as invalid
1586 constituent characters by the tokenizer.
1588 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1589 * new feature: reloading changed shared object files with
1590 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1591 * new feature: references to foreign variables and functions
1592 can now be compiled and loaded before the corresponding shared
1593 object file is loaded, as long as the foreign definitions are
1594 available at runtime.
1595 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1596 supported. (thanks to Dan Debertin)
1597 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1598 just keywords. (thanks to Vincent Arkesteijn)
1599 * fixed bug #331: structure-class instances corresponding to
1600 DEFSTRUCT forms are now created eagerly.
1601 * fixed bug #345: backtraces from calls to undefined functions work
1602 on x86 as well. Related bug #61 is now also partially fixed on x86
1603 (backtraces from throws to unknown catch tags.)
1604 * bug fix: lambda-list parsing is now stricter vrt. order and number
1605 of lambda-list keywords.
1606 * bug fix: as specified by AMOP, an error is signalled if a
1607 class-option appears multiple times in a DEFCLASS form. (reported
1609 * bug fix: RANDOM can be compiled when the compiler derives the type
1610 of its numeric argument as a disjoint set of small integers.
1611 (reported by Paul Dietz)
1612 * bug fix: starting a core saved with shared objects loaded when
1613 those objects are not available no longer causes threaded SBCL to
1614 hang. (reported by Sean Ross)
1615 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1617 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1618 argument, which must always be NIL. (reported by Kalle Niemitalo)
1619 * bug fix: printing 1.0d+23 no longer results in an error.
1620 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1621 * bug fix: structure slot setters preserve evaluation order. (thanks
1623 * fixed some bugs related to Unicode integration:
1624 ** RUN-PROGRAM can allow its child to take input from a Lisp
1625 stream. (reported by Stefan Scholl)
1626 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1627 locales. (reported by Ken Causey)
1628 * fixed some bugs revealed by Paul Dietz' test suite:
1629 ** INCF, DECF and REMF evaluate their place form as specified in
1631 ** FORMATTER expands ~{ ~} iteration directives with V or #
1632 parameters correctly.
1633 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1634 consequent uses no arguments correctly.
1635 ** the system has a partial understanding of the (COMPLEX RATIO)
1637 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1640 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1641 * new feature: a build-time option (controlled by the :SB-UNICODE
1642 keyword feature, enabled by default) for building the system with
1643 support for the entire 21-bit character space defined by the
1645 * new feature: the system now has rudimentary external-format
1646 support; the primary user-visible change at this time is that
1647 characters with the high bit set (such as lower-case-e-acute) will
1648 print correctly to a terminal in a UTF-8 environment.
1649 * minor incompatible change: BASE-CHAR no longer names a class;
1650 however, CHARACTER continues to do so, as required by ANSI.
1651 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1652 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1653 has been moved to the SB-EXT package (temporarily re-exported from
1655 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1656 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1658 * fixed bug #351: better error-handlind and reporting for malformed
1660 * fixed bug #350: bignum-printing is now more memory-efficient,
1661 allowing printing of very large bignums, eg. (expt 2 10000000).
1662 (reported by Bruno Haible)
1663 * fixed bug #302: better primitive-type selection for intersection
1665 * fixed bug #308: non-graphic characters now all have names, as
1666 required. (reported by Bruno Haible)
1667 * bug fix: redefining a class with different superclasses now correctly
1668 removes it from the direct-subclasses of its previous superclasses.
1669 (reported by David Morse)
1670 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1671 new value now works. (reported by Bruno Haible)
1672 * bug fix: correct canonicalization of multiple non-standard slot
1673 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1674 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1675 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1676 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1677 now exists, an signals an error.
1678 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1679 during macroexpansion for non-iteration variables as well. (reported
1680 by Bruno Haible for CMUCL)
1681 * bug fix: Cyclic structures and unprintable objects in compiler
1682 messages no longer cause errors. (reported by Bruno Haible)
1683 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1684 and READ-FROM-STRING all now return a primary value of NIL if
1685 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1686 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1687 (reported by Bruno Haible for CMUCL)
1688 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1689 do not designate a condition. (reported by Bruno Haible for
1691 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1692 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1693 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1694 permitting resolution in favour of any of the conflicting symbols.
1695 (reported by Bruno Haible for CMUCL)
1696 * bug fix: EQUAL compiler optimizations is less aggressive on
1697 strings which can potentially compare true despite having distinct
1698 specialized array element types.
1699 * bug fix: unit enumerations can be defined without dividing by
1700 zero. (thanks to Vincent Arkesteijn)
1701 * FORMAT compile-time argument count checking has been enhanced.
1702 (report from Bruno Haible for CMUCL)
1703 * a partial workaround for the bug 262: the compiler does not try to
1704 inline-expand a local function doing RETURN-FROM from a deleted
1705 BLOCK. (thanks to Peter Denno for the bug report and to David
1706 Wragg for the simple test case)
1707 * fixed some bugs revealed by Paul Dietz' test suite:
1708 ** RENAME-PACKAGE allows all package designators as new package
1710 ** constraint propagation and lambda variable substitution are
1711 more cautious in dealing with partially deleted code.
1712 ** compiler sometimes lost reoptimization passes.
1713 ** CERROR, given a condition as condition designator, passes its
1714 remaining arguments to the continue format control without
1716 ** Case-altering FORMAT directives work correctly on non-ASCII
1718 ** The REAL type specifier handles bounds outside the floating
1719 point ranges without signalling FLOATING-POINT-OVERFLOW.
1720 ** Functions with IR1-transformations can create intercomponent
1721 references to global functions.
1722 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1724 ** FORMAT ~R treats a nil value for its first parameter correctly.
1726 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1727 * enhancement: saving cores with foreign code loaded is now
1728 supported on x86/NetBSD and sparc/Linux in addition to the previously
1729 supported platforms.
1730 * bug fix: on some platforms repeated installations caused multiple
1731 copies of HTML documentation to be installed -- should not happen
1732 any more. (reported by Stefan Scholl)
1733 * bug fix: parsing self-recursive alien record types multiple times
1734 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1735 original patch by Helmut Eller for CMUCL)
1736 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1737 (thanks to Richard Kreuter)
1738 * bug fix: defining classes whose accessors are methods on existing
1739 generic functions in other (locked) packages no longer signals
1740 bogus package lock violations. (reported by François-René Rideau)
1741 * bug fix: special variables as DEFMETHOD parameters no longer have
1742 associated bogus type declarations. (reported by David Wragg and
1744 * bug fix: read-write consistency on streams of element-type
1745 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1746 * bug fix: redefiniton of the only method of a generic function with
1747 no DEFGENERIC no longer emits a full WARNING. In addition,
1748 redefinition of generic functions with no DEFGENERIC to an
1749 incompatible lambda list now signals an error. (thanks to Zach
1751 * bug fix: DEFGENERIC now works even when there's a function of the
1752 same name in an enclosing lexical environment. (thanks to Zach
1754 * fixed compiler failure, caused by instrumenting code during
1755 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1756 * optimization: added loop analysis and improved register allocation
1757 * fixed some bugs revealed by Paul Dietz' test suite:
1758 ** POSITION on displaced vectors with non-zero displacement
1759 returns the right answer.
1760 ** (SIMPLE-STRING) is a valid type specifier for sequence
1762 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1764 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1765 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1767 ** The justification version of the FORMAT ~< directive treats
1768 non-zero minpad parameter correctly.
1770 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1771 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1772 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1773 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1774 the supported interface.
1775 * new feature: Single-stepping of code compiled with DEBUG 2 or
1776 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1777 * new feature: saving cores with foreign code loaded is now
1778 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1779 Timothy Moore's work for CMUCL)
1780 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1781 parameters to * instead of NIL if no initform is supplied.
1782 (reported by Johan Bockgaard)
1783 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1784 correctly when FUNCALL appears as the car of the form. Note:
1785 despite this FUNCALL forms are not currently subject to
1786 compiler-macro expansion. (port of Raymond Toy's fix for the
1787 same from CMUCL, reported by Johan Bockgaard)
1788 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1789 (thanks for Teemu Kalvas)
1790 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1791 case instead of terminating the iteration (thanks for Julian
1792 Squires, Sean Champ and Raymond Toy)
1793 * bug fix: incorrect expansion of defgeneric that caused a style
1794 warning. (thanks for Zach Beane)
1795 * bug fix: slot accessor effective method computation works properly
1796 for classes with multiple non-standard applicable methods on
1797 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1798 * on x86 compiler supports stack allocation of results of LIST and
1799 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1800 CMUCL implementation by Gerd Moellmann)
1801 * fixed some bugs revealed by Paul Dietz' test suite:
1802 ** FORMAT strings with both the ~<~:;~> form of the justification
1803 directive and pretty-printing directives cause an error.
1805 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1806 * incompatible change: the internal functions
1807 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1808 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1809 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1810 instead of the old functions.
1811 * new feature: on platforms where "dladdr" is available foreign
1812 function names now appear in backtraces. (based on Helmut Eller's
1814 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1815 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1817 * documentation: ASDF manual is now distributed with SBCL.
1818 * bug fix: SBCL can now load its contributed modules with REQUIRE
1819 even if the system-provided entries have been removed from
1820 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1822 * bug fix: backtraces involving undefined functions or assembly
1823 routines are more informative. (thanks to Brian Downing)
1824 * bug fix: mutually referent alien structures now work correctly.
1825 (reported by Rick Taube)
1826 * bug fix: structures defined by WITH-ALIEN can be referred to
1827 within other definitions in the same WITH-ALIEN.
1828 * bug fix: division operators (MOD, TRUNCATE and the like) with
1829 constant zero divisors and integer dividends no longer generate
1831 * bug fix: provide default methods for INPUT-STREAM-P and
1832 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1833 * bug fix: improve the detection and resolution of MOP metacycles
1834 (where effective method computation depends on the generic function
1835 having its effective method computed). (reported by Bruno Haible)
1836 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1837 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1838 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1839 * bug fix: pathnames with relative directory components can be
1840 represented relative to default pathnames.
1841 * optimization: in taking the GCD of bignums, reduce the two bignums
1842 to approximately the same size (using Euclid's algorithm) before
1843 applying the more sophisticated binary GCD. (thanks to Juho
1845 * optimization: COUNT on bitvectors now operates word-at-a-time.
1846 * optimization: ASH with a positive, but not necessarily constant,
1847 (leftwards) shift, when in a modular context, is compiled to a
1849 * fixed some bugs revealed by Paul Dietz' test suite:
1850 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1851 if the corresponding argument is NIL.
1852 ** FORMAT directives accepting a minpad parameter treat negative
1854 ** literal commas as character parameters to FORMAT directives are
1856 ** literal spaces directly after ~<Newline> directives within a
1857 format-logical-block (~:< ~@:>) do not induce :FILL-style
1858 conditional newlines.
1859 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1860 as specified: it no longer includes conditional newlines).
1861 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1863 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1864 :PER-LINE-PREFIX argument does not evaluate to a string.
1866 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1867 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1868 the manual for details; for now, package locks can be disabled by
1869 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1870 if no major problems are found then it is likely that they will be
1871 compiled in unconditionally.
1872 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1873 unsupported operators on all platforms. To load a shared library
1874 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1875 object file, link it into a shared library outside of SBCL and
1876 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1878 * bug workaround: ROOM T and the GC were not working together
1879 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1880 reporting facility aren't preserved. That reporting has been
1881 disabled, so now until and unless someone figures out how to make it
1882 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1883 * minor incompatible change: as threatened around sbcl-0.8.0, the
1884 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1885 The symbols are also exported from SB-PCL for backwards
1886 compatibility, but more so than before SB-PCL should be treated as
1887 an implementation-internal package.
1888 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1890 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1891 It is known as of this release not to work on the Alpha, however.
1892 * fixed bug #167: errors signalled due to illegal syntax in method
1893 bodies are now more legible.
1894 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1895 designators and can hence be used with TYPEP.
1896 * fixed bug #333: CHECK-TYPE now ensures that the type error
1897 signalled, if any, has the right object to be accessed by
1898 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1899 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1900 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1902 * fixed bug #334: programmatic addition of slots using specialized
1903 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1904 and :ALLOCATION :CLASS slots.
1905 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1906 float exponents. (rereported by Peter Seibel)
1907 * fixed a bug: #\Space (and other whitespace characters) are no
1908 longer considered to be macro characters in standard syntax by
1909 GET-MACRO-CHARACTER.
1910 * fixed bug: initialization of condition class metaobjects no longer
1911 causes an instance of the condition to be created. (reported by
1913 * fixed bug: it is now possible to have slots such that
1914 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1915 description is neither :INSTANCE nor :CLASS.
1916 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1917 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1918 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1919 non-standard SLOT-DEFINITION-ALLOCATION values.
1920 * fixed bugs: various race conditions handling exiting threads.
1921 CL-PPCRE's thread torture test now passes.
1922 * fixed arguably-a-bug: GC time proportional to number of threads in
1923 system even when most of them are idle
1924 * optimization: improved performance of BIT and SBIT on bit-vectors.
1925 * .fasl file incompatibility: The fasl file version number has
1926 been incremented because of changes associated with package locks.
1928 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1929 * minor incompatible change: the system no longer provides
1930 optimization hints (conditions which are TYPEP
1931 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1932 mode; these hints are still emitted when higher SPEED optimization
1934 * new contrib module: a sampling profiler (profiling by statistical
1935 sampling, rather than by instrumenting functions) is available as
1936 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1937 * the behaviour of the standard function ED is now customizeable by
1938 third parties through a hook variable: see ED's documentation
1939 string for information on the protocol.
1940 * the compiler no longer emits efficiency notes for (FUNCALL X)
1941 when the type of X is uncertain under default optimization
1943 * fixed bug 276: mutating a binding of a specialized parameter to a
1944 method to something that is not TYPEP the specializer is now
1946 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1947 exhibit the required behaviour on the x86 platform. (thanks to
1948 Peter van Eynde, Eric Marsden and Bruno Haible)
1949 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1950 even for difficult arguments. (reported by Peter Graves)
1951 * fixed bug 141a: the backquote printer now descends quoted
1953 * fixed another bug in backquote printing: no more destructive
1954 modification of the form's list structure. (reported by Brian
1956 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1957 move between its address being taken and the call to
1958 interrupt_thread, fixing a crashing race condition.
1959 * the SB-POSIX contrib implementation has been adjusted so that it
1960 no longer exhibits ridiculously poor performance when constructing
1961 instances corresponding to C structs.
1963 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1964 * minor incompatible change: the sb-grovel contrib now treats C
1965 structures as alien (in the sense of SB-ALIEN) objects rather than
1966 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1967 has implications for memory management of client code
1968 (sb-grovel-returned objects must now be manually managed) and for
1969 type safety (alien objects now have full types).
1970 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1971 used to control emission of compiler diagnostics, rather than the
1972 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1973 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1974 quality should be considered deprecated.
1975 * install.sh now installs the user manual as well
1976 * (not quite a new documentable feature, but worth considering in
1977 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1978 beginnings of a semantically meaningful condition hierarchy is
1979 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1981 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1982 designator as the defaults argument.
1983 * fixed bug: Displaced arrays whose displaced-to array has become
1984 too small now cause ARRAY-DIMENSION to signal an error, providing
1985 for safer bounds-checking. (reported by Bruno Haible)
1986 * fixed bug: DEFCLASS slot definitions with identical :READER and
1987 :WRITER names now signal a reasonable error. (reported by Thomas
1989 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1991 * fixed bug: Invalid slot specification errors now print correctly.
1992 (thanks to Zach Beane)
1993 * fixed bug 320: Shared to local slot value transfers in class
1994 redefinitions now happen corrently from superclasses as
1995 well. (reported by Bruno Haible)
1996 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1998 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1999 improper lists correctly. (reported by Bruno Haible)
2000 * fixed bug 313: source-transform for <fun-name> was erroneously
2001 applied to a call of a value of a variable with name <fun-name>.
2002 (reported by Antonio Menezes Leitao)
2003 * fixed bug 307: The obsolete instance protocol ensures that
2004 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2005 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2006 remove dead unknown-values globs from the middle of the stack.
2007 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2008 easier to resume long computations after using *BREAK-ON-SIGNALS*
2009 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2010 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2011 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2012 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2013 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2014 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2016 * fixed a (fairly theoretical) bug in string printing: if
2017 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2018 does not have array-element-type equal to the most general string
2020 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2021 function name. (thanks to Nikodemus Siivola)
2022 * fixed bug: compile-time format string checker failed on
2023 non-closed ~{. (reported by Thomas F Burdick)
2024 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2025 don't warn on legal loop code involving more than one aggregate
2026 boolean. (thanks to Nikodemus Siivola)
2027 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2028 in keyword position to MAKE-PACKAGE was not being reported as
2030 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2031 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2033 * fixed bugs 280 and 312: the checking for multiple definitions in a
2034 file is less likely to become confused by uses of inline
2036 * fixed bug: the #S reader macro performs the keyword coercion
2037 specified for slot names. (reported by Kalle Niemitalo)
2038 * fixed bug: lambda lists may contain symbols whose names start with
2039 & but are not lambda-list-keywords; their occurrence triggers a
2041 * fixed bug 321: define-method-combination argument lambda lists do
2042 not cause invalid code to be generated when &optional or &aux
2043 variables have default values. (reported by Bruno Haible)
2044 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2045 have CLOS classes; this appears also to have fixed reported
2046 instability in stack exhaustion detection.
2047 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2048 MUFFLE-WARNING when no associated restart is present is now
2050 * optimization: rearranged the expansion of various defining macros
2051 so that each expands into only one top-level form in a
2052 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2054 * optimization: used a previously little-used slot in symbols to
2055 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2057 * fixed some bugs revealed by Paul Dietz' test suite:
2058 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2060 ** Fixed an optimization bug related to inheritance of initforms
2061 from local to shared slots.
2062 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2063 ** CAST optimizer forgot to flush argument derived type.
2064 ** print/read consistency on floats is now orders of magnitude
2065 more likely. (thanks also to Bruno Haible for a similar report
2067 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2068 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2069 for objects of type REAL. Make it so.
2070 ** SXHASH is specified (infelicitously) to respect similarity,
2071 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2072 it so. (thanks to Markus Ziegler)
2073 ** on the Alpha, the compiler succeeds in compiling functions
2074 returning a known number of arguments greater than 63.
2075 ** fixed handling of invalid NIL arguments in keyword position
2077 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2079 ** provided a readably-printable representation for RANDOM-STATE
2081 ** ensured that pathnames, where they have a namestring, always
2082 print using #P"..." syntax.
2084 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2085 * Support for the forthcoming 2.0 version of the NetBSD kernel
2086 running on x86 hardware has been added. (thanks to Perry
2087 E. Metzger most immediately, and others for their past work)
2088 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2089 current release version; SBCL's OpenBSD support had been broken
2090 since about the time of OpenBSD's switch to ELF binary format.)
2091 * [placeholder for DX summary]
2092 ** user code with &REST lists declared dynamic-extent, under high
2093 speed or space and low safety and debug optimization policy.
2094 * The manual has been converted to Texinfo format and the debugger
2095 chapter from the cmucl manual has been added.
2096 * A facility has been added to extract documentation strings from
2097 sbcl and store them as Texinfo-formatted snippets for inclusion in
2098 the manual (via Texinfo's @include directive)
2099 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2100 the test case to Dave Roberts)
2101 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2102 pointers. (reported by Sean Ross)
2103 * bug fix: PROFILE output is printed nicely even for large numerical
2104 values. (thanks to Zach Beane)
2105 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2106 greater than 32 handle EOF correctly.
2107 * bug fix: on X86 an immediate argument of the IMUL instruction is
2108 correctly printed in disassembly. (thanks to Lutz Euler)
2109 * bug fix: class slots in redefined classes preserve their old
2110 values. (thanks to Bruno Haible and Nikodemus Siivola)
2111 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2112 succeeds. (reported by Marco Baringer)
2113 * fixed some bugs revealed by Paul Dietz' test suite:
2114 ** READ-SEQUENCE now works on ECHO-STREAMs.
2115 ** RATIONALIZE works more according to its specification. (thanks
2118 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2119 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2120 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2121 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2122 should matter to you only if you rebind the printer control
2123 variables and then find you want different bindings in the
2124 debugger than in the ordinary execution of your program.)
2125 * The runtime build system has been tweaked to support building
2126 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2127 assembler and linker. (thanks to Nikodemus Siivola)
2128 * Unbound, undefined, undeclared variables now trigger full
2129 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2130 more often programmer error than deliberate exploitation of undefined
2132 * optimization: the hash algorithm for strings has changed to one
2133 that is less vulnerable to spurious collisions. (thanks to Juho
2135 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2136 needless bounds checking. (thanks to Juho Snellman)
2137 * optimization: implemented multiplication as a modular
2138 (UNSIGNED-BYTE 32) operation on the PPC backend.
2139 * fixed some bugs revealed by Paul Dietz' test suite:
2140 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2142 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2143 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2144 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2146 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2148 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2150 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2151 their output stream on EOF from read.
2152 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2153 have been read to end-of-file.
2154 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2156 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2157 description of determination of which consecutive characters
2159 ** Printing the "Space" character with escaping on now yields "#\\ ",
2160 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2161 ** Reading floating-point numbers with *READ-BASE* set to a number
2162 less than 10 works correctly.
2163 ** Reading floating-point numbers with *READ-BASE* set to a number
2164 more than 10 works correctly.
2165 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2166 the readtable currently in effect.
2168 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2169 * minor incompatible change: parsing of namestrings on a physical
2170 (Unix) host has changed; numbers after the final #\. in a
2171 namestring are no longer interpreted as a version field. This is
2172 intented to be largely invisible to the user, except that the
2173 meaning of the namestring "*.*.*" has changed: it now refers to a
2174 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2175 should usually be replaced by
2176 (make-pathname :name :wild :type :wild :version :wild)
2177 with the added benefit that this is more likely to be portable.
2178 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2179 option to OPEN now signals an error if the file being opened
2180 exists; this may have an impact on existing code.
2181 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2182 signals received, on the PowerPC platforms (both Linux and
2183 Darwin). (thanks to Pierre Mai for pointing out the location of
2185 * several fixes on OS X: The system now builds and runs cleanly on
2186 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2187 (thanks to Brian Mastenbrook)
2188 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2189 second-resolution: integer multiples of 1/3600 between -24 and 24.
2190 (thanks to Vincent Arkesteijn)
2191 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2192 type when called with 1 argument; PEEK-CHAR checked type of
2193 PEEK-TYPE only after having read first character from a
2194 stream. (reported by Peter Graves)
2195 * bug fix: the garbage collector now has much better locality
2196 behaviour, and in particular no longer treats all memory as being
2197 exhausted when many small objects point to each other in a deeply
2199 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2200 recognized as being TYPEP their class.
2201 * bug fix: the PUSHNEW documentation string has been corrected.
2202 (thanks to Vincent Arkesteijn)
2203 * bug fix: defaulting of the value for the last of an atypically
2204 large number of multiple values being bound was not being
2205 performed correctly on the Alpha or PPC platforms
2206 * optimization: implemented multiplication as a modular
2207 (UNSIGNED-BYTE 32) operation on the x86 backend.
2208 * optimization: SEARCH on simple-base-strings can now be open-coded.
2209 (see also contrib/compiler-extras.lisp for inspiration for
2210 teaching the compiler about the Boyer-Moore algorithm).
2211 * value, returned by MAX (and MIN) called with several EQUALP, but
2212 not EQL, arguments now does not depend on compiler settings.
2213 * fixed some bugs revealed by Paul Dietz' test suite:
2214 ** in stack analysis liveness information is propagated from
2215 non-local entry points.
2216 ** pathwise CAST removing failed when the CAST node did not start
2218 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2219 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2221 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2222 host is already defined.
2223 ** RENAME-FILE works on streams instead of signalling an internal
2225 ** PEEK-CHAR uses the current readtable when determining whether
2226 or not a character is whitespace.
2227 ** MERGE-PATHNAMES handles the case when the pathname does not
2228 specify a name while the default-pathname specifies a version.
2229 ** Pathnames now stand a better chance of respecting print/read
2231 ** Attempting to use standardized file system operators with a
2232 pathname with invalid :DIRECTORY components signals a
2234 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2236 ** DIRECTORY on logical pathnames is more correct.
2237 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2238 signal a TYPE-ERROR if their argument is not a stream.
2239 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2240 designator argument does not designate a stream.
2241 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2242 examining the synonym.
2243 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2245 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2246 element-type larger than ([UN]SIGNED-BYTE 32).
2248 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2249 * When built with the :SB-FUTEX feature, threaded builds now take
2250 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2251 for faster/more reliable mutex and condition variable support.
2252 * Incompatible change (but one you probably shouldn't have been using
2253 anyway): the interface and code for arbitrating between multiple
2254 threads in the same user session has been redesigned.
2255 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2256 error when called without an explicit environment argument.
2257 (thanks to Nikodemus Siivola)
2258 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2259 confused about their position. (thanks to Adam Warner and Gerd
2261 * bug fix: Pretty printing backquoted forms with unquotations in the
2262 argument list position of various code constructs such as LAMBDA
2263 now works correctly. (reported by Paul Dietz)
2264 * bug fix: Pretty printing unquotations no longer loses all
2265 stream position information.
2266 * optimization: performance of string output streams is now less
2267 poor for multiple small sequence writes.
2268 * optimization: performance of CSUBTYPEP in the presence of complex
2269 expressions involving CONS and NOT many times has been improved.
2270 (reported by Paul Dietz)
2271 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2273 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2274 streams for the REPL, for the debugger, and so on. (thanks to
2276 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2277 * fixed some bugs revealed by Paul Dietz' test suite:
2278 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2279 is ignored if it is too small, rather than propagating through
2280 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2281 ** extremely complex negations of CONS types were not being
2282 sufficiently canonicalized, leading to inconsistencies in
2284 ** VALUES tranformer lost derived type.
2286 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2287 * fixed a bootstrapping bug: the build process no longer assumes
2288 that the various BOOLE-related constants have the same value in
2289 host and target lisps. (noted by Paul Dietz' test suite on an
2290 SBCL binary built from CLISP)
2291 * The system can now be dynamically linked on the MIPS platform,
2292 which enables dynamic loading of foreign code from Lisp. (thanks
2293 to Ralf Baechle for discussions on the MIPS ABI)
2294 * The system now records debugging information for its own source
2295 files in a filesystem-position-independent manner, relative to
2296 the "SYS" logical host.
2297 * fixed a compiler bug: MV-LET convertion did not check references
2298 to the "max args" entry point. (reported by Brian Downing)
2299 * tweaked disassembly notes to be less confident about proclaiming
2300 some instruction as an LRA. (thanks to Brian Downing)
2301 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2302 now each have their own history, command character, and other
2303 characteristics. (thanks to David Lichteblau)
2304 * fixed some bugs revealed by Paul Dietz' test suite:
2305 ** compiler failure in compiling LOGAND expressions including a
2307 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2308 shift greater than 32.
2309 ** FLUSH-DEST did not mark blocks for type check regeneration.
2310 ** HANDLER-CASE failed to accept declarations in handler clauses
2311 in some circumstances.
2313 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2314 * New code in contrib/sb-introspect (still probably not entirely
2315 stable yet) provides some support for smart Lisp development
2316 environments like SLIME.
2317 * The conditions signalled for errors occurring when loading .fasl
2318 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2319 in a way which should help ASDF recover gracefully.
2320 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2321 stuff has been cleaned up. If you code contrib/ stuff, this might
2322 affect you, and you can look at contrib/README, contrib/STANDARDS,
2323 and/or the 0.8.4.27 diff to check.
2324 * In full calls the compiler now does not generate checks for declared
2325 argument types for all arguments.
2326 * various threading fixes
2327 ** and some experimental patches which didn't make it into
2328 the main tree for this release, but which are shipped in
2329 contrib/experimental-thread.patch as a possible fix for some
2330 failures (deadlock, spinning...) in GC-intensive multithreaded
2332 * fixed PPC build problem (source code incompatibility of different
2333 library versions): added offsetof() hackery which attempts to divine
2334 where glibc maintainers put uc_mcontext today
2335 * fixed bug 282: compiler does not trust type assertions while passing
2336 arguments to a full call.
2337 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2338 &OPTIONAL VALUES type parameter.
2339 * fix bug 214: algorithm for noting rejected templates is now more
2340 similar to that of template seletion. (also reported by rydis on
2342 * fixed bug 141b: printing backquoted information readably and prettily
2343 inserts a space where necessary.
2344 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2345 (CHARACTER 10) are now reported as errors, rather than propagated
2346 as unknown types. (reported by piso on #lisp)
2347 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2348 with values NIL and :ERROR. (thanks to Milan Zamazal)
2349 * fixed bug 191c: CLOS now does proper keyword argument checking as
2350 described in CLHS 7.6.5 and 7.6.5.1.
2351 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2352 longer raise an error; further, using a list as a for-as-arithmetic
2353 counter now raises a meaningful error.
2354 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2355 now understood by sequence creation functions such as MAKE-SEQUENCE
2357 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2358 a STRING-INPUT-STREAM.
2359 * compiler enhancement: SIGNUM is now better able to derive the type
2361 * type declarations inside WITH-SLOTS are checked. (reported by
2363 * fixed some bugs revealed by Paul Dietz' test suite:
2364 ** incorrect optimization of TRUNCATE for a positive first
2365 argument and negative second.
2366 ** compiler failure in let-convertion during flushing dead code.
2367 ** compiler failure while deriving type of TRUNCATE on an
2368 interval, containing 0.
2369 ** ASH of a negative bignum by a negative bignum count now returns
2371 ** intersection of CONS types now canonicalizes properly, fixing
2372 inconsistencies in SUBTYPEP.
2374 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2375 * incompatible change: The --disable-debugger command line
2376 option now clobbers the debugger at a more fundamental
2377 level, by redefining #'INVOKE-DEBUGGER instead of by
2378 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2379 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2380 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2381 enter the debugger REPL and then suffer recursive errors
2382 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2383 while under the new system, BREAK in --disable-debugger mode
2384 terminates the system just as an unhandled error would.
2385 * fixed compiler performance when processing loops with a step >1;
2386 * bug fix: DOCUMENTATION now retrieves generic function
2387 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2388 support has been systematized, and now supports the methods
2389 specified by ANSI, along with a default method and a method for
2390 slot documentation. (reported by Nathan Froyd)
2391 * bug fix: effective methods associated with a generic function are
2392 no longer cached over a change of that generic function's method
2393 combination. (reported by Andreas Fuchs)
2394 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2395 names a non-existent directory. (thanks to Andreas Fuchs)
2396 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2397 the PPC platform, lead to stack corruption; this has been fixed.
2398 (reported by Rainer Joswig)
2399 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2400 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2401 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2402 arithmetic in general, is now much more reliable on the PPC
2404 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2405 platform now returns the right answer.
2406 * optimization: restored some effective method precomputation in
2407 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2408 precomputation is now tunable.
2409 * optimization: compiler-internal data structure use has been
2410 reviewed, and changes have been made that should improve the
2411 performance of the compiler by about 20%.
2412 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2413 simple-base-strings and simple-bit-vectors is improved.
2414 * optimization: the optimization of 32-bit logical and arithmetic
2415 functions introduced in version 0.8.3 on the x86 has been
2416 implemented on the mips, ppc and sparc platforms; an
2417 implementation of the same facility, but for 64-bit arithmetic,
2418 has been added for the alpha.
2419 * microoptimization: the compiler is better able to make use of the
2420 x86 LEA instruction for multiplication by constants.
2421 * bug fix: in some situations compiler did not report usage of
2422 generic arithmetic in (SPEED 3) policy.
2423 * bug 145b fix: compiler used wrong type specifier while converting
2424 MEMBER-types to numeric.
2425 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2427 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2428 index argument now works properly on the Alpha platform.
2429 * bug fix: floating point exception treatment on the Alpha platform
2431 * bug fix: FILE-POSITION works much better on string input and
2432 output streams. (thanks to Nikodemus Siivola)
2433 * bug fix: many threading/garbage collection symptoms sorted.
2434 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2435 might be pseudo-atomic.
2436 * internal change: Stopping for GC is now done with signals not
2437 ptrace. GC is now done in whichever thread wanted it, instead of
2439 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2441 * bug fix: result form in DO is not contained in the implicit
2443 * incompatible change: ICR structure is changed; the value part of
2444 CONTINUATION is now called LVAR; corresponding functions are
2445 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2446 * added type deriver for ISQRT (thanks to Robert E. Brown).
2447 * bug fix: better support for loading from the command line when an
2448 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2450 * fixed some bugs revealed by Paul Dietz' test suite:
2451 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2452 ** ROUND and FROUND now give the right answer when given very
2453 small float arguments.
2454 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2456 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2457 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2458 ** LCM should return a non-negative integer.
2459 ** PARSE-INTEGER returned the index of a terminator instead of the
2460 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2461 ** PARSE-INTEGER returned an incorrect index being applied to a
2463 ** LCM with two arguments of 0 returns 0 rather than signalling
2465 ** unsigned addition of a 32-bit constant with the high bit set no
2466 longer causes an internal compiler error on the x86.
2467 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2468 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2469 with negative last argument.
2470 ** byte specifiers with zero size and position no longer cause
2471 an error during type derivation.
2472 ** bignum multiplication on the Alpha platform now returns the
2474 * porting: The system now builds on SuSE AMD64, although it still
2475 generates a 32-bit binary.
2476 * .fasl file incompatibility: The fasl file version number has
2477 been incremented (because of the changes to internal compiler
2478 data structures referred to above).
2480 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2481 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2482 more accurately, on the Darwin kernel running on PowerPC hardware.
2483 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2484 * Compiler code deletion notes now signal a condition of type
2485 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2486 an associated MUFFLE-WARNING restart.
2487 * The compiler now performs limited argument count validation of
2488 constant format strings in FORMAT, and where appropriate in ERROR,
2489 CERROR and WARN. (thanks to Gerd Moellmann)
2490 * New ASDF-INSTALL contrib can be used for automatic download and
2491 installation of third-party Lisp code from CCLAN or other sites
2493 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2494 which forces another thread to execute a function supplied by the
2496 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2497 now accept and act upon their :ELEMENT-TYPE keyword argument.
2498 (reported by Martin Atzmueller, Edi Weitz)
2499 * bug fix: FILE-POSITION now accepts position designators up to
2500 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2501 is the greater. (thanks to Patrik Nordebo)
2502 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2503 Kalle Olavi Niemitalo)
2504 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2505 printable. (reported by Eric Marsden)
2506 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2507 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2508 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2509 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2510 (thanks to Stig Sandoe)
2511 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2512 UNREAD-CHAR now correctly includes the unread character in the
2513 target sequence. (thanks to Gerd Moellmann)
2514 * bug fix in threaded builds: the system can now be suspended and
2515 resumed by shell job control with minimal disruption.
2516 * bug fixes in times and timezones >2038AD
2517 * better handling of "where is GNU make?" problem in build scripts
2518 (thanks to Nikodemus Siivola)
2519 * new optimization: inside a named function any reference to a
2520 function with the same name is considered to be a self-reference;
2521 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2522 optimization quality.
2523 * new optimization on x86: logical functions and + now have
2524 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2525 used when the result is truncated to 32 bits.
2526 * VALUES declaration is partially enabled.
2527 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2528 * fixed some bugs revealed by Paul Dietz' test suite:
2529 ** The system now obeys the constraint imposed by
2530 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2531 types form a lattice under type intersection.
2532 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2533 ** ASSOC now ignores NIL elements in an alist.
2534 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2535 and (1+ MOST-POSITIVE-FIXNUM) answers.
2536 ** The addition of a method with invalid qualifiers to a generic
2537 function does not cause an error to be signalled immediately;
2538 a warning is signalled, and the error is generated only on
2539 calling the generic function.
2540 * changed .fasl file version number, in part to add type codes for
2541 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2542 obscure ANSI requirements
2544 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2545 * fixed bug 148: failure to inline-expand a local function left
2546 garbage, confusing the compiler.
2547 * fixed bugs 3cd: structure slot readers perform type check if the
2548 slot can have an invalid value (i.e. it is either not initialized
2549 or can be written with a less specific slot writer).
2550 * bug fix: the compiler now traps array references to elements off
2551 the end of an array; previously, the bounds checking in some
2552 circumstances could go off-by-one.
2553 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2555 * type declarations for array element types now obey the description
2556 on the CLHS page "Declaration TYPE", as per discussions on
2557 sbcl-help around 2003-05-08. This means that a declaration
2558 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2559 declaration, all references to BAR will be asserted or assumed
2560 (with THE, so dependent on compiler policy) to involve objects of
2561 type FOO. Note that no such declaration is implied in
2562 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2563 * declared types of functions from the "Conditions"
2564 chapter. (reported by Paul Dietz)
2565 * bug fix: CERROR accepts a function as its first argument.
2566 * bug fix: NTH an NTHCDR accept a bignum as index
2567 arguments. (reported by Adam Warner)
2568 * optimization: character compare routines now optimize comparing
2569 against a constant character. (reported by Gilbert Baumann)
2570 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2571 argument now works properly on the MIPS platform.
2572 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2573 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2574 declared to have a functional type, does not cause an error
2576 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2577 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2578 * fixed bug: if last continuation of a deleted block has a
2579 destination, this destination should be deleted too. (reported by
2581 * fixed a bug in the bootstrap process: the host compiler's values
2582 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2583 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2584 test case from Patrik Nordebo)
2585 * improved the ability of the disassembler on the PPC platform to
2586 provide helpful disassembly notes.
2587 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2588 the class in more cases than previously.
2589 * bug fix: FILE-POSITION now understands :START and :END for
2590 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2591 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2592 8) data. (thanks to David Lichteblau)
2593 * bug fix: it is possible to add a method to a generic function
2594 without lambda list.
2595 * bug fix: reader failed to signal END-OF-FILE inside an
2596 object representation. (reported by Nikodemus Siivola)
2597 * fixed some bugs revealed by Paul Dietz' test suite:
2598 ** LAST and [N]BUTLAST should accept a bignum.
2599 ** condition slot accessors are methods.
2600 ** (VECTOR NIL) is a subtype of STRING.
2602 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2603 * minor incompatible change: some nonsensical specialized lambda
2604 lists (used in DEFMETHOD) which were previously ignored now signal
2606 * minor incompatible change: the system is now aware of the types of
2607 variables in the COMMON-LISP package, and will signal errors for
2608 most violations of these type constraints (where previously they
2609 were silently accepted).
2610 * minor incompatible change: COMPILE-FILE now uses the freedom
2611 afforded (ANSI 3.2.2.3) to use derived function types for
2612 functions defined in the same file. This also permits the system
2613 to warn on static type mismatches and function
2614 redefinition. (Currently it does not work with high DEBUG level.)
2615 * minor incompatible change: VALUES declaration is disabled.
2616 * When issuing notes, the compiler now signals a condition of type
2617 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2618 restart for use in user handlers. It is expected that the
2619 COMPILER-NOTE condition will eventually become a condition
2620 supertype to a hierarchy of note types, which will then be
2621 handleable in a similar fashion. However, at the moment, no such
2622 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2623 but it's an implementation detail, not a classification for the
2625 * Changes in type checking closed the following bugs:
2626 ** type checking of unused values (192b, 194d, 203);
2627 ** template selection based on unsafe type assertions (192c, 236);
2628 ** type checking in branches (194bc).
2629 * A short form of VALUES type specifier has ANSI meaning (it has
2630 increased the number of situations when SBCL cannot perform type
2632 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2633 keywords or constants is permissible.
2634 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2635 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2636 classes. (thanks to Antonio Martinez)
2637 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2638 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2639 argument to a pretty-print stream. Instead, it leaves any such
2640 operations to DESCRIBE-OBJECT methods.
2641 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2643 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2644 lambda lists. (thanks to David Lichteblau)
2645 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2646 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2647 (thanks to Gerd Moellmann)
2648 * fixed bug 246: increased compilation speed of long
2649 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2651 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2653 * DEFCONSTANT now throws a condition of type
2654 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2655 constant to a non-EQL value; CONTINUE and ABORT restarts
2656 respectively change and preserve the value.
2657 * fixed bug 63: The code walker, part of the implementation of CLOS,
2658 is now better at handling symbol macros.
2659 * bug fix: There is no longer an internal implementation type named
2660 CL:LENGTH. (reported by Raymond Toy)
2661 * bug fix: In macro-like defining macros/special operators the
2662 implicit block does not enclose the lambda list.
2663 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2664 enclosed in parentheses) are not suitable as type specifiers, and
2665 their use properly signals an error now.
2666 * bug fix: An argument count mismatch for a type specifier in code
2667 being compiled no longer causes an unhandled error at compile
2668 time, but signals a compile-time warning.
2669 * fixed simple vector readable printing
2670 * bug fix: DESCRIBE takes more care over whether the class
2671 precedence list slot of a class is bound before accessing it.
2672 (reported by Markus Krummenacker)
2673 * bug fix: FORMATTER can successfully compile pretty-printer format
2674 strings which use variants of the ~* directive inside.
2675 * bug fix: SEARCH now applies its TEST predicate to the elements of
2676 the arguments in the correct order. (thanks to Wolfhard Buss)
2677 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2678 in type inference. (thanks to Robert E. Brown)
2679 * bug fix: Reading in symbols with an explicit package name of ""
2680 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2681 * fixed some bugs revealed by Paul Dietz' test suite:
2682 ** NIL is now allowed as a structure slot name.
2683 ** Arbitrary numbers, not just REALs, are allowed in certain
2684 circumstances in LOOP for-as-arithmetic clauses.
2685 ** Multiple class redefinitions before slot access no longer
2686 causes a type error.
2687 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2688 association between the name and a class.
2689 ** Generic functions with non-standard method-combination and over
2690 five methods all of which return constants no longer return NIL
2691 after the first few invocations. (thanks to Gerd Moellmann)
2692 ** CALL-NEXT-METHOD with no arguments now passes the original
2693 values of the arguments, even in the presence of assignment.
2694 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2696 ** DISASSEMBLE works with closures and funcallable instances.
2697 ** ADD-METHOD now returns the generic function, not the new method.
2698 ** FIND-METHOD signals an error if the lengths of the specializers
2699 is incompatible with the generic function, even if the ERRORP
2701 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2702 which its argument is a member.
2703 ** DEFCLASS only redefines the class named by its class-name
2704 argument if that name is the proper name of the class;
2705 otherwise, it creates a new class.
2706 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2707 of the UNBOUND-SLOT condition to the name of the slot.
2708 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2709 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2710 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2711 SLOT-MAKUNBOUND in the specified fashion.
2713 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2714 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2715 cross-compilation host. As a consequence, we can now bootstrap our
2716 way up to SBCL starting with a bare gcc toolchain and human-readable
2717 source code (first the source to CLISP, then the source to SBCL).
2718 * A contributed module containing a partial implementation of the
2719 simple-streams interface has been included. (thanks to Rudi
2721 * A contributed module implementing the RFC1321 Message Digest
2722 Algorithm, known as MD5, has been included.
2723 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2724 no longer has any effect, as the code controlled by this feature
2725 has been deleted. (As far as we know, no-one has ever built using
2726 this feature, and its semantics were confused in any case).
2727 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2728 work on conditions (as required by the ANSI specification),
2729 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2730 expected behaviour on conditions. Users should note, however,
2731 that such behaviour is not required by the ANSI specification,
2732 and so use of this behaviour may render their code unportable.
2733 * fixed some bugs revealed by Paul Dietz' test suite:
2734 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2736 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2737 ** SXHASH on condition objects no longer returns NIL.
2738 ** :ALLOCATION :CLASS slots are better treated; their values are
2739 updated on class redefinition, and initforms inherited from
2740 superclasses are applied.
2741 ** REMOVE-METHOD returns its generic function argument even when
2742 no method was removed.
2743 ** SHARED-INITIALIZE now initializes the values of the requested
2744 slots, including those with :ALLOCATION :CLASS.
2745 ** ALLOCATE-INSTANCE now works on structure classes defined via
2746 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2748 ** SLOT-EXISTS-P now works on conditions, as well as structures
2750 ** MAKE-LOAD-FORM now has the required methods on
2751 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2752 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2753 keyword, and now implements the SLOT-NAMES argument.
2754 ** methods with &OPTIONAL arguments no longer allow too many
2755 arguments to be passed in the call without error.
2756 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2757 option is consistent with the required arguments of the generic
2758 function lambda list.
2759 * bug fix: REQUIRE accepts a string designator. (Thanks to
2761 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2762 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2763 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2764 * bug fix: The compiler now checks for duplicated variables in macro
2766 * bug fix: SETQ on globals returns the correct value.
2767 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2768 not a COMPILER-ERROR (followed by some other strange error on
2769 choosing the CONTINUE restart).
2770 * bug fix: make.sh and friends are now more consistent in the way that
2771 they look for GNU "make".
2773 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2774 * experimental native threads support (on x86 Linux >=2.4 only).
2775 This is not compiled in by default: you need to add :SB-THREAD to
2776 the target features. See the "Beyond ANSI" chapter of the manual
2778 * fix for longstanding nonANSIism: The old distinction between
2779 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2780 The return value from CL:FIND-CLASS is now a CLOS class, and
2781 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2782 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2783 * An interface to the MetaObject Protocol, as described in Kiczales,
2784 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2785 MIT Press, 1991, is available from the SB-MOP package.
2786 * incompatible change: the SB-PCL package should now be considered
2787 a private implementation detail, and no longer a semi-private MOP
2789 * minor incompatible change: due to rearrangement for threads, the
2790 control stack and binding stack are now allocated at arbitrary
2791 addresses instead of being hardcoded per-port. Users affected by
2792 this probably have to be doing advanced things with shared
2793 libraries, and will know who they are.
2794 * minor incompatible change: Previously, all --eval forms used were
2795 processed with READ before any of them were processed with EVAL.
2796 Now each --eval form is processed with both READ and EVAL before
2797 the next --eval form is processed. (Thus package operations like
2798 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2799 work as the user might reasonably expect.)
2800 * minor incompatible change: *STANDARD-INPUT* is now only an
2801 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2803 * minor incompatible change: Y-OR-N-P is now character-oriented,
2804 not line oriented. Also, YES-OR-NO-P now works without errors.
2805 (thanks to Antonio Martinez)
2806 * sb-aclrepl module improvements: an integrated inspector, added
2807 repl features, and a bug fix to :trace command.
2808 * Known functions, which cannot be open coded by the backend, are
2809 considered to be able to check types of their arguments. (fixing
2810 a bug report by Nathan J. Froyd)
2811 * fixed a bug in computing method discriminating functions: It is
2812 now possible to define methods specialized on classes which have
2813 forward-referenced superclasses. (thanks to Gerd Moellmann)
2814 * fixed evaluation order in optional entries (reported by Gilbert
2816 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2817 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2818 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2819 required-by-AMOP NAME argument, as well as CLASS and
2820 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2821 * fixed bug 20: DEFMETHOD can define methods using names that are
2822 not the proper names of classes to designate class specializers.
2823 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2824 Unix file descriptors, instead of blowing up. (thanks to Antonio
2826 * Garbage collection refactoring: user-visible change is that a
2827 call to the GC function during WITHOUT-GCING will not do garbage
2828 collection until the end of the WITHOUT-GCING. If you were doing
2829 this you were probably losing anyway.
2830 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2831 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2832 TYPEP the latter but not the former.
2833 * The compiler issues a full WARNING for calls to undefined functions
2834 with names from the CL package.
2835 * MAP-INTO for a vector destination is open coded. (reported by
2836 Brian Downing on c.l.l)
2837 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2838 documentation string.
2839 * fixed some bugs revealed by Paul Dietz' test suite:
2840 ** COPY-ALIST now signals an error if its argument is a dotted
2842 ** Condition slots are now accessed more correctly in the presence
2843 of multiple initargs for a given slot.
2844 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2845 exclude restarts of the same name associated with a different
2847 ** DEFCLASS of forward-referenced classes with another
2848 forward-referenced class in the superclasses list no longer
2850 ** Condition slots are now initialized once each, not multiple
2851 times. (thanks to Gerd Moellmann)
2852 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2854 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2855 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2856 arguments contain duplicated elements.
2857 ** RESTART-CASE understands local macros.
2858 ** RESTART-CASE associates exactly its own restarts with a condition.
2859 ** ENDP in safe mode checks its argument to be of type LIST.
2860 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2861 in question is unbound.
2862 ** Optimized MAKE-INSTANCE functions no longer cause internal
2863 assertion failures in the presence of duplicate initargs.
2864 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2865 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2866 propagated even in the ctor optimized implementation of
2868 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2870 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2871 * a better implementation of SXHASH on (simple) bit vectors,
2872 measured both in execution speed and in distribution of results
2873 over the positive fixnums, has been installed. Likewise, a better
2874 implementation of EQUAL for simple bit vectors is now available.
2875 * fixed CEILING optimization for a divisor of form 2^k.
2876 * fixed bug 240 (emitting extra style warnings "using the lexical
2877 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2878 by Antonio Martinez)
2879 * fixed SXHASH, giving different results for NIL depending on type
2880 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2881 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2882 variable. (found by Rolf Wester)
2883 * SBCL does not ignore type declarations for special
2884 variables. (reported by rif on c.l.l 2003-03-05)
2885 * some bug fixes in contrib/sb-aclrepl/
2886 * fixed some bugs revealed by Paul Dietz' test suite:
2887 ** a bug in the CONS type specifier, whereby the CAR and CDR
2888 types got intertwined, has been fixed;
2889 ** the type system is now able to reason about the interaction
2890 between INTEGER and RATIO types more completely;
2891 ** APPEND, [N]REVERSE and NRECONC check that those their
2892 arguments, which must be proper lists, are really so;
2893 ** An array specialized to be unable to hold elements has been
2894 implemented, as required -- yes, really -- by ANSI;
2895 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2896 on malformed property lists;
2898 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2899 * incompatible packaging change: in line with Unix convention,
2900 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2901 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2902 * REQUIRE and PROVIDE are now optionally capable of doing something
2903 useful. See the documentation string for REQUIRE.
2904 * infrastructure for a managed SBCL contrib system: contributed
2905 modules in this release include:
2906 ** the ASDF system definition facility;
2907 ** an interface to the BSD Sockets API;
2908 ** an ACL-like convenience interface to the repl;
2909 (thanks to Kevin Rosenberg)
2910 ** an implementation of ROTATE-BYTE, with efficient implementation
2912 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2913 gives a closer approximation to the right answer than previously.
2914 (thanks to Raymond Toy)
2915 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2916 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2917 optional environment arguments, as required by ANSI.
2918 * fixed bugs in other functions taking environment objects, allowing
2919 calls with an explicit NIL environment argument to be compiled
2921 * fixed bug 228: primary return values from
2922 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2923 COMPILE or FUNCTION.
2924 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2925 :TYPE will no longer signal errors on innocuous objects.
2926 * fixed bug 231b: SETQ is better at respecting type declarations in
2927 the lexical environment.
2928 * fixed a bug in DEFCLASS: classes named by symbols with no or
2929 unprintable packages can now be defined.
2930 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2931 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2932 * fixed compiler failure related to checking types of functions.
2933 (reported by Robert E. Brown)
2934 * the compiler is now much more consistent in its error-checking
2935 treatment of bounding index arguments to sequence functions: in
2936 (SAFETY 3) code, errors will be signalled in almost all cases if
2937 invalid sequence bounding indices are passed to functions defined
2938 by ANSI to operate on sequences.
2939 * fixed a bug in the build procedure: documentation of SBCL-specific
2940 packages is now preserved and available in the final Lisp image.
2941 * lifted FDEFINITION lookup out of loops in the implementation of
2942 many list operations. (thanks to Robert E. Brown)
2943 * fixed a bug in the reader: the #n# reader macro now works for
2944 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2945 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2946 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2947 E. Brown and Tony Martinez respectively)
2948 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2949 SLOT-BOUNDP: in particular, it is now able to optimize them much
2950 better, and is now not vulnerable to having packages renamed.
2951 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2952 always called when a slot is not present in an instance. (thanks
2954 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2955 longer signal an error on unreading a character following EOF on
2956 the previous constituent stream. (thanks to Tony Martinez)
2957 * fixed some bugs revealed by Paul Dietz' test suite:
2958 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2959 not just nonnegative fixnums;
2960 ** the logical bit-array operators such as BIT-AND now accept an
2961 explicit NIL for their "opt-arg" argument (to indicate a
2962 freshly-consed result bit-array);
2963 ** ELT now signals an error on an invalid sequence index in safe
2965 ** the type system is now cleverer about negations of numeric
2966 types, and consequently understands the BIGNUM and RATIO types
2968 ** the type system is now cleverer about the interaction between
2969 INTEGER and RATIO types: while bugs still remain, many more
2970 cases are accurately computed;
2971 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2972 if it is in the last clause;
2973 ** CONSTANTLY now correctly returns a side-effect-free function in
2975 ** DECLARE is no longer treated as a special-operator; in
2976 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2977 * incremented fasl file version number due to the change in the
2978 DEFSTRUCT-SLOT-DESCRIPTION structure.
2980 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2981 * minor incompatible change: code processed by the "interpreter" or
2982 EVAL now has a compilation optimization policy of (DEBUG 2)
2983 (changed from (DEBUG 1)) to improve debuggability of interactive
2984 development, and to allow the use of the debug RETURN command in
2986 * an experimental implementation of the RETURN command for the
2987 debugger has been included. (thanks to Frederik Kuivinen)
2988 * fixed bug 62: constraints were not propagated into a loop.
2989 * fixed bug in embedded calls of SORT (reported and investigated by
2991 * fixed some bugs revealed by Paul Dietz' test suite:
2992 ** printing and reading of arrays with some dimensions having
2993 length 0 (thanks to Gerd Moellmann);
2994 ** BOA constructor with &AUX argument without a default value does
2995 not cause a type error;
2996 ** CONSTANTP now returns true for all self-evaluating objects.
2998 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2999 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3000 accessors that are related by inheritance, as specified in the
3001 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3002 Valtteri Vuorikoski)
3003 * The compiler is now able to inline functions that were defined in
3004 a complex lexical environment (e.g. inside a MACROLET).
3005 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3007 * Support for the upcoming FreeBSD-5.0 release has been included.
3008 (thanks to Dag-Erling Smorgrav)
3009 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3010 effect when it is not in a toplevel context.
3011 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3012 stand a better chance of being correct. (thanks to Gerd
3014 * fixed bug in COERCE, which now signals an error on coercing a
3015 rational to a bounded real type which excludes the expected
3017 * The compiler is now able to derive types more accurately from the
3018 COERCE and COMPILE functions.
3019 * fixed bug 223: functional binding is considered to be constant
3020 only for symbols in the CL package.
3021 * fixed bug 231: SETQ did not check the type of a variable being set
3022 (reported by Robert E. Brown)
3023 * A new optimization for MAKE-INSTANCE has been included, fixing
3024 various bugs (including relating to :ALLOCATION :CLASS slots and
3025 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3027 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3028 ** As required by ANSI, LOOP now disallows anonymous collection
3029 clauses such as COLLECT I in conjunction with aggregate boolean
3030 clauses such as THEREIS (= I 1);
3031 ** LOOP now signals an error when any variable is reused in the
3032 same loop (including the potentially useful construct analogous
3033 to WITH A = 1 WITH A = (1+ A);
3034 ** IT is only a special loop symbol within the first clause of a
3035 conditional loop clause;
3036 ** LOOP with a typed iteration variable over a hashtable now
3037 signals a type error iff it should.
3038 * fixed some other bugs revealed by Paul Dietz' test suite:
3039 ** FILE-STREAM now names the class previously known as FD-STREAM;
3040 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3041 argument) no longer signals an error;
3042 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3043 of the slot symbol, rather than using the current package
3044 ((:CONC-NAME "") continues to intern the slot's name in the
3046 * incremented fasl file version number, because of the incompatible
3047 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3048 of the new implementation of DEFINE-COMPILER-MACRO.
3050 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3051 * Support for building SBCL for MIPS platforms running in
3052 little-endian mode has now been checked in, and basic
3053 functionality on said platforms verified.
3054 * minor incompatible change: PCL now records the pathname of a file
3055 in which methods and the like are defined, rather than its
3057 * minor incompatible change: TRUENAME now considers the truename of
3058 a file naming a directory to be the pathname with :DIRECTORY
3059 component indicating that directory.
3060 * minor incompatible change: a NAMED clause in the extended form of
3061 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3062 reason for the previous behaviour is unclear.
3063 * more systematization and improvement of CLOS and MOP conformance
3064 in PCL (thanks to Gerd Moellman and Pierre Mai):
3065 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3067 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3068 primary methods with no specializers;
3069 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3071 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3072 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3073 FORWARD-REFERENCED-CLASSes; error reporting on
3074 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3076 ** SXHASH on CLOS instances now uses a slot internal to the
3077 instance to return different numbers on distinct instances,
3078 while preserving the same return value through invocations of
3080 ** DEFMETHOD signals errors when methods with longer incongruent
3081 lambda lists are added to generic functions;
3082 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3083 CLASS, as specified in AMOP;
3084 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3085 on the order returned by the primary method for classes of
3086 class STANDARD-CLASS;
3087 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3088 * fixed some bugs shown by Paul Dietz' test suite:
3089 ** DOLIST puts its body in TAGBODY;
3090 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3092 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3093 value producing form;
3094 ** if more variables are given to PROGV than values, extra
3095 variables are bound and made to have no value;
3096 ** NSUBSTITUTE on list arguments gets the right answer with
3098 ** ELT signals an error of type TYPE-ERROR when the index argument
3099 is not a valid sequence index;
3100 ** LOOP signals (at macroexpansion time) an error of type
3101 PROGRAM-ERROR when duplicate variable names are found;
3102 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3103 ** FUNCALL of special-operators now cause an error of type
3105 ** PSETQ now works as required in the presence of side-effecting
3106 symbol-macro places;
3107 ** NCONC accepts any object as its last argument;
3108 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3110 ** loop-for-as-package does not require a package to be explicitely
3112 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3114 * fixed bug 166: compiler preserves "there is a way to go"
3115 invariant when deleting code.
3116 * fixed bug 172: macro lambda lists with required arguments after
3117 &REST arguments now cause an error to be signalled. (thanks to
3119 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3120 bug. (thanks to Matthew Danish)
3121 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3123 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3124 arguments are lexically rebound. (thanks to Gerd Moellmann and
3126 * fixed bug 194: error messages are now more informative when there
3127 is no primary method applicable in a call to a generic
3128 function. (thanks to Gerd Moellmann)
3129 * fixed bug in command line argument checking (thanks to Julian
3131 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3132 (thanks to Matthew Danish)
3133 * incremented fasl file version number, because of the
3134 SXHASH-related changes in the layout of CLOS data structures
3136 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3137 * minor incompatible change: The runtime (the Unix executable named
3138 "sbcl") is now much pickier about the .core files it will load.
3139 Essentially it now requires .core files to descend from the same
3140 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3141 as the runtime does. (The intent is to prevent the crashes which
3142 can occur, and which can even be reported as mysterious failures,
3143 when people patch the sources or change the build parameters
3144 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3145 sbcl and .core files.)
3146 * fixed bug: VALUES-LIST is no longer optimized away.
3147 * fixed bug 142: The FFI conversion of C string values to Lisp
3148 string values no longer conses excessively. (thanks to Nathan
3149 Froyd porting Raymond Toy's fix to CMU CL)
3150 * began to systematize and improve MOP conformance in PCL (thanks to
3151 Nathan Froyd, Gerd Moellman and Pierre Mai):
3152 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3154 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3155 ** FINALIZE-INHERITANCE is now called on class finalization;
3156 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3157 argument precedence order.
3158 * fixed bug 202: The compiler no longer fails on functions whose
3159 derived types contradict their declared type.
3160 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3161 so it can be non-toplevel.
3162 * The fasl file version number has changed (because of the new
3163 implementation of DEFMACRO).
3164 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3165 safe code, that any length requirement by their type-specifier
3166 argument is valid. The exceptions to this are described in bug
3168 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3169 functions now check the types of their inputs as required by ANSI.
3170 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3171 introduced symbol is DECLAREd to be SPECIAL.
3172 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3173 * fixed bug: :COUNT argument to sequence functions may be negative
3174 * fixed bug: body of DO-SYMBOLS may contain declarations
3175 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3176 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3177 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3178 symbol macro only once
3179 * fixed printing of call frame when argument list is unavailable
3180 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3181 * compiler no longer signals WARNING on unknown keyword
3184 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3185 * A beta-quality port to the mips architecture running Linux,
3186 based on the old CMUCL backend, has been made. It has been tested
3187 on a big-endian kernel, and works sufficiently well to be able to
3188 rebuild itself; it has not been tested in little-endian mode.
3189 * fixed an inconsistency between gencgc.c and purify.c which made
3190 dumping/loading .core files unreliable
3191 * fixed bug 120a: The compiler now deals correctly with IFs where
3192 the consequent is the same as the alternative, instead of
3193 misderiving the return type. (thanks to Alexey Dejneka)
3194 * fixed bug 113: Logical pathnames are now dumpable (the logical
3195 host is resolved at load-time, throwing an error if it is not
3197 * fixed bug 174: FORMAT's error message is slightly clearer when a
3198 non-printing character is used in a format directive.
3199 * fixed several bugs in compiler checking of type declarations, i.e.
3200 violations of the Python "declarations are assertions" principle
3201 (thanks to Alexey Dejneka)
3202 * fixed several bugs in PCL's error checking (thanks to Gerd
3204 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3206 * fixed bug in compilation of functions as first class values
3207 (thanks to Antonio Martinez-Shotton)
3208 * The compiler's handling TYPE-ERRORs which it can prove will
3209 inevitably happen at runtime has been cleaned up and corrected
3210 in several ways. (thanks to Alexey Dejneka)
3211 * improved argument type checking for various basic arithmetic
3212 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3213 lavished on them in the past that they can be compiled in many
3214 ways in different special cases
3215 * fixed bug 181: compiler checks validity of user supplied type
3217 * cleaned up code flushing in optimization: Function calls which
3218 should signal errors for safety purposes (e.g. which ANSI says
3219 should signal errors when their arguments are of incorrect type)
3220 are no longer optimized away.
3221 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3222 * incremented fasl file version number, because changes in the
3223 implementation of sequence functions like COERCE caused
3224 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3225 in old inline expansions) to become undefined. (Actually these
3226 changes were later undone, so we might very well be binary
3227 compatible with 0.7.7 after all, but leaving the version number
3228 incremented seemed like the simplest and most conservative
3231 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3232 * An alpha-quality port to the parisc architecture running Linux,
3233 based on the old CMUCL backend, has been made. This, even more so
3234 than the other backends, should be considered still a work in
3235 progress; known problems include that the Linux kernel in 64-bit
3236 mode does not propagate the correct sigcontext structure to
3237 userspace, and consequently SBCL on a parisc64 kernel will not
3239 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3240 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3241 suppressed.) Also now that the compiler is looking at declarations
3242 in the environment, it checks optimization declarations as well,
3243 and suppresses inlining when (> DEBUG SPEED).
3244 * More fixes have been made to treatment of floating point exception
3245 treatment and other Unix signals. In particular, floating point
3246 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3247 * The detection and handling of control stack exhaustion (infinite
3248 or very deeply nested recursion) has changed. Stack exhaustion
3249 detection is now done by write-protecting pages at the OS level
3250 and applies at all optimization settings; when found, a
3251 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3252 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3253 be caught using IGNORE-ERRORS.
3254 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3255 correctly when transforming recursive LABELS functions to LETs.
3256 (thanks to Alexey Dejneka)
3257 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3258 that are names of constants or global variables.
3259 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3260 alien routines with docstrings.
3261 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3262 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3264 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3265 (thanks to Eric Marsden)
3266 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3267 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3268 * Bug fix: The SPARC backend can now compile functions involving
3269 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3270 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3271 to a functions expecting a PATHNAME-DESIGNATOR.
3272 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3273 lambda lists. (thanks to Alexey Dejneka)
3274 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3275 functions so that EXPORTing the name of the class doesn't cause
3276 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3277 to Antonio Martinez for reporting this)
3278 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3279 constant-folded without failing an assertion. (thanks to Einar
3280 Floystad Dorum for reporting this)
3281 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3282 types are now dealt with more correctly by the compiler.
3283 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3284 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3285 behaviour analogous to RENAME-FILE. This puts its behaviour more
3286 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3288 * The fasl file version number has changed again. (because of the
3289 bug fix involving the names of PCL MAKE-INSTANCE functions)
3291 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3292 * bug fix: Floating point exceptions are treated much more
3293 consistently on the x86/Linux and PPC/Linux platforms.
3294 * Array initialization with :INITIAL-ELEMENT is now much faster for
3295 cases when the compiler cannot open code the array creation, but
3296 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3297 array accesses have also seen a speed increase.
3298 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3299 specified. (This was at the root of some bad interactions between
3300 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3302 * bug fix: Internal error arguments for undefined functions are now
3303 computed correctly on the PPC/Linux platform.
3304 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3305 Raymond Toy's patch for CMU CL)
3306 * Support for the Solaris 9 operating environment has been included
3307 (thanks to Daniel Merritt)
3308 * A very ugly but hopefully complete draft of the missing FFI chapter
3309 of the manual has been created by reformatting the corresponding
3310 CMU CL manual chapter into (currently very ugly and incoherent)
3311 DocBook and bringing it up to date for SBCL behavior. Thus, the
3312 manual is now essentially complete, at least by my extreme
3313 once-and-only-once standards, whereby it's acceptable to refer to
3314 the doc strings of SB-EXT functions as the primary documentation.
3315 * The fasl file version number has changed again, due to cleanup of
3316 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3317 is no longer a static symbol.)
3319 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3320 * SBCL now builds with OpenMCL (version 0.12) as the
3321 cross-compilation host; also, more progress has been made toward
3322 bootstrapping under CLISP.
3323 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3325 * bug 158 fixed: The compiler can now deal with integer loop
3326 increments different from 1; fixing this turned out also to fix
3328 * bug 169 fixed: no more bogus warnings about using lexical bindings
3329 despite the presence of perfectly good SPECIAL declarations (thanks
3330 to David Lichteblau)
3331 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3332 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3333 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3335 * bug 180 fixed: Method combination specifications no longer ignore
3336 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3337 * bug fix: Structure type predicate functions now check their argument
3338 count as they should.
3339 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3340 correctly. (thanks to Pierre Mai)
3341 * minor incompatible change: The --noprogrammer option is deprecated
3342 in favor of the new --disable-debugger option, which is very similar.
3343 (The major difference is that it takes effect at a slightly different
3344 time at startup, causing handling of errors in --sysinit and
3345 --userinit files will be affected differently.) The
3346 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3347 been added to allow this functionality to be controlled from ordinary
3348 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3349 maintainers, who might want to run non-interactive scripts to
3350 build SBCL cores which will later be used interactively.)
3351 * minor incompatible change: The LOAD function no longer, when given
3352 a wild pathname to load, loads all files matching that pathname.
3353 Instead, an error of type FILE-ERROR is signalled.
3355 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3356 * bug 147 fixed: The compiler preserves its block link/count
3357 invariants more correctly now so that it doesn't crash. (thanks
3359 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3361 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3362 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3363 no-value-for-this-character as NIL (as specified by ANSI).
3364 * HOST-NAMESTRING on physical pathnames now returns a string that is
3365 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3366 (thanks to Christophe Rhodes)
3367 * The Alpha port handles icache flushing more correctly. (thanks to
3369 * More progress has been made toward bootstrapping under CLISP. (thanks
3370 to Christophe Rhodes)
3371 * The fasl file format has changed again, because dynamic loading
3372 on OpenBSD (which has non-ELF object files) motivated some cleanups
3373 in the way that foreign symbols are transformed and passed around.
3374 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3375 is no longer treated as whitespace by the reader, but instead as
3376 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3377 |A
\7fB|, instead of A as it used to.
3379 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3380 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3381 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3382 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3383 this as of 0.7.1.45, but was left out of the previous news section
3384 (thanks to Dan Barlow)
3385 * SBCL now runs on the Solaris operating system on SPARC architectures
3386 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3387 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3388 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3389 * SPARC backend cleanups, allowing builds of cores optimized for V8
3390 and V9 SPARCS, and also emission of code targeted to a particular
3391 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3393 * SBCL is closer to bootstrapping under CLISP, thanks to various
3394 fixes by Christophe Rhodes.
3395 * The fasl file format has changed again, to allow the compiler's
3396 INFO database to support symbol macros.
3397 * The user manual (in doc/) is formatted into HTML more nicely.
3398 (thanks to coreythomas)
3399 * The system is smarter about SUBTYPEP relationships, especially
3400 those involving NOT types (including types such as ATOM which are
3401 represented internally using NOT types). Thus SUBTYPEP is less
3402 likely to return (VALUES NIL NIL) in general, and in particular
3403 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3405 * The fasl file format has changed again, because the internal
3406 representation of types now includes a new slot to support the new
3407 SUBTYPEP-of-NOT-types logic.
3408 * (not a change in the main branch of SBCL, but a related prototype
3409 which can hopefully be merged into the main branch of SBCL in the
3410 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3411 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3412 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3413 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3414 dispatch macro character combinations. (thanks to Alexey Dejneka)
3415 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3416 that superficially look like logical namestrings correctly.
3417 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3419 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3420 default supplied-p) for &optional and &key arguments. (thanks to
3423 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3424 * incompatible change: The compiler is now less aggressive about
3425 tail call optimization, doing it only when (> SPACE DEBUG) or
3426 (> SPEED DEBUG). (This is an incompatible change because there are
3427 programs which relied on the old CMU-CL-style behavior to optimize
3428 away their unbounded recursion which will now die of stack overflow.)
3429 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3430 for non-GENCGC systems has been increased to 20M (since that
3431 seems much closer to the likely performance optimum for modern
3432 systems than the old 4M value was)
3433 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3434 and *DEBUG-PRINT-LEVEL*
3435 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3436 of CMU CL's support for SPARC, and various endianness and other
3437 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3438 * new syntactic sugar for the Unix command line: --load foo.bar is now
3439 an alternate notation for --eval '(load "foo.bar")'.
3441 ** The system now detects stack overflow and handles it gracefully,
3442 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3443 optimization settings. (This is a good thing in general, and
3444 its introduction in this version should be particularly timely
3445 for anyone whose code fails because of suppression of tail
3447 ** The system now hunts for the C variable "environ" in a more
3448 devious way, to avoid segfaults when the C library version differs
3449 between compile time and run time. (thanks to Christophe Rhodes)
3450 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3451 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3452 ** The compiler no longer issues bogus style warnings for undefined
3453 classes in the same source file as the DEFCLASSes which defined
3454 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3456 ** fixes in CONDITION class precedence list for undefined function
3457 errors (thanks to Alexei Dejneka)
3458 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3459 correctly. (thanks to Dan Barlow)
3460 ** portability fixes aiming at bootstrapping under CLISP (thanks
3461 to Dave McDonald and Christophe Rhodes)
3462 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3463 ** fixes in type translation and and type inference (thanks to
3465 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3466 ** various fixes in the new ports (thanks to Dan Barlow)
3467 * several changes related to debugging:
3468 ** suppression of tail recursion, as noted above
3469 ** stack overflow detection, as noted above
3470 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3471 is now the default. (For some time encapsulation has been more
3472 reliable than the breakpoint-based :ENCAPSULATE NIL
3473 implementation, at least on X86 systems; and I just noticed that
3474 encapsulation also seems closer to the spirit of the ANSI
3477 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3479 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3480 up properly again. (There was a packaging bug in 0.7.0 which
3481 left their definitions in SB-SYS::LOAD-FOREIGN and
3482 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3483 vital for most things which interface to C-level interfaces,
3484 like extensions working with sockets or databases or
3485 Perl-compatible regexes or whatever, and the need to fix
3486 this bug is the main reason that 0.7.1 was released so
3488 ** DEFGENERIC is now choosier about the methods it redefines, so that
3489 reLOADing a previously-LOADed file containing DEFGENERICs does
3490 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3491 cycle now works as it should. (thanks to Alexey Dejneka)
3492 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3493 fixed by Christophe Rhodes. (He actually submitted this patch
3494 months ago, and I delayed until after 0.7.0.)
3495 ** Bug 111 (internal compiler confusion about runtime checks on
3496 FUNCTION types) was fixed by Alexey Dejneka.
3497 * Some internal cleanups (getting rid of variables which aren't
3498 needed now that the byte interpreter is gone) caused the fasl
3499 file format number to change again.
3501 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3502 * major incompatible change: The default fasl file extension, i.e. the
3503 default extension for files produced by COMPILE-FILE, has changed
3504 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3506 ** There are many changes in the implementation of the compiler.
3507 SBCL is now essentially a compiler-only implementation of ANSI
3508 Common Lisp. EVAL still "interprets" a few special cases, but
3509 almost all the interesting cases are handled by creating
3510 a LAMBDA expression, calling COMPILE on it, then calling
3511 FUNCALL on the result.
3512 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3513 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3514 Since the code is newer, there might still be some new bugs
3515 (though not as many as before Martin Atzmueller's fixes:-). But
3516 the new code is substantially simpler and clearer, and hopefully
3517 any remaining bugs will be simpler, less fundamental, and more
3518 fixable then the bugs in the old code.
3519 ** The revised compiler is still a little unsteady on its feet.
3521 *** The debugging information it produces (particularly the names
3522 of FUNCTION objects) is sometimes much less useful than what
3523 the old compiler produced.
3524 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3525 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3526 within a MACROLET) has been temporarily weakened.
3527 ** There are new compiler optimizations for various functions:
3528 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3529 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3530 *** the math functions TRUNCATE, FLOOR, and CEILING
3531 *** the function-of-all-trades COERCE
3532 Mostly these should be transparent, but there's one
3533 potentially-annoying problem (bug 117): when the compiler
3534 inline-expands a function and does type analysis on the result,
3535 it can create control paths which have type mismatches, and
3536 when it can't prove that those control paths aren't taken,
3537 it will issue WARNINGs about the type mismatches. This is
3538 a particular problem in practice for the new sequence functions.
3539 It's not clear how this should be fixed, and for now, a
3540 workaround is given in the entry for 117 in the BUGS file.
3541 ** (Because of the interaction between the two previous items --
3542 occasional inlining problems and new inline expansions -- some
3543 of the new sequence function optimizations won't really kick in
3544 completely until debugging information, and then inlining, are
3545 straightened out in some future version.)
3546 * minor incompatible changes:
3547 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3548 DIRECTORY no longer implicitly promotes NIL slots of its
3549 pathname argument to :WILD. In particular, when you ask for the
3550 contents of a directory (which you used to be able to do without
3551 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3552 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3553 ** changes in behavior that ANSI explicitly defines to be
3554 implementation dependent:
3555 *** The new compiler-only implementation still conforms with ANSI,
3556 but acts a little different than before. Besides the obvious
3557 changes in performance tradeoffs (that the cost per form passed
3558 to EVAL has gone up, and the cost per form executed by EVAL
3559 has gone down), the behavior of the system changes a little
3560 because there are no longer any interpreted function objects.
3561 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3562 e.g. doing COMPILE on the output of interactive DEFUN is
3564 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3566 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3567 has changed from (:CL) to NIL.
3568 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3569 e.g. #\Newline and #\Tab, don't have names specified in the
3570 ANSI Common Lisp standard, is now based on their ASCII symbolic
3571 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3572 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3573 are no longer used for output.
3574 ** changes in internal implementation constants:
3575 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3576 4 million. (If your application spends a lot of time GCing and
3577 you have a lot of RAM, you might want to experiment with
3578 increasing it even more.)
3579 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3580 However, almost all old code should still continue to work without
3581 immediate update, as SB-C-CALL is now a (deprecated) nickname
3583 ** Old operator names in the style DEF-FOO are now deprecated in
3584 favor of new corresponding names DEFINE-FOO, for consistency with
3585 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3586 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3587 internal symbols, but a few supported extensions like
3588 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3589 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3590 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3591 etc. as you get deeper into recursive calls to the debugger
3592 command loop, instead of the old "5]", "5]]", "5]]]"
3593 sequence. (I was motivated to do this when squabbles between
3594 ILISP and SBCL left me very deeply nested in the debugger. In the
3595 short term, this change will probably provoke more ILISP/SBCL
3596 squabbles, but hopefully it will be an improvement in the long run.)
3597 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3598 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3599 function, since declaiming return types involving aliens is
3600 (1) annoyingly messy to do by hand and (2) vital to efficient
3601 compilation of code which calls such functions.
3602 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3603 longer reexported by the SB-EXT package. They're solely useful
3604 for alien code, so it seems more logical that you should get
3605 them from the SB-ALIEN package, not in SB-EXT.
3606 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3607 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3608 features. Instead, the code that they used to control is always
3609 built into the system.
3610 * many other bug fixes
3611 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3612 advantage of the new EVAL-WHEN stuff and to clean them up in
3613 general, and they are now more ANSI-compliant in a number of
3614 ways. Martin Atzmueller is responsible for a lot of this.
3615 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3617 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3618 *** correct ERROR type for various file operations
3619 *** some fixes for Lisp streams
3620 *** DEFMETHOD syntax checking
3621 *** changing old weird representation of debug information as
3622 strings (which, among their other deficiencies, don't transform
3623 correctly when you rename packages, and don't change their
3624 print representation when you change things like *PACKAGE*
3625 and *PRINT-LENGTH*) to symbols and lists of symbols
3626 He also made several improvements and fixed several bugs in DESCRIBE.
3627 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3629 *** misbehavior of WRITE-STRING/WRITE-LINE
3630 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3631 and several other LOOP problems as well
3632 *** DIRECTORY when similar filenames are present
3633 *** DEFGENERIC with :METHOD options
3634 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3635 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3636 *** argument ordering in FIND with :TEST option
3637 *** mishandled package designator argument in APROPOS-LIST
3638 *** various problems in the backquote readmacro
3639 *** a bug in APROPOS
3640 *** probably some others that I'm not describing very well here,
3641 since the CVS log documents them by reference to sbcl-devel
3642 messages, and the SourceForge archives aren't working well.:-(
3643 ** Dan Barlow improved the Alpha port (and is making progress on the
3644 PPC port, for those of you who think different).
3645 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3646 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3647 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3648 continues to work on the SPARC port (for those of you in a position
3649 to look down upon our little PC-compatible boxes from a great height).
3650 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3651 and uses that copy, rather than the current dynamic value, when
3652 it's trying to decide whether to truncate output. Thus e.g.
3653 (let ((*print-lines* 50))
3654 (pprint-logical-block (stream nil)
3656 (let ((*print-lines* 8))
3657 (print (aref possiblybigthings i) stream)))))
3658 should now truncate the logical block only at 50 lines, instead of
3659 often truncating it at 8 lines, as it did before.
3660 * The doc/cmucl/ directory, containing old CMU CL documentation
3661 from the time of the fork, is no longer part of the base system.
3662 SourceForge has shut down its anonymous FTP service, and with it
3663 my original plan for distributing the old CMU CL documentation
3664 there. For now, if you need these files you can download an old
3665 SBCL source release and extract them from it.
3666 * The fasl file version number changed again, for dozens of reasons,
3667 some of which are apparent above.
3669 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3670 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3671 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3672 patch, so that DISASSEMBLE output is much nicer.
3673 * The code in the SB-PROFILE package now seems reasonably stable.
3674 I still haven't decided what the final interface should look like
3675 (I'd like PROFILE to interact cleanly with TRACE, since both
3676 facilities use function encapsulation) but if you have a need
3677 for profiling now, you can probably use it successfully with
3678 the current CMU-CL-style interface.
3679 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3680 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3681 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3682 link now returns the dangling link itself, and for similar
3683 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3684 link itself. (In these cases the old code signalled an error and
3685 looped endlessly, respectively.) Thus, DIRECTORY now works even
3686 in the presence of dangling and cyclic symbolic links.
3687 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3688 is now a supported extension again, since the consensus on
3689 sbcl-devel was that it can be useful for ordinary development
3690 work, not just for debugging SBCL itself.
3691 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3692 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3693 that currently-defined functions might be redefined later with
3694 different return types.
3695 * Hash tables can be printed readably, as inspired by CMU CL code
3696 of Eric Marsden and SBCL code of Martin Atzmueller.
3697 * better error handling in CLOS method combination, thanks to
3698 Martin Atzmueller porting Pierre Mai's CMU CL patches
3699 * more overflow fixes for >16Mbyte I/O buffers
3700 * A bug in READ has been fixed, so that now a single Ctrl-D
3701 character suffices to cause end-of-file on character streams.
3702 In particular, now you only need one Ctrl-D at the command
3703 line (not two) to exit SBCL.
3704 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3706 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3707 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3708 does the right thing.
3709 * The implementation of some type tests, especially for CONDITION
3710 types, is now tidier and maybe faster, due to CMU CL code
3711 originally by Douglas Crosher, ported by Martin Atzmueller.
3712 * Some math functions have been fixed, and there are new
3713 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3714 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3715 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3716 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3717 likely to be built into sbcl-0.7.0.)
3718 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3719 CL, ported to SBCL by Martin Atzmueller)
3720 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3721 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3722 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3723 * There's a new slam.sh hack to shorten the edit/compile/debug
3724 cycle for low-level changes to SBCL itself, and a new
3725 :SB-AFTER-XC-CORE target feature to control the generation of
3726 the after-xc.core file needed by slam.sh.
3727 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3728 COMPILE-FILE is no longer supported, so that now every function
3729 gets an entry point, so that block compilation looks a little
3730 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3731 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3732 returns the number of bytes consed since the system started,
3733 rather than the number consed since the first time the function
3734 was called. (The new definition parallels ANSI functions like
3735 CL:GET-INTERNAL-RUN-TIME.)
3736 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3737 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3738 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3739 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3740 * Fasl file version numbers are now independent of the target CPU,
3741 since historically most system changes which required version
3742 number changes have affected all CPUs equally. Similarly,
3743 the byte fasl file version is now equal to the ordinary
3746 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3747 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3748 is no longer recognized. I apologize for this, because it was
3749 listed in SB-EXT as a supported extension, but I found that
3750 its existing behavior was poorly specified, as well as incorrectly
3751 specified, and it looked like too much of a mess to straighten it
3752 out. I have enough on my hands trying to get ANSI stuff to work..
3753 * many patches ported from CMU CL by Martin Atzmueller, with
3754 half a dozen bug fixes in pretty-printing and the debugger, and
3755 half a dozen others elsewhere
3756 * fixed bug 13: Floating point infinities are now supported again.
3757 They might still be a little bit flaky, but thanks to bug reports
3758 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3759 as flaky as they were.
3760 * The --noprogrammer command line option is now supported. (Its
3761 behavior is slightly different in detail from what the old man
3762 page claimed it would do, but it's still appropriate under the
3763 same circumstances that the man page talks about.)
3764 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3765 are now supported, and enabled by default. Thus, the compiler can
3766 handle many floating point and complex operations much less
3767 inefficiently. (Thus e.g. you can implement a complex FFT
3769 * The compiler now detects type mismatches between DECLAIM FTYPE
3770 and DEFUN better, and implements CHECK-TYPE more correctly, and
3771 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3772 thanks to patches from Martin Atzmueller.
3773 * various fixes to make the cross-compiler more portable to
3774 ANSI-conforming-but-different cross-compilation hosts (notably
3775 Lispworks for Windows, following bug reports from Arthur Lemmens)
3776 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3777 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3778 fixed thanks to Pierre Mai's CMU CL patches.
3779 * fixes to keep the system from overflowing internal counters when
3780 it tries to use i/o buffers larger than 16M bytes
3781 * fixed bug 45a: Various internal functions required to support
3782 complex special functions have been merged from CMU CL sources.
3783 (When I was first setting up SBCL, I misunderstood a compile-time
3784 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3785 * improved support for type intersection and union, fixing bug 12
3786 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3787 more obscure bugs as well
3788 * some steps toward byte-compiling non-performance-critical
3789 parts of the system, courtesy of patches from Martin Atzmueller
3790 * Christophe Rhodes has made some debian packages of sbcl at
3791 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3792 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3793 stable, but are nonetheless usable. When he's ready, I'd be happy
3794 to add them to the SourceForge "File Releases" section. (And if
3795 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3796 * new fasl file format version number (because of changes in
3797 internal representation of (OR ..) types to accommodate the new
3798 support for (AND ..) types, among other things)
3800 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3801 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3803 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3805 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3806 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3807 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3808 * DESCRIBE now works on structure objects again.
3809 * Most function call argument type mismatches are now handled as
3810 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3811 know whether the function will be redefined before the call is
3812 executed. (The compiler could flag local calls with full WARNINGs,
3813 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3814 it doesn't keep track of enough information to know whether calls
3815 are local in this sense.)
3816 * Compiler output is now more verbose, with messages truncated
3817 later than before. (There should be some supported way for users
3818 to override the default verbosity, but I haven't decided how to
3819 provide it yet, so this behavior is still controlled by the internal
3820 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3821 src/compiler/ir1util.lisp.)
3822 * Fasl file format version numbers have increased again, because
3823 support for the Gray streams extension changes the layout of the
3824 system's STREAM objects.
3825 * The Gray subclassable streams extension now works, thanks to a
3826 patch from Martin Atzmueller.
3827 * The full LOAD-FOREIGN extension (not just the primitive
3828 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3829 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3830 but like most other programs, it defaults to copying the Unix
3831 environment from the original process instead of starting the
3832 new process in an empty environment.
3833 * Extensions which manipulate the Unix environment now support
3834 an :ENVIRONMENT keyword option which doesn't smash case or
3835 do other bad things. The CMU-CL-style :ENV option is retained
3836 for porting convenience.
3837 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3838 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3840 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3842 * A patch from Martin Atzmueller seems to have solved the SIGINT
3843 problem, and as far as we know, signal-handling now works cleanly.
3844 (If you find any new bugs, please report them!)
3845 * The system no longer defaults Lisp source file names to types
3846 ".l", ".cl", or ".lsp", but only to ".lisp".
3847 * The compiler no longer uses special default file extensions for
3848 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3849 seems to expect a single default extension for all compiled code,
3850 and there's no compelling reason to try to stretch the standard
3851 to allow two different extensions.) Instead, byte-compiled files
3852 default to the same extension as native-compiled files.
3853 * Fasl file format version numbers have increased again, because
3854 a rearrangement of internal implementation packages made some
3855 dumped symbols in old fasl files unreadable in new cores.
3856 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3857 many fewer weird special cases.
3858 * Bug #17 (differing COMPILE-FILE behavior between logical and
3859 physical pathnames) has been fixed, and some related misbehavior too,
3860 thanks to a patch from Martin Atzmueller.
3861 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3862 by Tim Moore, ported to SBCL by Martin Atzmueller.
3863 * Martin Atzmueller fixed several filesystem-related problems,
3864 including bug #36, in part by porting CMU CL patches, which were
3865 written in part by Paul Werkowski.
3866 * More compiler warnings in src/runtime/ are gone, thanks to
3867 more patches from Martin Atzmueller.
3868 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3871 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3873 * DESCRIBE now works on CONDITION objects.
3874 * The debugger now handles errors which arise when trying to print
3875 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3877 * The build system now uses an additional file, customize-target-features.lisp,
3878 to allow local modifications to the target *FEATURES* list. (The point of
3879 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3880 debugging features enabled, without having to worry about propagating it
3881 into everyone's system when I do a "cvs update".) When no
3882 customize-target-features.lisp file exists, the target *FEATURES* list
3883 should be constructed the same way as before.
3884 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3885 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3886 warn about behavior which is undefined under the ANSI spec. Note: This
3887 is specified by ANSI, but it's not very popular with programmers.
3888 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3889 macro in the SBCL sources for an example of a workaround which you
3890 might use to make portable ANSI-standard code which does what you want.
3891 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3892 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3893 when it's used as a non-toplevel form. (This is required in order
3894 to implement the DEFCONSTANT-EQX macro.)
3895 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3896 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3897 funny things, due to bugs in the implementation of EVAL-WHEN
3898 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3899 non-toplevel DEFCONSTANTs are uncommon.)
3900 * The core file version number and fasl file version number have been
3901 incremented, because the old noncompliant DEFCONSTANT behavior involved
3902 calling functions which no longer exist, and because I also took the
3903 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3904 * fixed bug 1 (error handling before read-eval-print loop starts), and
3905 redid debugger restarts and related debugger commands somewhat while
3907 ** The QUIT debugger command is gone, since it did something
3908 rather different than the SB-EXT:QUIT command, and since it never
3909 worked properly outside the main toplevel read/eval/print loop.
3910 Invoking the new TOPLEVEL restart provides the same functionality.
3911 ** The GO debugger command is also gone, since you can just invoke
3912 the CONTINUE restart directly instead.
3913 ** The TOP debugger command is also gone, since it's redundant with the
3914 FRAME 0 command, and since it interfered with abbreviations for the
3916 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3918 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3919 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3920 some porting work by Martin Atzmueller)
3921 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3922 with a boatload of patches which he ported to SBCL. Now that those
3924 ** The system tries to make sure that its low-priority messages
3925 are prefixed by semicolons, to help people who like to use
3926 syntax highlighting in their ILISP buffer. (This patch
3927 was originally due to Raymond Toy.)
3928 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3929 patches originally written by Raymond Toy.
3930 ** The compiler understands coercion between single-value and
3931 multiple-VALUES type expressions better, getting rid of some very
3932 weird behavior, thanks to patches originally by Robert MacLachlan
3933 and Douglas Crosher.
3934 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3935 lambda lists, thanks to a patch originally by Pierre Mai.
3936 ** The system no longer bogusly warns about "abbreviated type
3938 ** The compiler gets less confused by inlining and RETURN-FROM,
3939 thanks to some patches originally by Tim Moore.
3940 ** The system no longer hangs when dumping circular lists to fasl
3941 files, thanks to a patch originally from Douglas Crosher.
3942 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3943 undefined function error.
3944 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3945 in structure slot initforms) for now, documented workaround instead:-|
3946 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3947 * fixed bug 5: added stubs for various Gray stream functions called
3948 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3949 installed, at least appropriate type errors are generated
3950 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3951 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3952 consistently in DEFMETHOD forms.
3953 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3954 it doesn't seem to affect SBCL after all
3955 * The C runtime system now builds with better optimization and many
3956 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3958 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3960 * The system is now under CVS at SourceForge (instead of the
3961 CVS repository on my home machine).
3962 * The new signal handling code has been tweaked to treat register
3963 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3964 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3965 downstream, e.g. in debug-int.lisp, has implicit dependencies
3966 on the unsignedness of integer representation of machine words,
3967 and that caused the system to bomb out with infinite regress
3968 when trying to recover from type errors involving signed values,
3969 e.g. (BUTLAST '(1 2 3) -1).)
3970 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3971 (This was one of the bugs Peter Van Eynde reported back in July.)
3972 * The system now uses code inspired by Colin Walters' O(N)
3973 implementation of MAP (from the cmucl-imp@cons.org mailing
3974 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3975 inline the MAP operation, and there is more than one
3976 sequence argument to the MAP call (so that it can't just
3977 do ETYPECASE once and for all based on the type of the
3978 single sequence argument). (The old non-inline implementation
3979 of the general M-argument sequence-of-length-N case required
3980 O(M*N*N) time when any of the sequence arguments were LISTs.)
3981 * The QUIT :UNIX-CODE keyword argument has been renamed to
3982 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3983 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3985 * Raymond Wiker's patches to port dynamic loading from Linux to
3986 FreeBSD have been added.
3987 * The BUGS file is now more nearly up to date, thanks in large part
3988 to Martin Atzmueller's review of it.
3989 * The debugger now flushes standard output streams before it begins
3990 its output ("debugger invoked" and so forth).
3991 * The core version number and fasl file version number have both
3992 been incremented, because of incompatible changes in the layout
3994 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3995 used to not be called for a saved Lisp image.)
3996 * Martin Atzmueller's version of a patch to fix a compiler crash,
3997 as posted on sbcl-devel 13 September 2000, has been installed.
3998 * Instead of installing Martin Atzmueller's patch for the
3999 compiler transform for SUBSEQ, I deleted the compiler transform,
4000 and transforms for some similar consing operations.
4001 * A bug in signal handling which kept TRACE from working on OpenBSD
4003 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4004 away by constant folding
4005 * The system now defines its address space constants in one place
4006 (in the Lisp sources), and propagates them automatically elsewhere
4007 (through GENESIS and the sbcl.h file). Therefore, patching the
4008 address map is less unnecessarily tedious and error-prone. The
4009 Lisp names of address space constants have also been systematized.
4010 * CVS tags like dollar-Header-dollar have been removed from
4011 the sources, because they have never saved me trouble and
4012 they've been source of trouble working with patches and other
4013 diff-related operations.
4014 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4015 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4017 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4019 * The system has been ported to OpenBSD.
4020 * The system now compiles with a simple "sh make.sh" on the systems
4021 that it's supported on. I.e., now you no longer need to tweak
4022 text in the target-features.lisp-expr and symlinks in src/runtime/
4023 by hand, the make.sh takes care of it for you.
4024 * The system is no longer so grossly inefficient when compiling code
4025 involving vectors implemented as general (not simple) vectors (VECTOR T),
4026 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4027 which dares to use the various sequence functions on non-simple
4028 vectors, takes less of a performance hit.
4029 * There is now a primitive type predicate VECTOR-T-P
4030 to test for the (VECTOR T) type, so that e.g.
4031 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4032 can now be compiled with some semblance of efficiency. (The old code
4033 turned the type declaration into a full call to %TYPEP at runtime!)
4034 * AREF on (VECTOR T) is still not fast, since it's still compiled
4035 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4036 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4037 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4038 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4039 without making full calls to SUBTYPEP at runtime.
4040 (Some analogous efficiency issues for non-simple vectors specialized to
4041 element types other than T, or for non-simple multidimensional arrays,
4042 have not been addressed. They could almost certainly be handled the
4043 same way if anyone is motivated to do so.)
4044 * The changes in array handling break binary compatibility, so
4045 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4046 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4047 instead of (VALUES T T).
4048 * By following the instructions that Dan Barlow posted to sbcl-devel
4049 on 2 July 2000, I was able to enable primitive dynamic object
4050 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4051 functionality is not implemented (since it calls ld to resolve
4052 library references automatically, requiring RUN-PROGRAM for its
4053 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4054 try to resolve library references) is now supported.
4055 * The system now flushes the standard output streams when it terminates,
4056 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4057 them at several other probably-convenient times, e.g. in each pass of
4058 the toplevel read-eval-print loop, and after evaluating a form given
4059 as an "--eval" command-line option. (These changes were motivated by a
4060 discussion of stream flushing issues on cmucl-imp in August 2000.)
4061 * The source transform for TYPEP of array types no longer assumes
4062 that an array whose element type is a not-yet-defined type
4063 is implemented as an array of T, but instead punts, so that the
4064 type will be interpreted at runtime.
4065 * There is now some support for cross-compiling in make.sh: each of
4066 the phases of make.sh has its own script. (This should be transparent
4067 to people doing ordinary, non-cross-compile builds.)
4068 * Since my laptop doesn't have hundreds of megabytes of memory like
4069 my desktop machine, I became more motivated to do some items on
4070 my to-do list in order to reduce the size of the system a little:
4071 ** Arrange for various needed-only-at-cold-init things to be
4072 uninterned after cold init. To support this, those things have
4073 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4074 symbols with such names are now uninterned after cold init).
4075 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4076 fasl files for cold load.
4077 ** Remove the old compiler structure pooling code (which used to
4078 be conditional on the target feature :SB-ALLOC) completely.
4079 ** Redo the representation of some data in cold init to be more compact.
4080 (I also looked into supporting byte compiled code at bootstrap time,
4081 which would probably reduce the size of the system a lot, but that
4082 looked too complicated, so I punted for now.)
4083 * The maximum signal nesting depth in the src/runtime/ support code has
4084 been reduced from 4096 to 256. (I don't know any reason for the very
4085 large old value. If the new smaller value turns out to break something,
4086 I'll probably just bump it back up.)
4087 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4089 * Many, many bugs reported by Peter Van Eynde have been added to
4090 the BUGS list; some have even been fixed.
4091 * While enabling dynamic object file loading, I tried to make the
4092 code easier to understand, renaming various functions and variables
4093 with less ambiguous names, and changing some function calling
4094 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4095 * While trying to figure out how to do the OpenBSD port, I tried to
4096 clean up some of the code in src/runtime/. In particular, I dropped
4097 support for non-POSIX signal handling, added various comments,
4098 tweaked the code to reduce the number of compilation warnings, and
4099 renamed some files to increase consistency.
4100 * To support the new automatic configuration functionality in make.sh,
4101 the source file target-features.lisp-expr has been replaced with the
4102 source file base-target-features.lisp-expr and the machine-generated
4103 file local-target-features.lisp-expr.
4104 * fixed a stupid quoting error in make.sh so that using CMU CL
4105 "lisp -batch" as cross-compilation host works again
4107 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4109 * DESCRIBE no longer tries to call itself recursively to describe
4110 bound/fbound values, so that it no longer fails on symbols which are
4111 bound to themselves (like keywords, T, and NIL).
4112 * DESCRIBE now works on generic functions.
4113 * The printer now prints less-screwed-up representations of closures
4114 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4115 * A private symbol is used instead of the :EMPTY keyword previously
4116 used to mark empty slots in hash tables. Thus
4117 (DEFVAR *HT* (MAKE-HASH-TABLE))
4118 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4119 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4120 now does what ANSI says that it should. (You can still get
4121 similar noncompliant behavior if bang on the hash table
4122 implementation with all the symbols you get back from
4123 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4124 This breaks binary compatibility, since tests for equality to
4125 :EMPTY are wired into things like the macroexpansion of
4126 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4128 * There's now a minimal placeholder implementation for CL:STEP,
4129 as required by ANSI.
4130 * An obscure bug in the interaction of the normal compiler, the byte
4131 compiler, inlining, and structure predicates has been patched
4132 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4133 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4135 * Missing ordinary arguments in a macro call are now detected even
4136 when the macro lambda list contains &KEY or &REST.
4137 * The debugger no longer complains about encountering the top of the
4138 stack when you type "FRAME 0" to explicitly instruct it to go to
4139 the top of the stack. And it now prints the frame you request even
4140 if it's the current frame (instead of saying "You are here.").
4141 * As specified by ANSI, the system now always prints keywords
4142 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4144 * The default initial SIZE of HASH-TABLEs is now smaller.
4145 * Type information from CLOS class dispatch is now propagated
4146 into DEFMETHOD bodies, so that e.g.
4147 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4149 is now basically equivalent to
4150 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4151 (DECLARE (TYPE SINGLE-FLOAT X))
4153 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4154 operation, without having to do run-time type dispatch.
4155 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4156 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4157 * Since I don't seem to be making big file reorganizations very often
4158 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4159 is overflowing my ability to conveniently back them up), I've finally
4160 checked the system into CVS. (The CVS repository is on my home system,
4161 not at SourceForge -- putting it on SourceForge might come later.)
4162 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4163 high-level GC-NOTIFY-FOO functions send their output. (There's
4164 still very little control of where low-level verbose GC functions
4165 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4166 less than it used to -- the GC-NOTIFY-FOO functions are now under
4167 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4168 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4169 in only one place in the source code, and propagates it automatically
4170 everywhere that it's needed. Thus e.g. when I bump the version from
4171 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4172 * The C source files now include boilerplate legalese and documentation
4173 at the head of each file (just as the Lisp source files already did).
4174 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4175 to his page will be replaced with a link to his new CLiki service.
4177 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4179 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4180 * The build process now looks for GNU make under the default name "gmake",
4181 instead of "make" as it used to. If GNU make is not available as "gmake"
4182 on your system, you can change this default behavior by setting the
4183 GNUMAKE environment variable.
4184 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4185 can build without error under CMU CL.
4187 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4189 * There is now a partial SBCL user manual (with some new text and some
4190 text cribbed from the CMU CL manual).
4191 * The beginnings of a profiler have been added (starting with the
4192 CMU CL profiler and simplifying and cleaning up). Eventually the
4193 main interface should be through the TRACE macro, but for now,
4194 it's still accessed through vaguely CMU-CL-style functions and macros
4195 exported from the package SB-PROFILE.
4196 * Some problems left over from porting CMU CL to the new
4197 cross-compilation bootstrap process have been cleaned up:
4198 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4199 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4200 part when getting rid of the compiler *BACKEND* stuff.)
4201 ** The value of *NULL-TYPE* was screwed up, because it was
4202 being initialized before the type system knew the final
4203 definition of the 'NULL type. This screwed up several key
4204 optimizations in the compiler, causing inefficiency in all sorts
4205 of places. (I found it because I wanted to understand why
4206 GET-INTERNAL-RUN-TIME was consing.)
4207 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4208 PROCLAIM FTYPE information. Unfortunately this broke binary
4209 compatibility again, since now the forms output by DEFGENERIC
4210 to refer to functions which didn't exist in 0.6.3.
4211 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4212 can use the new (as of 0.6.3) transform for SOME into MAP into
4214 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4215 values to (INTEGER 0 1000000), so that the time code will no longer
4216 occasionally get blown up by Linux returning 1000000 microseconds
4217 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4218 its output conform to the ANSI spec. (Alas, this makes its output
4219 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4220 * A full call to MAP NIL with a single sequence argument no longer conses.
4221 * fixes to problems pointed out by Martin Atzmueller:
4222 * The manual page no longer talks about multiprocessing as though
4223 it were currently supported.
4224 * The ILISP support patches have been removed from the distribution,
4225 because as of version 5.10.1, ILISP now supports SBCL without us
4226 having to maintain patches.
4227 * added a modified version of Raymond Toy's recent CMU CL patch for
4228 EQUALP comparison of HASH-TABLE
4230 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4232 * The system still can't cross-compile itself with
4233 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4234 entails), but at least it can compile more of itself that way
4235 than it used to be able to, and various buglets which were uncovered
4236 by trying to cross-compile itself that way have now been fixed.
4237 * This release breaks binary compatibility again. This time
4238 at least I've incremented the FASL file format version to 2, so that the
4239 problem can be detected reliably instead of just causing weird errors.
4240 * various new style warnings:
4241 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4242 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4243 ** using the lexical binding of a variable named in the *FOO* style
4244 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4245 as specified by ANSI.
4246 * *RANDOM-STATE* is no longer automatically initialized from
4247 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4248 default behavior of the system is to repeat its behavior every time
4249 it's run. If you'd like to change this behavior, you can always
4250 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4251 old convention there was no comparably easy way to get the system to
4252 repeat its behavior every time it was run.
4253 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4254 FTYPE declarations has been removed, in favor of their ANSI
4255 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4256 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4257 the types of their sequence arguments can be determined at compile time.
4258 This is done through a new open code expansion for MAP which eliminates
4259 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4260 argument types can be determined at compile time.
4261 * The optimizer now transforms COERCE into an identity operation when it
4262 can prove that the coerced object is already of the correct type. (This
4263 can be a win for machine generated code, including the output of other
4264 optimization transforms, such as the MAP transform above.)
4265 * Credit information has been moved from source file headers into CREDITS.
4266 * Source file headers have been made more standard.
4267 * The CASE macro now compiles without complaining even when it has
4270 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4272 * (Note that the way that the PCL macroexpansions were rewritten
4273 to accommodate the change in DEFGENERIC below breaks binary
4274 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4275 not run under sbcl-0.6.2. Once we get out of alpha releases,
4276 i.e. hit release 1.0.0, we'll probably try to maintain binary
4277 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4278 and sbcl-1.4.4. Until then, however, it might be fairly common
4279 for maintenance releases to break binary compatibility.)
4280 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4282 * The reporting of unhandled signals has been changed to print some
4283 explanatory text as well as the report form. (Previously only
4284 the report form was printed.)
4285 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4286 it defines, so that the compiler no longer issues undefined function
4287 warnings for compiled-but-not-yet-loaded generic functions.
4288 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4289 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4290 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4291 section of the ANSI spec are supported.
4292 * The "" nickname for the "KEYWORD" package has been removed.
4293 The reader still handles symbol tokens which begin with a package marker
4294 as keywords, but it doesn't expose its mechanism for doing so in the
4295 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4296 * The system now issues STYLE-WARNINGs for contradictory TYPE
4297 proclamations. (Warnings for contradictory FTYPE proclamations would
4298 be nice too, but those can't be done usefully unless the type system
4299 is made smarter about FUNCTION types.)
4300 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4301 systematized, so that "*target-*.lisp is supposed to exist only on the
4302 target and imply that there's a related file which exists on the
4303 host, and *host-*.lisp is supposed to exist only on the host and imply
4304 that there's a related file which exists on the target. This involves a
4305 lot of renaming. Hopefully the acute confusion caused by the renaming
4306 will be justified by the reduction in chronic confusion..
4307 ** runtime-type.lisp -> early-target-type.lisp
4308 ** target-type.lisp -> late-target-type.lisp
4309 ** early-host-format.lisp -> early-format.lisp
4310 ** late-host-format.lisp -> late-format.lisp
4311 ** host-error.lisp -> misc-error.lisp
4312 ** early-error.lisp -> early-target-error.lisp
4313 ** late-error.lisp -> late-target-error.lisp
4314 ** host-defboot.lisp -> early-defboot.lisp
4315 ** code/misc.lisp -> code/target-misc.lisp
4316 ** code/host-misc.lisp -> code/misc.lisp
4317 ** code/numbers.lisp -> code/target-numbers.lisp
4318 ** code/early-numbers.lisp -> numbers.lisp
4319 ** early-host-type.lisp -> early-type.lisp
4320 ** late-host-type.lisp -> late-type.lisp
4321 ** host-typep.lisp -> typep.lisp
4322 ** load.lisp -> target-load.lisp
4323 ** host-load.lisp -> load.lisp
4324 ** host-disassem.lisp -> disassem.lisp
4325 ** host-insts.lisp -> insts.lisp
4326 ** byte-comp.lisp -> target-byte-comp.lisp
4327 ** host-byte-comp.lisp -> byte-comp.lisp
4328 ** host-signal.lisp -> signal.lisp
4329 ** host-defstruct.lisp -> defstruct.lisp
4330 ** late-target-type.lisp -> deftypes-for-target.lisp
4331 Furthermore, several other previously target-only files foo.lisp (e.g.
4332 hash-table.lisp and random.lisp) have been split into a target-and-host
4333 foo.lisp file and a target-only target-foo.lisp file, with their key type
4334 definitions in the target-and-host part, so that the cross-compiler will
4335 know more about target types.
4336 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4337 gone away. In their place are various *BACKEND-FOO* variables
4338 corresponding to the slots of the old structure.
4339 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4340 into the target SBCL has been fixed.
4341 * The chill.lisp system for loading cold code into a running SBCL
4343 * Support for the CMU CL "scavenger hook" extension has been removed.
4344 (It was undocumented and unused in the CMU CL sources that SBCL was
4345 derived from, and stale in sbcl-0.6.1.)
4346 * Various errors in the cross-compiler type system were detected
4347 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4348 (enabling various consistency checks). Many of them were fixed,
4349 but some hard problems remain, so the compiler is back to
4350 running without *TYPE-SYSTEM-INITIALIZED* for now.
4351 * As part of the cross-compiler type system cleanup, I implemented
4352 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4353 * I have started adding UNCROSS calls throughout the type system
4354 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4355 on cross-compiler input files will be able to go away, and various
4357 * CONSTANTP now returns true for quoted forms (as explicitly required
4360 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4362 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4363 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4364 (MAKE-LIST -1). In the longer term, it ought to become true
4365 that declarations are assertions even at SAFETY 1. For now, it's not
4366 quite true even at SAFETY 3, but it's at least more nearly true..
4367 (Note that this change seems to increases the size of the system by
4368 O(5%) and to decrease the speed of the compiler by 20% or more.)
4369 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4370 for the problem of printing dozens of lines of distracting information
4371 about low-level system machinery as part of the top stack frame
4372 on entry to the debugger when an undefined function was called.
4373 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4374 is not reset to COMMON-LISP-USER.
4375 * Compilation of stuff related to dyncount.lisp has been made conditional
4376 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4377 smaller. The various dyncount-related symbols have been moved into
4378 a new "SB-DYNCOUNT" package.
4379 * tty-inspect.lisp has been renamed to inspect.lisp.
4380 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4381 feature has gone away. (When we eventually port to other flavors of
4382 libc and/or Unix, we'll try to make the differences between flavors
4383 invisible at the user level.)
4384 * Various other *FEATURES* tags, and/or their associated conditionals,
4385 have been removed if obsolescent, or given better documentation, or
4386 sometimes given more-mnemonic names.
4388 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4390 * tidied up "make.sh" script
4391 * tidied up system directory structure
4392 * better "clean.sh" behavior
4393 * added doc/FOR-CMUCL-DEVELOPERS
4394 * many many small tweaks to output format, e.g. removing possibly-confusing
4395 trailing #\. character in DESCRIBE-INSTANCE
4396 * (EQUALP #\A 'A) no longer signals an error.
4397 * new hashing code, including EQUALP hashing
4398 * tidied up Lisp initialization and toplevel
4399 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4400 * command line argument processing
4401 * added POSIX-GETENV function to deal with Unix-ish environment variables
4402 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4403 terminating SBCL on EOF
4404 * non-verbose GC by default
4405 * There is no more "sbcl" shell script; the sbcl file is now the C
4406 runtime executable (just like CMU CL).
4407 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4409 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4410 symbol and package information as Lisp native symbol and package objects
4411 instead of strings naming symbols and strings naming packages. This way,
4412 whenever packages are renamed (as in warm init), debug information is
4413 transformed along with everything else.
4414 * tweaked the optimization policy declarations which control the building
4415 of SBCL itself. Now, among other things, the system no longer saves
4416 source location debugging information. (This helps two problems at once
4417 by reducing SBCL size and by keeping SBCL from trying to look for its
4418 sources -- which may not exist -- when reporting errors.)
4419 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4420 debugging and testing purposes
4421 * cleaned up printing, making the printer call PRINT-OBJECT for
4422 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4423 methods, giving nearly-ANSI behavior
4424 * converted almost all special variables to use *FOO* naming convention
4425 * deleted PARSE-TIME functionality, since it can be done portably
4426 * moved some files out of cold init into warm init
4427 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4429 * regularized formatting of source files
4430 * added an install.sh script
4431 * fixed ridiculous memory usage of cross-compiler by making
4432 compiler/alloc.lisp not try to do pooling unless it can hook
4433 itself into the GC of the cross-compilation host. Now the system
4434 builds nicely on my old laptop.
4435 * added :SB-ALLOC in target-features.lisp-expr
4436 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4437 was not implemented)
4438 * re-did condition handling and note reporting in the compiler. Notes
4439 are no longer handled by signalling conditions. Style warnings
4440 and warnings are handled more correctly and reported in such a way
4441 that it's easy to find one or the other in your output (so that you
4442 can e.g. figure out which of many problems caused COMPILE-FILE to
4444 * changed the severity of several compiler warnings from full WARNING
4445 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4446 compiler note reporting so that it doesn't use the condition system
4447 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4448 COMPILE-FILE command)
4449 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4450 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4451 no longer supported, since their implementation was deeply intertwingled
4452 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4453 * removed bogus "support" for compiler macros named (SETF FOO), and
4454 removed the compiler macro for SETF INFO (but only after making a fool
4455 of myself on the cmucl-imp mailing list by posting a bogus patch for
4456 DEFINE-COMPILER-MACRO..)
4457 * Compiled files containing forms which have side effects on the Lisp
4458 reader (such as DEFPACKAGE forms) are now handled more correctly.
4459 (Compiler queuing of top level lambdas has been suppressed by setting
4460 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4461 * deleted various currently-unused source files, e.g. gengc.lisp. They
4462 may be added back at some point e.g. when porting to other architectures,
4463 but until they are it's distracting to distribute them and to try to
4465 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4466 there were so many of them they're just distractions, and UNCROSS is
4467 known to be able to handle the current sources
4468 * moved PROFILE functionality into TRACE, so that it will be clear
4469 how the wrapping and unwrapping of functions when you profile them
4470 interacts with the wrapping and unwrapping of functions when you
4471 trace them. (Actually, the functionality isn't there yet, but at least
4472 the interface specification is there. Hopefully, the functionality will
4473 arrive with some future maintenance release.)
4474 * removed host-oops.lisp
4475 * changed signature of QUIT function to allow UNIX-CODE argument
4476 * fixed READ-SEQUENCE bug
4477 * tweaked verbose GC output so that it looks more like the progress
4478 output that ANSI specifies for functions like LOAD
4479 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4480 * added <http://sbcl.sourceforge.com> to the banner information printed by