1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
3 * minor incompatible change: the MSI installer on Windows no longer
4 associates .lisp and .fasl files with the installed SBCL.
5 * minor incompatible change: :UNIX is no longer present in *FEATURES*
6 on Windows. (thanks to Luis Oliviera)
7 * new feature: SB-CLTL2 contrib module now implements
8 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
9 * optimization: scavenging weak pointers is now more efficient,
10 requiring O(1) instead of O(N) per weak pointer to identify
11 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
12 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
13 names, when the class of the instance is a direct instance
14 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
15 as slow as the constant slot-name case.)
16 * optimization: member type construction is now O(N) instead
18 * optimization: UNION and NUNION are now O(N+M) for large
19 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
21 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
22 now more readable in environments like Slime which display it.
23 (thanks to Tobias C. Rittweiler)
24 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
25 was able to infer the correct argument type for the object on which
26 the CAS operation was being performed.
28 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
29 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
30 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
31 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
33 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
34 and generic function now print the object identity for anonymous
35 instances. (thanks to Attila Lendvai)
36 * enhancement: as an extension to MOP, SBCL now supports SETF of
37 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
38 (thanks to Attila Lendvai)
39 * enhancement: sb-cover has an alternate annotation mode which puts
40 the form annotations at the car of the form, instead of annotating
41 it completely. The benefit of this mode is that it shows better
42 which forms were instrumented by the compiler.
43 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
44 by printer control variables. (reported by Dan Corkill)
45 * bug fix: system leaked memory when delayed output was performed by
46 the OS in smaller chunks then expected. (thanks to David Smith)
47 * bug fix: system leaked memory when file streams were not closed
49 * bug fix: large objects written to slow streams that were modified
50 after the write could end up with the modified state written to
51 the underlying file descriptor.
52 * bug fix: multiple threads operating in parallel on the same stream
53 could cause buffer-overflows.
54 * bug fix: source location information is stored correctly
55 (broken since 1.0.6). This bug would generally show up as the
56 Slime debugger higlighting the wrong form when the sldb-show-source
58 * bug fix: Branch forms are again annotated as branches in the sb-cover
60 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
61 on x86-64. (reported by Gregory Vanuxem)
62 * bug fix: EQUALP could return wrong results for structures with raw
63 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
64 word). (reported by Vjacheslav Fyodorov)
65 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
68 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
69 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
70 atomic compare-and-swap operations on threaded platforms.
71 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
72 allows assining a global minimum value to optimization qualities
73 (overriding proclamations and declarations).
74 * enhancement: closed over variables can be stack-allocated on x86
76 * performance bug fix: GETHASH and (SETF GETHASH) are once again
78 * optimization: slot definition lookup is now O(1). This speeds up
79 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
80 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
81 * optimization: ASSOC and MEMBER can now be open-coded for all
82 combinations of keyword arguments when second argument is constant
83 and SPEED >= SPACE. In other cases a specialized version is
85 * bug fix: using obsoleted structure instances with TYPEP and
86 generic functions now signals a sensible error.
87 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
88 (reported by Kristoffer Kvello)
89 * bug fix: backtrace construction is now more careful when making
90 lisp-objects from pointers on the stack, to avoid creating bogus
91 objects that can be seen by the GC.
92 * bug fix: defaulting of values in contexts expecting more than 7
93 variables now works on x86-64. (reported by Christopher Laux)
94 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
96 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
97 as the property-list of a symbol.
98 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
99 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
100 situations requested, are once again file-compileable. (reported
103 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
104 * MOP improvement: support for user-defined subclasses of
105 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
106 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
107 the expansion of DEFMETHOD, is responsible for generating a form
108 which creates a list of specializers when evaluated. Additional
109 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
110 debugging and introspective support.
111 * minor incompatible change: the (unsupported) spinlock interface
112 has changed: free spinlock now has the value NIL, and a held spinlock
113 has the owning thread as its value.
114 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
115 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
117 * enhancement: name of a socket-stream is now "a socket" instead of
119 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
120 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
121 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
122 for generic functions. (Thanks to Tobias C. Rittweiler)
123 * optimization: bignum printing speed has been improved by 20-40%
124 (depending on the bignum size.)
125 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
127 * bug fix: the cache used by the CLOS to store precomputed effective
128 methods, slot offsets, and constant return values is now thread and
130 * bug fix: generic function dispatch function updating is now thread
131 and interrupt safe (in the sense that the known issues have been
133 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
134 * bug fix: interning EQL-specializers is now thread and interrupt safe.
135 * bug fix: asdf systems with dependencies to the SB-POSIX or
136 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
137 * bug fix: interrupt safety of applicable method computation has been
140 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
141 * new contrib: sb-cover, an experimental code coverage tool, is included
143 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
144 significantly faster.
145 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
146 produce more efficient code on x86-64 if the compiler can determine
147 that (+ INDEX <constant>) does not require a bounds check and FOO
148 has an element type at least 8 bits wide.
149 * enhancement: a new, experimental synchronous timeout facility is
150 provided. Refer to SB-SYS:WITH-DEADLINE for details.
151 * enhancement: when a symbol name conflict error arises, the
152 conflicting symbols are always printed with a package prefix.
153 (thanks to Kevin Reid)
154 * enhancement: stepping is now once again supported on the SPARC and
155 MIPS platforms. (It is also now more likely to work on CheneyGC
157 * enhancement: sb-sprof can now also track and report accurate call
159 * bug fixes: the treatment of non-standard subclasses of
160 SB-MOP:SPECIALIZER is more correct.
161 * incompatible change: PURIFY no longer copies the data from the
162 dynamic space into the static and read-only spaces on platforms
163 that use the generational garbage collector
164 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
166 * bug fix: GC race condition occasionally resulting in crashes with
167 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
169 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
171 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
172 a STYLE-WARNING to be signalled (regression from 1.0.4.)
173 * bug fix: an asynchronous interrupt could previously leave the
174 system running with GC inhibited.
175 * bug fix: a DECLARE form evaluated at top-level now causes an error
176 rather than silently (or verbosely) returning NIL.
177 * bug fix: trying to dissassemble functions compiled at high DEBUG could
178 result in a error being signalled due to source form lookup errors.
179 (reported by Peter Graves)
181 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
182 * incompatible change: removed writer methods for host-ent-name,
183 host-ent-addresses -- changing the values did not update the DNS
185 * minor incompatible change: changed experimental JOIN-THREAD interface
186 * documentation: the manual now lists reader and writer methods
187 in class slot documentation sections. (thanks to Richard M Kreuter)
188 * documentation: unwinding from asyncronous events has been
189 documented as unsafe.
190 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
191 in multithreaded application code.
192 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
193 platforms. (thanks to James Anderson for the optimization hint)
194 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
196 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
197 variants no longer cons.
198 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
199 their NOT- variants no longer cons.
200 * optimization: EQUAL hash tables no longer use SXHASH for objects
201 of all data types, but instead use an EQL hash for types for which
202 EQUAL is the same as EQL.
203 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
204 are significantly faster.
205 * optimization: new STRING-OUTPUT-STREAM implementation is much
206 faster for large outputs and conses 30% less on average.
207 * enhancement: XREF information is now collected for references made
208 to global variables using SYMBOL-VALUE and a constant argument.
209 * enhancement: SIGINT now causes a specific condition
210 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
211 * bug fix: FILE-POSITION used to signal an error in some cases where
212 ANSI requires it to return NIL.
213 * bug fix: ADJUST-ARRAY is now interrupt-safe.
214 * bug fix: adding and removing fd-handlers is now interrupt-safe.
215 * bug fix: inlined calls to C now ensure 16byte stack alignment on
217 * bug fix: bad type declaration in the CLOS implementation has
218 been fixed. (reported by James Anderson)
219 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
220 (thanks to Lutz Euler)
221 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
222 required by the CLHS. (reported by Samium Gromoff)
223 * bug fix: dead unbound variable references now signal an error.
224 (reported by Marco Monteiro)
225 * bug fix: / with an unused value was being deleted in safe code.
226 (thanks to Marco Monteiro and Kevin Reid)
227 * bug fix: number of characters that can be written onto a single
228 line in a file is unlimited.
229 * bug fix: some GC deadlocks caused by asynchronous interrupts have
230 been fixed by inhibiting interrupts for when GC is disbled.
231 * bug fix: some interrupt safety issues with GETHASH, (SETF
232 GETHASH), CLRHASH and REMHASH have been fixed.
233 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
234 specifier no longer causes infinite recursion.
235 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
236 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
237 * bug fix: an &environment argument with the correct variable information
238 is passed to macros that are expanded during byte compilation.
239 (reported by Samium Gromoff)
240 * bug fix: a logic error was causing the hash function for bignums to
241 have excessive amounts of collisions. (reported by Faré Rideau)
242 * bug fix: modifying the contents of an array could change the return
243 value of SXHASH on that array, which is only allowed for strings
244 and bit vectors. (bug introduced in 0.9.16)
245 * improvement: the x86-64/darwin port now passes all tests (except
246 for the debugger tests) but should still be considered
247 experimental until this is fixed.
248 * improvement: a style-warning is signaled for CASE (etc) clauses with
249 duplicate keys. (thanks to Kevin Reid)
250 * improvement: macos/x86 and macos/x86-64 now use mach exception
251 handlers for dealing with illegal instructions (for trapping and
252 error handling) and memory protection violations (for GC).
254 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
255 * new platform: experimental support for x86-64/darwin (MacOS).
256 * incompatible change: the thread-safe (on most platforms) getaddrinfo
257 and getnameinfo sockets functions are used instead of gethostbyaddr
258 and gethostbyname, on platforms where the newer functions are available.
259 As a result, the ALIASES field of HOST-ENT will always be NIL on these
261 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
262 don't print the startup banner, but behave as if --noinform was passed
263 as a command line argument. (thanks to Kevin Reid)
264 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
265 * optimization: code using alien values with undeclared types is much faster.
266 * optimization: the compiler is now able to open code SEARCH in more cases.
267 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
268 * bug fix: using standardized COMMON-LISP special variables as loop
269 variables no longer signals bogus package lock violations. (reported
271 * bug fix: declaring local loop variables to be of a range-limited type
272 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
273 (reported by Andras Simon)
274 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
275 bugs remain on x86-64.)
276 * bug fix: the #= and ## reader macros now interact reasonably with
277 funcallable instances.
278 * bug fix: type-checks for function arguments were compiled using the
279 compiler policy settings of the wrong lexical environment (in 1.0.2
281 * bug fix: SHADOW accepts characters as string designators, as required
282 by the spec (thanks to Eric Marsden)
283 * bug fix: fixed GC safety issues when foreign functions are called with
284 non-base strings as arguments
285 * bug fix: more consistent error output for fatal-compiler-warnings, like
287 * bug fix: functions with proper names were showing up as NIL in
288 backtraces (reported by Edi Weitz, regression in 1.0.2)
290 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
291 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
292 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
293 produce more efficient code on the x86 if the compiler can determine
294 that (+ INDEX <constant>) does not require a bounds check and FOO
295 has an element type at least 8 bits wide.
296 * bug fix: references to undefined variables are handled the same way
297 in toplevel forms as in the normal compiler
298 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
300 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
301 (thanks to Jon Buller)
302 * improvement: faster compilation times for complex functions
303 * improvement: added readlink support to SB-POSIX (thanks to Richard
306 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
307 * improvement: experimental support for mach exception handling on
308 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
310 * improvement: support for GBK external format.
311 (thanks to Chun Tian (binghe))
312 * improvement: the debugger now displays variables that have been closed
313 over, in code compiled with (DEBUG 2) or higher
314 * improvement: support for executable cores on NetBSD (thanks to
316 * new feature: added a RESTART-FRAME debugger command
317 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
318 be used to provide an implementation for FILE-POSITION on Gray streams
319 (thanks to Eric Marsden)
320 * new feature: add syslog, openlog and closelog support to SB-POSIX
321 (thanks to Richard Kreuter)
322 * optimization: the function call overhead in code compiled with
323 a high DEBUG optimization setting is significantly
324 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
325 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
326 (thanks to Tony Martinez)
327 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
328 evaluated once (reported by Kevin Reid)
329 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
330 works again (thanks to Kilian Sprotte)
331 * bug fix: an error is signaled for tagbodies with duplicate tags
332 (thanks to Stephen Wilson)
333 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
334 * bug fix: Win32 port can now handle foreign code unwinding Lisp
335 stack frames from alien callbacks.
336 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
337 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
338 65k pages (thanks to David Woodhouse)
339 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
341 changes in sbcl-1.0.1 relative to sbcl-1.0:
342 * new platform: FreeBSD/x86-64, including support for threading.
343 * new feature: the compiler stores cross-referencing information
344 abount function calls (who-calls), macroexpansion (who-macroexpands)
345 and special variables (who-binds, who-sets, who-references) for code
346 compiled with (< SPACE 3). This information is available through the
347 sb-introspect contrib.
348 * new feature: users may subclass SEQUENCE, and have instances of
349 these classes interoperate with standard Common Lisp functions if
350 a number of methods are defined. (This feature is experimental
351 and the interface subject to change based on feedback from SBCL
352 users and the general community)
353 * improvement: sb-sprof traces call stacks to an arbitrary depth on
354 x86/x86-64, rather than the previous fixed depth of 8
355 * improvement: another pthread back-end of mutex "pthread-futex".
356 * bug fix: non-ascii command-line arguments are processed correctly
357 (thanks to Yaroslav Kavenchuk)
358 * bug fix: non-required arguments were not passed correctly when a method
359 defined using DEFMETHOD was called from a mop-generated method using
360 CALL-NEXT-METHOD (reported by Pascal Costanza)
361 * bug fix: recursion is now permitted in accessors through
362 SLOT-UNBOUND. (reported by Pascal Costanza)
363 * bug fix: an error was signaled at startup if the HOME environment
364 variable was defined, but had an empty value (reported by Peter Van Eynde)
365 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
366 signal a PROGRAM-ERROR, not a BUG.
367 * bug fix: SB-POSIX:READDIR works when built with large file support.
368 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
370 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
371 for better automatic indentation support. (thanks to Matt Pillsbury)
372 * optimization: loading generic functions no longer takes O(n^2) time,
373 proportional to the amount of methods in the generic function
374 (reported by Todd Sabin and Jeremy Brown)
375 * optimization: the FIND and POSITION family of sequence functions
376 are significantly faster on arrays whose element types have been
378 * improvements to the Windows port:
379 ** Intermittent heap corruption problems have been fixed. (thanks
380 to Alastair Bridgewater)
381 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
383 ** Lisp is able to unwind foreign exception frames from alien
384 callbacks. (thanks to Alastair Bridgewater)
386 changes in sbcl-1.0 relative to sbcl-0.9.18:
387 * improvement: experimental support for threading on FreeBSD/x86.
388 (thanks to NIIMI Satoshi)
389 * improvement: runtime option --dynamic-space-size can be used
390 to set the size of the dynamic space reserved on startup.
391 * improvement: floating point modes in effect are now saved in
392 core, and restored on startup.
393 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
394 startup, not time since first call to GET-INTERNAL-REAL-TIME.
395 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
396 threads are not running after *SAVE-HOOKS* have run.
397 * improvement: writes to CLOS instance slots are type-checked in code
398 compiled with (SAFETY 3)
399 * improvement: floating-point exception handling on FreeBSD (thanks to
401 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
402 (thanks to Zach Beane)
403 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
405 * improvement: added support for the Shift-JIS external format.
406 (contributed by NIIMI Satoshi)
407 * improvement: callbacks are supported on Linux/PPC. (thanks to
409 * bug fix: compiler bug triggered by a (non-standard) VALUES
410 declaration in a LET* was fixed. (reported by Kaersten Poeck)
411 * bug fix: file compiler no longer confuses validated and already
412 dumped structurres. (reported by Kaersten Poeck)
413 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
414 fill-pointer signals a type-error as required. (thanks to
416 * bug fix: disassemly of funcallable instances works.
417 * bug fix: single stepping on PPC.
418 * bug fix: fix thread-safety problems in the type system (generally
419 manifesting as nonsensical errors like "STRING is a bad type specifier
420 for sequences" or "The value 1 is not of type FIXNUM")
421 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
422 (reported by Josip Gracin)
423 * bug fix: an error is signaled for attempts to displace arrays with
424 incompatible element types (thanks to Mario Mommer)
425 * bug fix: more correct handling of wide characters in debug info
426 (bug reported by Attila Lendvai and fixed by Juho Snellman)
427 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
428 and don't cause extra consing
429 * optimization: MAP and MAP-INTO are significantly faster on vectors
430 whose elements types have been declared.
431 * Improvements to SB-SPROF:
432 ** Support for allocation profiling
433 ** Reduced profiling overhead, especially for long profiling runs
434 * Improvements to the Windows port:
435 ** floating point exceptions are now reported correctly.
436 ** stack exhaustion detection works partially.
437 ** more accurate GET-INTERNAL-REAL-TIME.
438 ** better RUN-PROGRAM behaviour in case of errors in the child
440 ** PROBE-FILE now simplifies pathnames correctly.
441 ** DIRECTORY now works correctly with :WILD-INFERIORS.
442 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
443 1st 1970 (although time timezone data will be unreliable).
444 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
446 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
447 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
448 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
450 * bug fix: two potential GC deadlocks affecting threaded builds.
451 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
452 returns T (reported by Anton Kazennikov)
453 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
454 with non-variable places
455 * bug fix: remove a race condition in the setting of
456 funcallable-instance functions, this should make threaded CLOS
457 code more stable against memory faults.
458 * bug fix: corruption of specials when unbinding is interrupted by an
459 asynchronous unwind (reported by Hannu Koivisto)
460 * improvement: the debugger will now also display local variables that
461 are only used once, for code compiled with a DEBUG optimization quality
464 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
465 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
466 * incompatible change: External-format support for FFI calls. The
467 SB-ALIEN:C-STRING no longer implies an ASCII
468 external-format. Instead the string is subject to external-format
469 conversion. Additionally return values of type SB-ALIEN:C-STRING
470 are by default of type (SIMPLE-ARRAY CHARACTER), not
471 SIMPLE-BASE-STRING. For an alien type that behaves like the old
472 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
473 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
474 * incompatible change: SB-EXT package no longer contains the
475 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
476 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
478 * incompatible change: the single-stepper is no longer available
479 on Alpha, Mips and Sparc platforms.
480 * minor incompatible change: the direct superclasses of
481 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
482 not (STANDARD-OBJECT FUNCTION). This makes the
483 class-precedence-lists of GENERIC-FUNCTION and
484 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
486 * minor incompatible change: the default stream external format on
487 non-unicode SBCL is no longer detected from the locale, but is always
489 * new feature: Add a version of evaluator that uses an interpreter instead
490 of the compiler. EVAL still uses the compiler by default, to switch it
491 to use the interpreter, set the value of the variable
492 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
493 * minor incompatible change: the single-stepper REPL has been merged
494 with the normal debugger (see the "Stepping" heading of the debugger help
495 for more details). The debugger command STEP will no longer switch
496 to the single-stepper REPL.
497 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
498 for a type now works.
499 * bug fix: loading of default sysinit file works. (thanks to Leonid
501 * bug fix: better detection of circularities in the file-compiler.
502 (reported by Marco Monteiro)
503 * bug fix: the CL pathname functions now work with files that have
504 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
505 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
506 systems with Unix98 pty semantics.
507 * bug fix: ASDF-INSTALL will now work with bsd tar.
508 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
510 * bug fix: timers expiring in dead threads no longer cause a
511 type-error (reported by Paul "Nonny Mouse").
512 * bug fix: thanks to more lightweight single-stepper instrumentation,
513 code compiled with (DEBUG 3) will compile and execute significantly faster,
514 and will have more accurate type-inferencing than before
515 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
516 whose bindings are modified
517 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
518 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
519 ** replace dummy implementations of CL:MACHINE-INSTANCE and
520 CL:SOFTWARE-VERSION with proper version
522 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
523 * feature: implemented the READER-METHOD-CLASS and
524 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
525 as specified by AMOP.
526 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
528 * optimization: faster LOGCOUNT implementation on x86 and x86-64
529 (thanks to Lutz Euler)
530 * optimization: hashing of general arrays and vectors has been
531 improved. (reported by Any Fingerhut)
532 * enhancement: SB-INTROSPECT is now able to find definitions of
533 profiled functions. (thanks to Troels Henriksen)
534 * enhancement: compiler-macro expansion applies now to FUNCALL forms
535 as well, allowing compiler-macros for SETF-functions to expand.
536 * enhancement: step-instrumentation no longer wraps known
537 single-value functions with multiple-value context, allowing
538 better type inference.
539 * fixed bug #337: use of MAKE-METHOD in method combination now works
540 even in the presence of user-defined method classes. (reported by
541 Bruno Haible and Pascal Costanza)
542 * fixed bug #339(c): if there are applicable methods not part of any
543 long-form method-combination group, call INVALID-METHOD-ERROR.
544 (reported by Bruno Haible)
545 * fixed bug #361: the :FUNCTION initarg in the protocol for
546 initialization of methods can now be used to override
547 internally-produced optimized functions. (reported by Bruno
549 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
550 system-provided lambda expression no longer cause warnings about
551 unbound #:|pv-table| symbols.
552 * bug fix: improved the handling of type declarations and the
553 detection of violations for keyword arguments with non-constant
555 * bug fix: potentially erronous calls to PATHNAME and
556 MERGE-PATHNAMES were being flushed in some cases.
557 (reported by Richard Kreuter)
558 * bug fix: compiled calls to TYPEP were mishandling obsolete
559 instances. (reported by James Bielman and Attila Lendvai)
560 * bug fix: format strings with ~> without matching ~< no longer
561 trigger an AVER, but signal an understandable error instead.
562 (reported by Antonio Martinez)
563 * bug fix: specifying an output-file without a pathname-type for
564 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
565 of input-file instead of "fasl". (reported by Robert Dodier)
566 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
567 (reported by James Y Knight).
568 * bug fix: compiler-macros-function did not consider the environment
569 argument for shadowing by local functions.
570 * bug fix: compiler-macros expansion was inhibited by local INLINE
572 * bug fix: inline expansions of known functions were subject to
573 step-instrumentation in high debug policies, leading to problems
575 * bug fix: compiler failed to differentiate between different CONS
577 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
578 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
579 * bug fix: #368: incorrect use of expressed vs. upgraded array
581 * bug fix: #306a: more precise unions of array types.
582 * thread-safety improvements:
583 ** CONDITION-WAIT could return early on Linux, if the thread was
584 interrupted and subsequently continued with SIGCONT.
585 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
587 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
588 * added support for the ucs-2 external format. (contributed by Ivan
590 * minor incompatible change: pretty printing of objects of type
591 (cons symbol) is, in the default pprint-dispatch-table, now
592 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
594 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
595 called later in a class's lifetime, possibly as late as when the
596 first instance of the class is created. Previously,
597 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
598 class became finalizeable.
599 * fixed bug: FILE-POSITION sometimes returned inconsistent results
600 for multibyte external-format streams. (thanks to "vbzoli")
601 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
602 with :ALLOCATION :CLASS inherited from superclasses of the
604 * fixed bug: anonymous classes can now be created using the :NAME
605 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
606 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
607 * fixed bug: core-files saved with :EXECUTABLE T can again be
608 executed when SBCL_HOME isn't set. (reported by James Knight)
609 * fixed bug: toplevel LOCALLY forms with declarations could
610 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
611 * fixed bug: printing from several different threads using different
612 values of *print-case* could cause invalid output, due to
613 some internal special variables of the printer not being bound
614 thread-locally (reported by Max Mikhanosha)
615 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
616 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
617 (reported by Pascal Costanza)
618 * fixed bug: FINALIZE-INHERITANCE is called from
619 REINITIALIZE-INSTANCE on classes when the class has previously
620 been finalized, as required by AMOP.
621 * minor code generation optimizations:
622 ** better register allocation in CLOS dispatching functions
623 ** overflow detection when coercing signed bytes to fixnums on x86-64
624 ** is now implemented with one IMUL instruction instead of three shifts
625 ** more efficient bit-vector access on x86 and x86-64
626 ** more efficient access to raw structure slots on x86 and x86-64
627 * fixed some bugs revealed by Paul Dietz' test suite:
628 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
631 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
632 * feature: thread support on Solaris/x86, and experimental thread support
634 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
636 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
637 default method for SLOT-UNBOUND.
638 * minor incompatible change: prevent the user from specializing the
639 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
640 somewhat counter to the intent of the protocol, I (CSR) think, and
641 additionally it just doesn't work in SBCL as currently
642 implemented, thanks to optimizations (that are always valid for
643 the other three SLOT-VALUEish functions, but not for the setter).
644 * minor incompatible change: the :SB-LDB feature is now enabled by
645 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
646 the low-level debugger.
647 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
648 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
649 down to an SB-SYS:FD-STREAM, allowing direct communication in
650 more cases. Temporary files and pipes are still used as a fallback
652 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
653 on the non-reentrant C functions localtime(3) and gmtime(3).
654 * bug fix: global optimization policy was not visible in LOCALLY and
656 * bug fix: class objects can be used as specializers in methods.
657 (reported by Pascal Costanza)
658 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
659 with a type error. (reported by blitz_ on #lisp)
660 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
661 failed with a type error. (reported by Pascal Bourguignon)
662 * bug fix: merging pathnames against defaults with :DIRECTORY
663 starting with '(:RELATIVE :BACK) should preserve the :BACK.
664 (reported by James Y Knight)
665 * bug fix: saving large (>2GB) cores on x86-64 now works
666 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
667 constant index and a (simple-array (signed-byte 32)) array
668 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
669 error (patch by Robert J. Macomber)
670 * feature: TIME now displays the amount of run-time spent in GC
671 * bug fix: The debugger now does a better job of respecting
672 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
673 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
674 workaround for bug 403.)
675 * bug fix: floating point exception handling now works on FreeBSD
676 (thanks to NIIMI Satoshi)
677 * fixed some bugs revealed by Paul Dietz' test suite:
678 ** MISC.641: LET-conversion were not supposed to work in late
680 * improvements to the Win32/x86 port:
681 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
682 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
684 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
685 * new feature: source path information is generated for macro-expansion
686 errors for use in IDE's like Slime (thanks to Helmut Eller)
687 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
688 * bug fix: compiler does not loop forever on an invalid type in
690 * improvement: compilation of most CLOS applications is significantly
692 * optimization: added a limited bytecode compiler for simple toplevel
693 forms, speeding up compilation and FASL loading
694 * bug fix: the statistical profiler now properly distinguishes anonymous
697 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
698 * minor incompatible change: in sbcl-0.9.11 (but not earlier
699 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
700 into subclasses of the finalized class. Now user calls to
701 FINALIZE-INHERITANCE finalize just the one class, and calls by the
702 system return before any subclasses are finalized.
703 * minor incompatible change: The reader no longer ignores errors
704 regarding non-existent packages in #+ and #- feature tests.
705 * new feature: command line options --no-sysinit, --no-userinit to
706 inhibit loading the corresponding init files
707 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
708 for sending data through UDP sockets (thanks to François-René Rideau)
709 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
710 error is signalled instead (thanks to François-René Rideau)
711 * minor incompatible change: Error signalling behaviour of lexical
712 operations violating package locks has changed slightly. Refer to
713 documentation on package locks for details.
714 * enhancement: EVAL can process IF-expressions without resorting to the
716 * bug fix: Some operations involving SIN, COS, or TAN weren't being
717 constant-folded properly due to missing out-of-line functions.
718 (reported by Mika Pihlajamäki)
719 * bug fix: LISTEN sometimes returned T even in cases where no data was
720 immediately available from the stream
721 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
722 were reversed. (reported by Levente Mészáros)
723 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
724 when given a vector displaced to an adjustable array. (reported
726 * bug fix: derivation of float boundaries from numbers outside the
727 appropriate float range (reported by John Wiseman)
728 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
730 * bug fix: precomputing cache entries for generic functions with
731 some subclasses of specializers as yet invalid does not attempt to
732 fill a cache line with a negative offset. (reported by Levente
734 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
735 allow condition class objects as arguments to DOCUMENTATION and
736 (SETF DOCUMENTATION); only find and set documentation for
737 structure names for the STRUCTURE doc-type. (suggested by Gary
739 * improvements to the Win32/x86 port:
740 ** (user-homedir-pathname) and default initialization file
741 locations now know about the user's "Documents and Settings"
742 directory (thanks to Yaroslav Kavenchuk)
743 ** run-program is implemented (thanks to Mike Thomas)
744 ** sockets support (thanks to Timothy Ritchey)
745 ** better backtrace support (thanks to Alastair Bridgewater)
746 ** sb-grovel supported
747 ** asdf-install and sb-posix work somewhat
748 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
749 * improvements to the Solaris/x86 port:
750 ** works on Solaris 11/Solaris Express
751 ** floating-point exception handling support
752 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
753 * fixed some bugs revealed by Paul Dietz' test suite:
754 ** REMOVE-METHOD returns its generic function argument even when
755 the method is not one of the generic functions' methods.
756 ** objects declared as MEMBER types can be admissible arguments to
758 ** printing characters should simply be printed by the FORMAT ~:C
760 ** compiler failure when compiling functions with hairy constant
761 defaults for optional parameters.
762 ** compiler produces wrong code when MAYBE-INLINE-expanding a
763 function, which is already optimized.
765 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
766 * new platform: experimental support for SBCL x86/Darwin, including
767 MacOS X 10.4.5 on Intel.
768 * new feature: Unicode character names are now known to the system
769 (through CHAR-NAME and NAME-CHAR).
770 * new feature: the filesystem location of SBCL's core file is
771 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
772 * minor incompatible change: the contrib modules SB-POSIX and
773 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
774 this change is to make it easier to distribute
775 location-independent binaries.
776 * bug fix: as implied by AMOP, standardized classes no longer have
777 slots named by external symbols of public packages. (reported by
779 * optimization: calling functions via constant symbols -- as in
780 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
781 via the function object as in (FUNCALL #'FOO).
782 * optimization: on x86 and x86-64, the calling convention for the
783 case of dealing with an unknown number of values has been altered
784 to be friendlier to the prediction heuristics implemented,
785 particularly with reference to CALL and RET pairing. (thanks to
786 Alastair Bridgewater)
787 * optimization: on x86, the code for access to thread-local symbol
788 values and binding/unbinding of thread-local symbols is smaller.
789 (thanks to Alastair Bridgewater)
790 * enhancement: CONSTANTP is now able to determine constantness of
791 more complex forms, including calls to constant-foldable standardized
792 functions and some special forms beyond QUOTE.
793 * fixed bug: occasional GC crashes on Solaris/x86
794 * optimization: x86-64 supports stack allocation of results of simple
795 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
796 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
797 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
798 (thanks to James Knight)
799 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
800 by the standard (thanks to Stephen Compall)
802 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
803 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
804 be used for bundling the runtime and the core file into one
805 executable binary. This feature is not currently supported on all SBCL
806 platforms. (thanks to James Bielman and NIIMI Satoshi)
807 * new feature: a generational or ephemeral garbage collector is now
808 the default on the PowerPC platform (both Linux and Darwin). The
809 old Cheney (stop and copy) collector is a build-time option.
810 (thanks to Cyrus Harmon, after Raymond Toy)
811 * minor incompatible change: the method by which SBCL finds its
812 contributed modules has changed; it no longer relies on symbolic
813 links from an $SBCL_HOME/systems directory, but searches directly
814 in the subdirectories of $SBCL_HOME.
815 * enhancement: the dynamic heap size on the Linux/PPC platform is
816 markedly larger, even using the older Cheney garbage collector.
817 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
818 certain complicated string types. (reported by Gary King)
819 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
820 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
821 * fixed bug: DOCUMENTATION on structures no longer signals an
822 error if no structure type of the right name exists. (reported by
824 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
825 applicable non-standard primary or :AROUND method on
826 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
827 no longer cause unbound variable errors. (reported by Kevin Reid)
828 * fixed bug: metacircle resolution in cases where methods have slots
829 added before the slots from STANDARD-METHOD. (reported by Jean
831 * fixed bug: the Power PC assembly code for calling into lisp
832 sometimes computed its return address wrongly (depending on the
833 15th bit of the address, fixed at link time). This used to
834 manifest itself in a segmentation violation while building PCL.
835 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
836 many others over the years)
837 * fixed bug: ROOM no longer reports silly numbers for stack usage if
838 the stack pointer (treated as a signed integer) is negative.
839 (thanks to Peter van Eynde)
841 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
842 * new platform: experimental support for the Windows operating
843 system has been added. (thanks to Alastair Bridgewater)
844 * fixed several bugs in and robustified the PPC FFI (including
845 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
846 * bug fix: allow non-simple string symbol names (reported by Paul
848 * bug fix: interrupt handling on NetBSD (thanks to Richard M
850 * bug fix: saving a core corrupted callbacks on x86/x86-64
851 * bug fix: closed a loophole in metacircularity detection and
852 grounding in the PCL implementation of CLOS.
853 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
854 (reported by tomppa on #lisp)
855 * bug fix: clear the direction flag on Lisp -> C transitions, as
856 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
857 (reported by Andrej Grozin and Hendrik Maryns)
858 * optimization: major improvements to GC efficiency on GENCGC platforms
859 * optimization: faster implementation of EQUAL
860 * optimization: emit more efficient opcodes for some common
861 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
862 * optimization: several other minor code-generation improvements on x86-64
863 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
865 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
866 * minor incompatible change: (SETF CLASS-NAME) and (SETF
867 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
868 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
869 by ANSI as a generic function, and for consistency (SETF
870 GENERIC-FUNCTION-NAME) is treated likewise.
871 * fixed bug #233.b: make constraint propagation notice when a variable
872 value is changed after it is referenced but before it is used
873 * fixed bug #296: no more arbitrary behaviour from filenames with
874 odd characters as --load arguments. (reported by Adam Warner)
875 * fixed bug #390: :CHARACTER-SET pathname components now work as
876 expected. (reported by Tim Daly Jr)
877 * fixed bug #391: complicated :TYPE intersections in slot
878 definitions no longer cause an error in PCL internals.
879 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
880 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
881 returning the number of octets which would be written to the
882 file-stream. (thanks to Robert J. Macomber)
883 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
884 arguments to the internal stream functions. (thanks to David
886 * bug fix: stack exhaustion is now handled on the x86/SunOS
887 platform. (thanks to Daisuke Homma)
888 * bug fix: more accurate ROOM results on GENCGC platforms
889 * optimization: improved type inference for arithmetic-for
890 index variables in LOOP
891 * optimization: faster floating-point SQRT on x86-64
892 * fixed some bugs revealed by Paul Dietz' test suite:
893 ** DOCUMENTATION returns NIL instead of "" for method combinations
894 that don't have a docstring
896 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
897 * minor incompatible change: (SETF CLASS-NAME) and (SETF
898 GENERIC-FUNCTION-NAME) are no longer generic functions, and
899 therefore are not customizeable by user code (as seems to be at
900 least permitted and maybe required by AMOP). As a consolation,
901 however, the SBCL implementation of these functions now calls
902 REINITIALIZE-INSTANCE as specified by AMOP.
903 * bug fix: slot-definition documentation is propagated as per ANSI
904 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
905 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
906 argument, as do eslotd initialization methods. (from Pascal
907 Costanza's "Closer" project)
908 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
909 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
911 * bug fix: it is now possible to have more than one subclass of
912 STANDARD-GENERIC-FUNCTION without causing stack overflow.
913 (reported by Bruno Haible, Pascal Costanza and others)
914 * bug fix: the dependent update protocol now works for generic
915 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
917 * bug fix: condition-class instances corresponding to
918 DEFINE-CONDITION forms are now created eagerly. (reported by
919 Kalle Olavi Niemitalo on comp.lang.lisp)
920 * bug fix: floating point printing is more accurate in some
921 circumstances. (thanks to Simon Alexander)
922 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
923 merged with *DEFAULT-PATHNAME-DEFAULTS*.
924 * bug fix: callbacks on OS X now preserve stack-alignment.
925 (thanks to Cyrus Harmon)
926 * enhancement: the x86-64 disassembler is much better at
927 disassembling SSE instructions. (thanks to Lutz Euler)
928 * enhancement: CHECK-TYPE on a variable now allows type-inference
929 (thanks to Vincent Arkesteijn)
930 * optimization: improved performance of EUC-JP external format.
931 (thanks to NIIMI Satoshi)
932 * optimization: performance improvements to IO on file streams of
933 :ELEMENT-TYPE CHARACTER
934 * optimization: much faster memory allocation on x86-64
935 * optimization: faster garbage collections (latency and throughput)
937 * optimization: faster bignum subtraction and division on x86 and x86-64
938 (thanks to Lutz Euler)
939 * optimization: more accurate type-inference for loops that use a
940 floating point index variable or a negative step.
942 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
943 * bug fix: add a workaround to SBCL looping infinitely at startup on
944 Linux kernels with apparently buggy implementations of personality().
945 (thanks to Svein Ove Aas)
946 * bug fix: Unicode symbols are correctly printed in LDB backtraces
947 (thanks to David Lichteblau)
948 * bug fix: local bindings shadow global symbol macros for
949 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
950 * optimization: non-open coded uses of numeric comparison operators
951 (e.g. >) no longer cons when called with more than one parameter
952 on platforms supporting dynamic-extent allocation.
953 * enhancement: saving cores with foreign code loaded is now supported
954 on MIPS/Linux in addition to the previously supported platforms.
955 * bug fix: FIND-RESTART now tests for activity, not applicability when given
956 a restart object as identifier. (reported by Helmut Eller for CMUCL)
957 * bug fix: division by zero in sb-sprof when no samples were collected
958 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
959 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
960 (thanks to Alexander Kjeldaas)
961 * new feature: ignoring the return values of destructive functions
962 like NREVERSE signals a compile-time style-warning.
963 (thanks to Kevin Reid)
964 * new feature: the alignment of alien structure fields can be
965 explicitly specified. (thanks to Cyrus Harmon)
966 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
967 and dump core on SIGQUIT
969 ** incompatible change: threads do not inherit values of specials
970 from their parents (see manual)
971 ** bug fix: threads stacks belonging to dead threads are freed by the
972 next exiting thread, no need to gc to collect thread stacks anymore
973 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
974 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
975 ** bug fix: dynamic variable and thread start related gc lossage
976 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
978 * fixed some bugs revealed by Paul Dietz' test suite:
979 ** SUBTYPEP is slightly more accurate on heinously complicated
980 CONS types where some of the members have uncertain (in the
981 NIL, NIL sense) type relationships to each other.
983 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
984 no longer purified unless :PURIFY T is explicitly specified.
985 ** Non-purified cores are significantly smaller than before
987 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
988 * new feature: timers based on Zach Beane's excellent timer package
989 * added support for the following external formats: koi8-u,
990 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
991 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
992 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
993 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
994 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
995 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
996 * incompatible change: a threaded SBCL will no longer revert to
997 non-threaded mode on non-NPTL systems, but refuse to start entirely.
998 * bug fix: interrupts are disabled until startup is complete; no
999 more sigsegvs when receiving a signal to soon
1000 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1002 * bug fix: add a workaround for the memory randomization features in
1003 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1004 workaround will only be in effect on systems with the proc filesystem
1006 * bug fix: printing objects of type HASH-TABLE signals a
1007 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1009 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1010 have been running for over 50 days. (reported by Gilbert Baumann)
1011 * bug fix: the logic for getting names of functions gets less
1012 confused when confronded with alternate-metaclass
1013 funcallable-instances. (reported by Cyrus Harmon)
1014 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1015 consistent, even on internal alternate-metaclass objects.
1016 * bug fix: flush closure information collected by physical
1017 environment analysis prepass before the main pass. (bug reported
1019 * bug fix: compiler pack phase does not modify a hashtable, which is
1020 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1021 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1022 may be expanded. (reported by Alexander <malishev> on c.l.l)
1023 * bug fix: no more style-warnings when compiling code using LAST
1024 under high-SPEED low-SPACE settings. (reported by David Wallin
1026 * fixed some incompatibilities between SBCL's MOP and the MOP
1028 ** the METAOBJECT class is now implemented;
1029 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1030 STANDARD-OBJECT, as required;
1031 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1032 now compatible (as required by VALIDATE-SUPERCLASS); there
1033 remains a consistency requirement over the presence or absence
1034 of the FUNCTION class in the superclass list of finalized
1035 classes; see the manual for more details;
1036 ** the :AROUND method for COMPUTE-SLOTS on
1037 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1038 requested slot ordering.
1040 ** bug fix: parent thread now can be gc'ed even with a live
1042 ** bug fix: binding a special with PROGV to no value is not
1043 the same as that symbol not having been bound (thanks to
1045 ** bug fix: binding specials is thread safe (thanks to
1047 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1048 and SB-SPROF do not die with 'no handler for signal XX in
1049 interrupt_handle_now(..)' anymore
1050 ** bug fix: WITH-TIMEOUT works with multiple running threads
1051 * fixed some bugs revealed by Paul Dietz' test suite:
1052 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1053 the :method-class keyword argument.
1055 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1056 * new port: the Solaris operating system on x86 processors is now
1057 mostly supported, though some rough edges in the environment
1058 remain. (thanks to Daisuke Homma)
1059 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1060 space for its heap. (thanks to Thiemo Seufer)
1061 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1062 closures. (thanks to Thiemo Seufer)
1063 * minor incompatible change: eof selects abort in the debugger.
1064 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1065 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1067 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1068 deprecated, and will go away in a future revision of SBCL.
1069 * minor incompatible change: GC-ON and GC-OFF are no longer
1070 implemented with a counter, it does not matter how many times gc
1071 is switched on or off
1072 * bug fix: discriminating functions for generic function classes
1073 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1074 make invalid assumptions about method precedence order. (reported
1076 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1077 platforms. (thanks to Thiemo Seufer)
1078 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1079 * bug fix: correct pathname printing with printer escaping is on.
1080 (thanks to Kevin Reid)
1081 * bug fix: complex VOP definitions in "user-space" no longer trigger
1082 package locks. (reported by Zach Beane)
1083 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1084 for funcallable-instance functions, and is no different from
1086 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1087 characters in the current readtable. (reported by Nicholas Neuss)
1088 * bug fix: SUBTYPEP on various CONS types returns more a more
1089 accurate acknowledgment of its certainty.
1090 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1091 lists in some cases. This partially fixes bug 384.
1092 * flush all standard streams before prompting in the REPL and the
1094 * bug fix: signal handling and triggering gc do not conflict
1095 directly anymore, in particular a high frequency sb-sprof does
1096 not prevent gc from running
1097 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1098 approximation for timezone and DST information between the
1099 universal time epoch and the smallest negative 32-bit time_t.
1100 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1101 year 1899 so long as the encoded time is non-negative.
1102 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1103 the correct answer for non-constant rotate amounts when performing
1104 an inline 32-bit rotation.
1106 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1107 there is only one thread in the session
1108 ** bug fix: memory leak for streams created in one thread and
1109 written to in another
1110 ** bug fix: lockup when compiled with gcc4
1111 ** bug fix: race that allows the gc to be triggered when gc is
1113 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1114 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1116 ** bug fix: finished threads are gc'ed properly
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1119 has arguments with a different set of applicable methods from
1120 the orignal arguments.
1121 ** The type error thrown by MAP now has a correct expected-type
1123 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1124 name a compiled function.
1125 ** Three MISC tests where a large bignum was improperly coerced to
1126 a float that couldn't represent that bignum during type
1127 derivation were fixed.
1128 ** SUBTYPEP can now handle types involving the negation of a
1129 list-form FUNCTION type.
1130 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1131 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1132 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1134 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1135 * New feature: Experimental support for bivalent streams: streams
1136 opened with :element-type :default now allow character and binary
1137 (unsigned-byte 8) I/O
1138 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1139 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1141 * bug fix: no more highly sporadic "couldn't check whether ~S is
1142 readable" when reading a stream and an interrupt hits in the middle
1143 of a select system call
1144 * compiler better recognizes complex arrays (reported by Eduardo
1146 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1147 for variables on SBCL built with linkage-tables. (reported by Luis
1149 * various error reporting improvements.
1150 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1151 (Thanks to Thiemo Seufer)
1152 * bug fix: nested reader invokations work correctly
1153 * bug fix: it is possible to have simultaneous references to foreign
1154 code and foreign data with the same name.
1156 ** added x86-64 support
1157 ** incompatible change: the threading api now works with thread
1158 objects instead of thread ids
1159 ** bug fix: threads are protected from signals and interruption when
1160 starting up or going down
1161 ** bug fix: a race where an exiting thread could lose its stack to gc
1162 ** bug fix: don't halt on infinite error in threads if possible
1163 ** fixed numerous gc deadlocks introduced in the pthread merge
1164 ** bug fix: fixed thread safety issues in read and print
1165 ** bug fix: debugger doesn't hang on session lock if interrupted at
1166 an inappropriate moment
1167 ** bug fix: run-program is now thread safe(r)
1168 ** bug fix: inner with-recursive-lock no longer releases the mutex
1169 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1170 * fixed some bugs revealed by Paul Dietz' test suite:
1171 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1172 EXPECTED-TYPE slots filled.
1173 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1175 ** MAKE-CONDITION accepts classes as type-designators.
1176 ** COMPILE may never return NIL.
1177 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1178 range before calling Unix time functions
1180 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1181 * numerous signal handling fixes to increase stability
1182 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1183 * minor incompatible change: we now correctly canonize default
1184 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1185 as per the MOP, rather than the historical (INITARG INITFUNCTION
1186 INITFORM). (reported by Bruno Haible)
1187 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1188 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1189 for more information.
1190 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1191 pathname is a directory pathname.
1192 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1193 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1195 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1196 :METACLASS argument in addition to a class name. (reported by
1197 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1198 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1199 (reported by Bruno Haible; patch from Gerd Moellmann)
1200 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1202 * bug fix: more cleanups to the floating point exception handling on
1203 x86-64 (thanks to James Knight)
1204 * bug fix: the compiler does not try to constant fold calls of
1205 COERCE and BIT-* functions when they return freshly constructed
1206 objects. (reported by Nikodemus Siivola and Paul Dietz)
1207 * optimization: improved the MIPS versions of generic (in the
1208 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1209 * optimization: direct conversion of (unsigned-byte 32) to floats on
1210 the PowerPC platform.
1211 * optimization: structure instances with raw slots now use less
1212 memory, and probably show better memory locality. (thanks to
1214 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1215 are treated as requests for stack allocation on the x86-64,
1216 Alpha-32, PPC and SPARC platforms.
1217 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1218 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1219 (thanks to Tony Martinez)
1221 ** gcing a dead thread can no longer lead to lockups
1222 ** threads block signals until they are set up properly
1223 ** errno is no longer shared by threads
1224 ** interrupt-thread restores the eflags register on x86
1225 ** fixed some lockups due to gc/thread interaction
1226 * fixed some bugs revealed by Paul Dietz' test suite:
1227 ** invalid dotted lists no longer raise a read error when
1228 *READ-SUPPRESS* is T
1229 ** signal an error if a symbol that names a declaration is used
1230 as the name of a type, or vice versa
1231 ** allow using the (declare (typespec var*)) abbreviation for
1232 (declare (type typespec var*)) with all type specifiers
1233 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1234 ** always return NIL from PPRINT-POP when OBJECT is NIL
1235 ** don't signal errors when pretty-printing malformed LABELS,
1236 FLET or MACROLET forms
1237 ** declarations in a DOLIST body are also in scope for the
1239 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1241 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1244 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1245 * fixed cross-compiler leakages that prevented building a 32-bit
1246 target with a 64-bit host compiler.
1247 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1248 opened with :IF-EXISTS :OVERWRITE.
1249 * fixed bug 281: error for an invalid qualifier in a short-form method
1250 combination method is not signalled until the faulty method is called.
1251 * bug fix: iteration variable type inferrer failed to deal with open
1252 intervals. (reported by Alan Shields)
1253 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1254 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1255 or more runtime options were provided to the sbcl binary.
1256 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1257 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1258 * optimization: there's now a fast-path for fixnum arguments in the
1259 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1260 * optimization: the code generated on x86-64 is more compact thanks
1261 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1262 * bug fix: floating-point exception handling now partly works on
1263 x86-64. (Thanks to James Knight)
1264 * improvement to the MIPS backend from Thiemo Seufer: C-style
1265 64-bit long long arguments and return values to alien functions
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** the type-error signalled from WARN has a filled-in DATUM slot.
1269 ** the type-error required when a stream is not associated with
1270 a file has the stream as its datum.
1271 ** type-errors on single-floats on x86-64 no longer have
1272 :INVALID-OBJECT as the datum
1273 ** the type-errors signalled for invalid function names now have
1274 a correct expected type
1275 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1276 for structures defined with a :TYPE. Documentation strings for
1277 typed structures are no longer immediately discarded
1278 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1279 broadcast streams and synonym streams. FILE-LENGTH now also works
1280 on broadcast streams.
1282 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1283 * incompatible change: the --noprogrammer option, deprecated since
1284 version 0.7.5, has been removed. Please use the equivalent
1285 --disable-debugger option instead.
1286 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1288 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1289 inoperational for a while now) has been completely removed.
1290 * null lexical environments are now printed as #<NULL-LEXENV>,
1291 significantly reducing the amount of clutter in typical backtraces.
1292 * documentation on weak pointers, finalization, and after GC hooks
1293 has been added to the manual.
1294 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1295 as other specialized array types, is much faster. SUBSEQ and
1296 COPY-SEQ on such arrays have also been sped up.
1297 * optimization: EQL is now more efficient when at least other argument
1298 is known to be of type (OR FIXNUM (NOT NUMBER)).
1299 * fixed bug: the runtime is now less vulnerable to changes in the
1300 size of the SBCL object on OS X, and virtual memory is reserved for
1301 all spaces that need to be at a fixed address.
1302 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1303 * fixed bug: finalizers and after GC hooks that cause consing are now
1305 * fixed bug: compiler error messages and summaries are now printed to
1306 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1307 * fixed inference of the upper bound of an iteration variable.
1308 (reported by Rajat Datta).
1309 * fixed bug 211e: calling local functions with duplicated constant
1310 keyword argument no longer causes a bogus style warning about an
1312 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1313 ftype declaration to be disregarded. (reported by Dave Roberts)
1314 * fixed bug 373: caused by erronous compilation of references to alien
1315 variables in the runtime on ppc/darwin.
1316 * fixed bug 376: CONJUGATE type deriver.
1317 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1318 (reported by Baughn on #lisp)
1319 * a cross-compiler bug on non-x86 platforms has been identified and
1320 fixed. (thanks to Bruno Haible)
1321 * improvements to the MIPS runtime code for increased stability.
1322 (thanks to Thiemo Seufer)
1323 * increased the maximimum compact environment size to allow
1324 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1325 * partial workaround for bug 135: don't name the function we're
1326 calling for hairy cases of EVAL, so as not to accumulate one environment
1327 entry per call to EVAL. (reported by Kevin Reid)
1328 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1329 * optimization: type testing for non-vector arrays should be faster.
1330 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1332 * bug fix: setting 31st element of a bit vector to zero did not work
1334 * bug fix: redefining a class definition which failed due to a
1335 previous accessor / function clash now works (but see BUGS entry
1336 #380 for more problems in this area). (thanks to Zach Beane)
1337 * the long form of DEFINE-METHOD-COMBINATION disables method group
1338 checking when given a single method group with pattern *. (thanks
1340 * on x86 compiler supports stack allocation of results of simple
1341 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1342 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1343 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1344 * fixed some bugs related to Unicode integration:
1345 ** the restarts for recovering from input and output encoding
1346 errors only appear when there is in fact such an error to
1348 * fixed some bugs revealed by Paul Dietz' test suite:
1349 ** MISC.549 and similar: late transformation of unsafe type
1350 assertions into derived types caused unexpected code
1352 ** SCALE-FLOAT type deriver is less wrong.
1353 ** type derivers for EXP, LOG and similar functions compute result
1354 types for complex arguments better.
1355 ** (MISC.563) CONJUGATE type deriver works for very restricted
1357 ** out-of-line type testers for character strings are available.
1358 ** EQUAL compiler transform understands specialness of objects
1360 ** accessing double-floats stored on the stack now works on x86-64.
1361 ** debugger internals could sometimes create invalid lispobjs,
1362 resulting in GC crashes.
1363 ** MISC.548: type check weakening can convert required type into
1365 ** initialization forms for bindings are not in scope of free special
1368 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1369 * incompatible change: thread support for non-NPTL systems has
1370 been removed - locking is buggy and unreliable. A threaded
1371 SBCL build will now warn at startup and refuse to create
1372 new threads, unless futex support is detected in the kernel
1373 * incompatible change: the top level REPL now has only an ABORT
1374 restart associated with it, not TOPLEVEL and ABORT as it used to.
1375 TOP and TOPLEVEL are now available as debugger commands for
1376 returning to the top level.
1377 * incompatible change: forms evaluated in the REPL now use the
1378 global optimization policy.
1379 * incompatible change: user- and system-initialization files are
1380 no longer processed with LOAD, but by READ and EVAL; hence the
1381 global optimization policy, startup package, readtable, etc,
1383 * The .fasl file format number has been incremented because of
1384 various incompatible changes.
1385 * internal entry point details and argument counts no longer appear
1386 in backtraces unless explicitly requested by setting
1387 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1388 * built-in and standard functions no longer have names like "top
1389 level local call to FOO".
1390 * fixed bug 32: functions defined in non-null lexical environments
1391 now have more legible printed representation
1392 * fixed bug 33: functions defined in non-null lexical environemnts
1393 are now more amenable to inspection by INSPECT.
1394 * workaround for bug 354: XEPs no longer appear in backtraces unless
1395 explicitly requested.
1396 * fixed bug: receiving the signal which results from attempting to
1397 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1398 SIGBUS on other *BSDs) on architectures where the C stack is also
1399 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1400 to debugger confusion or wild instability. (reported by Cheuksan
1402 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1403 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1404 (reported by Lutz Euler)
1405 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1406 compiled code. (reported by Paul Dietz)
1407 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1408 specialized parameter to a method to something that is not TYPEP
1409 the specializer is now possible.
1410 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1411 face of package deletion.
1412 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1413 pathname. (reported by David Tolpin on c.l.l)
1414 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1415 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1416 STRING-STREAM and FILE-STREAM, respectively.
1417 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1418 than previously; in addition, its test suite is now run on build.
1419 (thanks to Luke Gorrie)
1420 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1421 * optimization: added a immediate representation for single-floats
1423 * fixed some bugs revealed by Paul Dietz' test suite:
1424 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1425 ** attempting to create a package with a colliding nickname causes
1426 correctable errors to be signalled.
1427 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1428 ** several x86-64 backend bugs related to sign-extension of immediate
1431 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1432 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1433 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1435 * fixed loading of multiply forward-referenced layouts.
1436 (thanks to Cheuksan Wang)
1437 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1438 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1439 (see COMPILE-FILE documentation for details of :PRINT option)
1440 * fixed bugs 19 and 317: fixed-format floating point printing is
1441 more accurate. This also fixes a bug reported by Adam Warner
1442 related to the ~@F format directive.
1443 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1445 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1446 dispatch table if the from-char is a dispatch macro character.
1447 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1448 by the wordsize no longer ignore the last word. (reported by Lutz
1450 * fixed bug in type checking of dynamic-extent variables. (reported
1452 * optimization: sequence traversal functions use their freedom to
1453 coerce function designators to functions.
1454 * optimization: code with many calls to CLOS methods specialized on
1455 CLOS classes has had redundant type checks removed.
1456 * optimization: added declarations to speed up operations that access
1457 the internal character database (for example STRING-UPCASE)
1458 * optimization: comparison operations between floats and sufficiently small
1459 fixnums no longer create extra rationals
1460 * fixed some bugs related to Unicode integration:
1461 ** portions of multibyte characters at the end of buffers for
1462 character-based file input are correctly transferred to the
1463 start of the buffer at the next read.
1464 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1465 passing it through to OPEN.
1466 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1467 argument given to internal calls to OPEN.
1468 ** the built-in comment readers (introduced by character sequences
1469 ";" and "#|") are more forgiving to encoding errors; they will
1470 STYLE-WARN and then attempt to resync the stream at a character
1471 boundary. (thanks to Teemu Kalvas)
1472 * fixed some bugs in the x86-64 port:
1473 ** Negative short int return values from c-calls are sign-extended
1475 ** The stack is aligned to 16-bytes for c-calls, as required by
1476 the ABI. (thanks to Cheuksan Wang)
1477 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1478 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1479 for enums. (thanks to Vincent Arkesteijn)
1480 ** Multiple small optimizations and bugfixes for floating point
1482 * fixed some bugs revealed by Paul Dietz' test suite:
1483 ** Space, Tab, Linefeed, Return and Page have the invalid
1484 secondary constituent character trait.
1485 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1487 ** WITH-INPUT-FROM-STRING should only update the index place on
1489 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1490 works more reliably.
1491 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1492 with one-dimensional arrays with fill pointers.
1493 ** TYPE-OF failed on a complex with an integer realpart and a
1495 ** compiler failure during type inference for the code of form
1496 (IF (EQL X (THE ...)) ...) (MISC.535).
1498 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1499 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1500 platform. The port supports 61-bit fixnums, large memory spaces
1501 and reloading shared object files.
1502 * enhancement: saving cores with foreign code loaded is now
1503 supported on ppc/Darwin in addition to the previously supported
1505 * enhancement: the statistical profiler now walks deeper into the
1506 call stack for more meaningful call-graphs and accrued time
1507 reports (x86/x86-64 only). It also now reports time spent in
1509 * enhancement: it is now possible to trace most individual methods
1510 of a generic function in addition to tracing the generic function
1512 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1514 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1515 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1516 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1517 produces an error. (thanks to Vincent Arkesteijn)
1518 * bug fix: NAMESTRING on pathnames with :WILD components in their
1519 directories works correctly. (thanks to Artem V. Andreev)
1520 * fixed bug 125: compiler preserves identity of closures. (reported
1522 * bug fixed: functions with &REST arguments sometimes failed with
1523 "Undefined function" when compiled with (DEBUG 3). (reported by
1525 * bug fix: overflow during compiling of setting element of a bit
1526 vector with constant index and value. (reported by Timmy Douglas)
1527 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1528 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1529 * on x86 compiler partially supports stack allocation of dynamic-extent
1531 * GO and RETURN-FROM do not check the extent of their exit points
1532 when compiled with SAFETY 0.
1533 * fixed some bugs related to Unicode integration:
1534 ** encoding and decoding errors are now much more robustly
1535 handled; it should now be possible to recover even from invalid
1536 input or output to the terminal. (thanks to Teemu Kalvas)
1537 ** provided a first cut at implementing STRING-TO-OCTETS and
1538 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1539 ** altered the SB-MD5 contributed module slightly, changing the
1540 interface just enough for it to be supportable for builds where
1541 lisp characters are not eight bits.
1542 * fixed some bugs revealed by Paul Dietz' test suite:
1543 ** the FORMATTER-generated functions for ~V[ conditionals require
1544 the correct number of arguments.
1545 ** READ-FROM-STRING returns the mandated second value when applied
1546 to displaced strings.
1547 ** the #\Rubout and #\Backspace characters are treated as invalid
1548 constituent characters by the tokenizer.
1550 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1551 * new feature: reloading changed shared object files with
1552 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1553 * new feature: references to foreign variables and functions
1554 can now be compiled and loaded before the corresponding shared
1555 object file is loaded, as long as the foreign definitions are
1556 available at runtime.
1557 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1558 supported. (thanks to Dan Debertin)
1559 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1560 just keywords. (thanks to Vincent Arkesteijn)
1561 * fixed bug #331: structure-class instances corresponding to
1562 DEFSTRUCT forms are now created eagerly.
1563 * fixed bug #345: backtraces from calls to undefined functions work
1564 on x86 as well. Related bug #61 is now also partially fixed on x86
1565 (backtraces from throws to unknown catch tags.)
1566 * bug fix: lambda-list parsing is now stricter vrt. order and number
1567 of lambda-list keywords.
1568 * bug fix: as specified by AMOP, an error is signalled if a
1569 class-option appears multiple times in a DEFCLASS form. (reported
1571 * bug fix: RANDOM can be compiled when the compiler derives the type
1572 of its numeric argument as a disjoint set of small integers.
1573 (reported by Paul Dietz)
1574 * bug fix: starting a core saved with shared objects loaded when
1575 those objects are not available no longer causes threaded SBCL to
1576 hang. (reported by Sean Ross)
1577 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1579 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1580 argument, which must always be NIL. (reported by Kalle Niemitalo)
1581 * bug fix: printing 1.0d+23 no longer results in an error.
1582 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1583 * bug fix: structure slot setters preserve evaluation order. (thanks
1585 * fixed some bugs related to Unicode integration:
1586 ** RUN-PROGRAM can allow its child to take input from a Lisp
1587 stream. (reported by Stefan Scholl)
1588 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1589 locales. (reported by Ken Causey)
1590 * fixed some bugs revealed by Paul Dietz' test suite:
1591 ** INCF, DECF and REMF evaluate their place form as specified in
1593 ** FORMATTER expands ~{ ~} iteration directives with V or #
1594 parameters correctly.
1595 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1596 consequent uses no arguments correctly.
1597 ** the system has a partial understanding of the (COMPLEX RATIO)
1599 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1602 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1603 * new feature: a build-time option (controlled by the :SB-UNICODE
1604 keyword feature, enabled by default) for building the system with
1605 support for the entire 21-bit character space defined by the
1607 * new feature: the system now has rudimentary external-format
1608 support; the primary user-visible change at this time is that
1609 characters with the high bit set (such as lower-case-e-acute) will
1610 print correctly to a terminal in a UTF-8 environment.
1611 * minor incompatible change: BASE-CHAR no longer names a class;
1612 however, CHARACTER continues to do so, as required by ANSI.
1613 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1614 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1615 has been moved to the SB-EXT package (temporarily re-exported from
1617 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1618 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1620 * fixed bug #351: better error-handlind and reporting for malformed
1622 * fixed bug #350: bignum-printing is now more memory-efficient,
1623 allowing printing of very large bignums, eg. (expt 2 10000000).
1624 (reported by Bruno Haible)
1625 * fixed bug #302: better primitive-type selection for intersection
1627 * fixed bug #308: non-graphic characters now all have names, as
1628 required. (reported by Bruno Haible)
1629 * bug fix: redefining a class with different superclasses now correctly
1630 removes it from the direct-subclasses of its previous superclasses.
1631 (reported by David Morse)
1632 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1633 new value now works. (reported by Bruno Haible)
1634 * bug fix: correct canonicalization of multiple non-standard slot
1635 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1636 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1637 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1638 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1639 now exists, an signals an error.
1640 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1641 during macroexpansion for non-iteration variables as well. (reported
1642 by Bruno Haible for CMUCL)
1643 * bug fix: Cyclic structures and unprintable objects in compiler
1644 messages no longer cause errors. (reported by Bruno Haible)
1645 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1646 and READ-FROM-STRING all now return a primary value of NIL if
1647 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1648 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1649 (reported by Bruno Haible for CMUCL)
1650 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1651 do not designate a condition. (reported by Bruno Haible for
1653 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1654 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1655 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1656 permitting resolution in favour of any of the conflicting symbols.
1657 (reported by Bruno Haible for CMUCL)
1658 * bug fix: EQUAL compiler optimizations is less aggressive on
1659 strings which can potentially compare true despite having distinct
1660 specialized array element types.
1661 * bug fix: unit enumerations can be defined without dividing by
1662 zero. (thanks to Vincent Arkesteijn)
1663 * FORMAT compile-time argument count checking has been enhanced.
1664 (report from Bruno Haible for CMUCL)
1665 * a partial workaround for the bug 262: the compiler does not try to
1666 inline-expand a local function doing RETURN-FROM from a deleted
1667 BLOCK. (thanks to Peter Denno for the bug report and to David
1668 Wragg for the simple test case)
1669 * fixed some bugs revealed by Paul Dietz' test suite:
1670 ** RENAME-PACKAGE allows all package designators as new package
1672 ** constraint propagation and lambda variable substitution are
1673 more cautious in dealing with partially deleted code.
1674 ** compiler sometimes lost reoptimization passes.
1675 ** CERROR, given a condition as condition designator, passes its
1676 remaining arguments to the continue format control without
1678 ** Case-altering FORMAT directives work correctly on non-ASCII
1680 ** The REAL type specifier handles bounds outside the floating
1681 point ranges without signalling FLOATING-POINT-OVERFLOW.
1682 ** Functions with IR1-transformations can create intercomponent
1683 references to global functions.
1684 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1686 ** FORMAT ~R treats a nil value for its first parameter correctly.
1688 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1689 * enhancement: saving cores with foreign code loaded is now
1690 supported on x86/NetBSD and sparc/Linux in addition to the previously
1691 supported platforms.
1692 * bug fix: on some platforms repeated installations caused multiple
1693 copies of HTML documentation to be installed -- should not happen
1694 any more. (reported by Stefan Scholl)
1695 * bug fix: parsing self-recursive alien record types multiple times
1696 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1697 original patch by Helmut Eller for CMUCL)
1698 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1699 (thanks to Richard Kreuter)
1700 * bug fix: defining classes whose accessors are methods on existing
1701 generic functions in other (locked) packages no longer signals
1702 bogus package lock violations. (reported by François-René Rideau)
1703 * bug fix: special variables as DEFMETHOD parameters no longer have
1704 associated bogus type declarations. (reported by David Wragg and
1706 * bug fix: read-write consistency on streams of element-type
1707 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1708 * bug fix: redefiniton of the only method of a generic function with
1709 no DEFGENERIC no longer emits a full WARNING. In addition,
1710 redefinition of generic functions with no DEFGENERIC to an
1711 incompatible lambda list now signals an error. (thanks to Zach
1713 * bug fix: DEFGENERIC now works even when there's a function of the
1714 same name in an enclosing lexical environment. (thanks to Zach
1716 * fixed compiler failure, caused by instrumenting code during
1717 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1718 * optimization: added loop analysis and improved register allocation
1719 * fixed some bugs revealed by Paul Dietz' test suite:
1720 ** POSITION on displaced vectors with non-zero displacement
1721 returns the right answer.
1722 ** (SIMPLE-STRING) is a valid type specifier for sequence
1724 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1726 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1727 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1729 ** The justification version of the FORMAT ~< directive treats
1730 non-zero minpad parameter correctly.
1732 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1733 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1734 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1735 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1736 the supported interface.
1737 * new feature: Single-stepping of code compiled with DEBUG 2 or
1738 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1739 * new feature: saving cores with foreign code loaded is now
1740 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1741 Timothy Moore's work for CMUCL)
1742 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1743 parameters to * instead of NIL if no initform is supplied.
1744 (reported by Johan Bockgaard)
1745 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1746 correctly when FUNCALL appears as the car of the form. Note:
1747 despite this FUNCALL forms are not currently subject to
1748 compiler-macro expansion. (port of Raymond Toy's fix for the
1749 same from CMUCL, reported by Johan Bockgaard)
1750 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1751 (thanks for Teemu Kalvas)
1752 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1753 case instead of terminating the iteration (thanks for Julian
1754 Squires, Sean Champ and Raymond Toy)
1755 * bug fix: incorrect expansion of defgeneric that caused a style
1756 warning. (thanks for Zach Beane)
1757 * bug fix: slot accessor effective method computation works properly
1758 for classes with multiple non-standard applicable methods on
1759 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1760 * on x86 compiler supports stack allocation of results of LIST and
1761 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1762 CMUCL implementation by Gerd Moellmann)
1763 * fixed some bugs revealed by Paul Dietz' test suite:
1764 ** FORMAT strings with both the ~<~:;~> form of the justification
1765 directive and pretty-printing directives cause an error.
1767 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1768 * incompatible change: the internal functions
1769 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1770 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1771 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1772 instead of the old functions.
1773 * new feature: on platforms where "dladdr" is available foreign
1774 function names now appear in backtraces. (based on Helmut Eller's
1776 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1777 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1779 * documentation: ASDF manual is now distributed with SBCL.
1780 * bug fix: SBCL can now load its contributed modules with REQUIRE
1781 even if the system-provided entries have been removed from
1782 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1784 * bug fix: backtraces involving undefined functions or assembly
1785 routines are more informative. (thanks to Brian Downing)
1786 * bug fix: mutually referent alien structures now work correctly.
1787 (reported by Rick Taube)
1788 * bug fix: structures defined by WITH-ALIEN can be referred to
1789 within other definitions in the same WITH-ALIEN.
1790 * bug fix: division operators (MOD, TRUNCATE and the like) with
1791 constant zero divisors and integer dividends no longer generate
1793 * bug fix: provide default methods for INPUT-STREAM-P and
1794 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1795 * bug fix: improve the detection and resolution of MOP metacycles
1796 (where effective method computation depends on the generic function
1797 having its effective method computed). (reported by Bruno Haible)
1798 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1799 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1800 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1801 * bug fix: pathnames with relative directory components can be
1802 represented relative to default pathnames.
1803 * optimization: in taking the GCD of bignums, reduce the two bignums
1804 to approximately the same size (using Euclid's algorithm) before
1805 applying the more sophisticated binary GCD. (thanks to Juho
1807 * optimization: COUNT on bitvectors now operates word-at-a-time.
1808 * optimization: ASH with a positive, but not necessarily constant,
1809 (leftwards) shift, when in a modular context, is compiled to a
1811 * fixed some bugs revealed by Paul Dietz' test suite:
1812 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1813 if the corresponding argument is NIL.
1814 ** FORMAT directives accepting a minpad parameter treat negative
1816 ** literal commas as character parameters to FORMAT directives are
1818 ** literal spaces directly after ~<Newline> directives within a
1819 format-logical-block (~:< ~@:>) do not induce :FILL-style
1820 conditional newlines.
1821 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1822 as specified: it no longer includes conditional newlines).
1823 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1825 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1826 :PER-LINE-PREFIX argument does not evaluate to a string.
1828 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1829 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1830 the manual for details; for now, package locks can be disabled by
1831 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1832 if no major problems are found then it is likely that they will be
1833 compiled in unconditionally.
1834 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1835 unsupported operators on all platforms. To load a shared library
1836 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1837 object file, link it into a shared library outside of SBCL and
1838 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1840 * bug workaround: ROOM T and the GC were not working together
1841 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1842 reporting facility aren't preserved. That reporting has been
1843 disabled, so now until and unless someone figures out how to make it
1844 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1845 * minor incompatible change: as threatened around sbcl-0.8.0, the
1846 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1847 The symbols are also exported from SB-PCL for backwards
1848 compatibility, but more so than before SB-PCL should be treated as
1849 an implementation-internal package.
1850 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1852 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1853 It is known as of this release not to work on the Alpha, however.
1854 * fixed bug #167: errors signalled due to illegal syntax in method
1855 bodies are now more legible.
1856 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1857 designators and can hence be used with TYPEP.
1858 * fixed bug #333: CHECK-TYPE now ensures that the type error
1859 signalled, if any, has the right object to be accessed by
1860 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1861 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1862 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1864 * fixed bug #334: programmatic addition of slots using specialized
1865 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1866 and :ALLOCATION :CLASS slots.
1867 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1868 float exponents. (rereported by Peter Seibel)
1869 * fixed a bug: #\Space (and other whitespace characters) are no
1870 longer considered to be macro characters in standard syntax by
1871 GET-MACRO-CHARACTER.
1872 * fixed bug: initialization of condition class metaobjects no longer
1873 causes an instance of the condition to be created. (reported by
1875 * fixed bug: it is now possible to have slots such that
1876 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1877 description is neither :INSTANCE nor :CLASS.
1878 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1879 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1880 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1881 non-standard SLOT-DEFINITION-ALLOCATION values.
1882 * fixed bugs: various race conditions handling exiting threads.
1883 CL-PPCRE's thread torture test now passes.
1884 * fixed arguably-a-bug: GC time proportional to number of threads in
1885 system even when most of them are idle
1886 * optimization: improved performance of BIT and SBIT on bit-vectors.
1887 * .fasl file incompatibility: The fasl file version number has
1888 been incremented because of changes associated with package locks.
1890 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1891 * minor incompatible change: the system no longer provides
1892 optimization hints (conditions which are TYPEP
1893 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1894 mode; these hints are still emitted when higher SPEED optimization
1896 * new contrib module: a sampling profiler (profiling by statistical
1897 sampling, rather than by instrumenting functions) is available as
1898 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1899 * the behaviour of the standard function ED is now customizeable by
1900 third parties through a hook variable: see ED's documentation
1901 string for information on the protocol.
1902 * the compiler no longer emits efficiency notes for (FUNCALL X)
1903 when the type of X is uncertain under default optimization
1905 * fixed bug 276: mutating a binding of a specialized parameter to a
1906 method to something that is not TYPEP the specializer is now
1908 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1909 exhibit the required behaviour on the x86 platform. (thanks to
1910 Peter van Eynde, Eric Marsden and Bruno Haible)
1911 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1912 even for difficult arguments. (reported by Peter Graves)
1913 * fixed bug 141a: the backquote printer now descends quoted
1915 * fixed another bug in backquote printing: no more destructive
1916 modification of the form's list structure. (reported by Brian
1918 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1919 move between its address being taken and the call to
1920 interrupt_thread, fixing a crashing race condition.
1921 * the SB-POSIX contrib implementation has been adjusted so that it
1922 no longer exhibits ridiculously poor performance when constructing
1923 instances corresponding to C structs.
1925 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1926 * minor incompatible change: the sb-grovel contrib now treats C
1927 structures as alien (in the sense of SB-ALIEN) objects rather than
1928 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1929 has implications for memory management of client code
1930 (sb-grovel-returned objects must now be manually managed) and for
1931 type safety (alien objects now have full types).
1932 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1933 used to control emission of compiler diagnostics, rather than the
1934 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1935 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1936 quality should be considered deprecated.
1937 * install.sh now installs the user manual as well
1938 * (not quite a new documentable feature, but worth considering in
1939 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1940 beginnings of a semantically meaningful condition hierarchy is
1941 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1943 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1944 designator as the defaults argument.
1945 * fixed bug: Displaced arrays whose displaced-to array has become
1946 too small now cause ARRAY-DIMENSION to signal an error, providing
1947 for safer bounds-checking. (reported by Bruno Haible)
1948 * fixed bug: DEFCLASS slot definitions with identical :READER and
1949 :WRITER names now signal a reasonable error. (reported by Thomas
1951 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1953 * fixed bug: Invalid slot specification errors now print correctly.
1954 (thanks to Zach Beane)
1955 * fixed bug 320: Shared to local slot value transfers in class
1956 redefinitions now happen corrently from superclasses as
1957 well. (reported by Bruno Haible)
1958 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1960 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1961 improper lists correctly. (reported by Bruno Haible)
1962 * fixed bug 313: source-transform for <fun-name> was erroneously
1963 applied to a call of a value of a variable with name <fun-name>.
1964 (reported by Antonio Menezes Leitao)
1965 * fixed bug 307: The obsolete instance protocol ensures that
1966 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1967 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1968 remove dead unknown-values globs from the middle of the stack.
1969 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1970 easier to resume long computations after using *BREAK-ON-SIGNALS*
1971 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1972 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1973 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1974 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1975 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1976 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1978 * fixed a (fairly theoretical) bug in string printing: if
1979 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1980 does not have array-element-type equal to the most general string
1982 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1983 function name. (thanks to Nikodemus Siivola)
1984 * fixed bug: compile-time format string checker failed on
1985 non-closed ~{. (reported by Thomas F Burdick)
1986 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1987 don't warn on legal loop code involving more than one aggregate
1988 boolean. (thanks to Nikodemus Siivola)
1989 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1990 in keyword position to MAKE-PACKAGE was not being reported as
1992 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1993 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1995 * fixed bugs 280 and 312: the checking for multiple definitions in a
1996 file is less likely to become confused by uses of inline
1998 * fixed bug: the #S reader macro performs the keyword coercion
1999 specified for slot names. (reported by Kalle Niemitalo)
2000 * fixed bug: lambda lists may contain symbols whose names start with
2001 & but are not lambda-list-keywords; their occurrence triggers a
2003 * fixed bug 321: define-method-combination argument lambda lists do
2004 not cause invalid code to be generated when &optional or &aux
2005 variables have default values. (reported by Bruno Haible)
2006 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2007 have CLOS classes; this appears also to have fixed reported
2008 instability in stack exhaustion detection.
2009 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2010 MUFFLE-WARNING when no associated restart is present is now
2012 * optimization: rearranged the expansion of various defining macros
2013 so that each expands into only one top-level form in a
2014 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2016 * optimization: used a previously little-used slot in symbols to
2017 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2019 * fixed some bugs revealed by Paul Dietz' test suite:
2020 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2022 ** Fixed an optimization bug related to inheritance of initforms
2023 from local to shared slots.
2024 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2025 ** CAST optimizer forgot to flush argument derived type.
2026 ** print/read consistency on floats is now orders of magnitude
2027 more likely. (thanks also to Bruno Haible for a similar report
2029 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2030 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2031 for objects of type REAL. Make it so.
2032 ** SXHASH is specified (infelicitously) to respect similarity,
2033 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2034 it so. (thanks to Markus Ziegler)
2035 ** on the Alpha, the compiler succeeds in compiling functions
2036 returning a known number of arguments greater than 63.
2037 ** fixed handling of invalid NIL arguments in keyword position
2039 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2041 ** provided a readably-printable representation for RANDOM-STATE
2043 ** ensured that pathnames, where they have a namestring, always
2044 print using #P"..." syntax.
2046 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2047 * Support for the forthcoming 2.0 version of the NetBSD kernel
2048 running on x86 hardware has been added. (thanks to Perry
2049 E. Metzger most immediately, and others for their past work)
2050 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2051 current release version; SBCL's OpenBSD support had been broken
2052 since about the time of OpenBSD's switch to ELF binary format.)
2053 * [placeholder for DX summary]
2054 ** user code with &REST lists declared dynamic-extent, under high
2055 speed or space and low safety and debug optimization policy.
2056 * The manual has been converted to Texinfo format and the debugger
2057 chapter from the cmucl manual has been added.
2058 * A facility has been added to extract documentation strings from
2059 sbcl and store them as Texinfo-formatted snippets for inclusion in
2060 the manual (via Texinfo's @include directive)
2061 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2062 the test case to Dave Roberts)
2063 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2064 pointers. (reported by Sean Ross)
2065 * bug fix: PROFILE output is printed nicely even for large numerical
2066 values. (thanks to Zach Beane)
2067 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2068 greater than 32 handle EOF correctly.
2069 * bug fix: on X86 an immediate argument of the IMUL instruction is
2070 correctly printed in disassembly. (thanks to Lutz Euler)
2071 * bug fix: class slots in redefined classes preserve their old
2072 values. (thanks to Bruno Haible and Nikodemus Siivola)
2073 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2074 succeeds. (reported by Marco Baringer)
2075 * fixed some bugs revealed by Paul Dietz' test suite:
2076 ** READ-SEQUENCE now works on ECHO-STREAMs.
2077 ** RATIONALIZE works more according to its specification. (thanks
2080 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2081 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2082 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2083 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2084 should matter to you only if you rebind the printer control
2085 variables and then find you want different bindings in the
2086 debugger than in the ordinary execution of your program.)
2087 * The runtime build system has been tweaked to support building
2088 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2089 assembler and linker. (thanks to Nikodemus Siivola)
2090 * Unbound, undefined, undeclared variables now trigger full
2091 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2092 more often programmer error than deliberate exploitation of undefined
2094 * optimization: the hash algorithm for strings has changed to one
2095 that is less vulnerable to spurious collisions. (thanks to Juho
2097 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2098 needless bounds checking. (thanks to Juho Snellman)
2099 * optimization: implemented multiplication as a modular
2100 (UNSIGNED-BYTE 32) operation on the PPC backend.
2101 * fixed some bugs revealed by Paul Dietz' test suite:
2102 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2104 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2105 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2106 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2108 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2110 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2112 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2113 their output stream on EOF from read.
2114 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2115 have been read to end-of-file.
2116 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2118 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2119 description of determination of which consecutive characters
2121 ** Printing the "Space" character with escaping on now yields "#\\ ",
2122 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2123 ** Reading floating-point numbers with *READ-BASE* set to a number
2124 less than 10 works correctly.
2125 ** Reading floating-point numbers with *READ-BASE* set to a number
2126 more than 10 works correctly.
2127 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2128 the readtable currently in effect.
2130 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2131 * minor incompatible change: parsing of namestrings on a physical
2132 (Unix) host has changed; numbers after the final #\. in a
2133 namestring are no longer interpreted as a version field. This is
2134 intented to be largely invisible to the user, except that the
2135 meaning of the namestring "*.*.*" has changed: it now refers to a
2136 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2137 should usually be replaced by
2138 (make-pathname :name :wild :type :wild :version :wild)
2139 with the added benefit that this is more likely to be portable.
2140 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2141 option to OPEN now signals an error if the file being opened
2142 exists; this may have an impact on existing code.
2143 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2144 signals received, on the PowerPC platforms (both Linux and
2145 Darwin). (thanks to Pierre Mai for pointing out the location of
2147 * several fixes on OS X: The system now builds and runs cleanly on
2148 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2149 (thanks to Brian Mastenbrook)
2150 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2151 second-resolution: integer multiples of 1/3600 between -24 and 24.
2152 (thanks to Vincent Arkesteijn)
2153 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2154 type when called with 1 argument; PEEK-CHAR checked type of
2155 PEEK-TYPE only after having read first character from a
2156 stream. (reported by Peter Graves)
2157 * bug fix: the garbage collector now has much better locality
2158 behaviour, and in particular no longer treats all memory as being
2159 exhausted when many small objects point to each other in a deeply
2161 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2162 recognized as being TYPEP their class.
2163 * bug fix: the PUSHNEW documentation string has been corrected.
2164 (thanks to Vincent Arkesteijn)
2165 * bug fix: defaulting of the value for the last of an atypically
2166 large number of multiple values being bound was not being
2167 performed correctly on the Alpha or PPC platforms
2168 * optimization: implemented multiplication as a modular
2169 (UNSIGNED-BYTE 32) operation on the x86 backend.
2170 * optimization: SEARCH on simple-base-strings can now be open-coded.
2171 (see also contrib/compiler-extras.lisp for inspiration for
2172 teaching the compiler about the Boyer-Moore algorithm).
2173 * value, returned by MAX (and MIN) called with several EQUALP, but
2174 not EQL, arguments now does not depend on compiler settings.
2175 * fixed some bugs revealed by Paul Dietz' test suite:
2176 ** in stack analysis liveness information is propagated from
2177 non-local entry points.
2178 ** pathwise CAST removing failed when the CAST node did not start
2180 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2181 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2183 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2184 host is already defined.
2185 ** RENAME-FILE works on streams instead of signalling an internal
2187 ** PEEK-CHAR uses the current readtable when determining whether
2188 or not a character is whitespace.
2189 ** MERGE-PATHNAMES handles the case when the pathname does not
2190 specify a name while the default-pathname specifies a version.
2191 ** Pathnames now stand a better chance of respecting print/read
2193 ** Attempting to use standardized file system operators with a
2194 pathname with invalid :DIRECTORY components signals a
2196 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2198 ** DIRECTORY on logical pathnames is more correct.
2199 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2200 signal a TYPE-ERROR if their argument is not a stream.
2201 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2202 designator argument does not designate a stream.
2203 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2204 examining the synonym.
2205 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2207 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2208 element-type larger than ([UN]SIGNED-BYTE 32).
2210 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2211 * When built with the :SB-FUTEX feature, threaded builds now take
2212 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2213 for faster/more reliable mutex and condition variable support.
2214 * Incompatible change (but one you probably shouldn't have been using
2215 anyway): the interface and code for arbitrating between multiple
2216 threads in the same user session has been redesigned.
2217 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2218 error when called without an explicit environment argument.
2219 (thanks to Nikodemus Siivola)
2220 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2221 confused about their position. (thanks to Adam Warner and Gerd
2223 * bug fix: Pretty printing backquoted forms with unquotations in the
2224 argument list position of various code constructs such as LAMBDA
2225 now works correctly. (reported by Paul Dietz)
2226 * bug fix: Pretty printing unquotations no longer loses all
2227 stream position information.
2228 * optimization: performance of string output streams is now less
2229 poor for multiple small sequence writes.
2230 * optimization: performance of CSUBTYPEP in the presence of complex
2231 expressions involving CONS and NOT many times has been improved.
2232 (reported by Paul Dietz)
2233 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2235 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2236 streams for the REPL, for the debugger, and so on. (thanks to
2238 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2239 * fixed some bugs revealed by Paul Dietz' test suite:
2240 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2241 is ignored if it is too small, rather than propagating through
2242 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2243 ** extremely complex negations of CONS types were not being
2244 sufficiently canonicalized, leading to inconsistencies in
2246 ** VALUES tranformer lost derived type.
2248 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2249 * fixed a bootstrapping bug: the build process no longer assumes
2250 that the various BOOLE-related constants have the same value in
2251 host and target lisps. (noted by Paul Dietz' test suite on an
2252 SBCL binary built from CLISP)
2253 * The system can now be dynamically linked on the MIPS platform,
2254 which enables dynamic loading of foreign code from Lisp. (thanks
2255 to Ralf Baechle for discussions on the MIPS ABI)
2256 * The system now records debugging information for its own source
2257 files in a filesystem-position-independent manner, relative to
2258 the "SYS" logical host.
2259 * fixed a compiler bug: MV-LET convertion did not check references
2260 to the "max args" entry point. (reported by Brian Downing)
2261 * tweaked disassembly notes to be less confident about proclaiming
2262 some instruction as an LRA. (thanks to Brian Downing)
2263 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2264 now each have their own history, command character, and other
2265 characteristics. (thanks to David Lichteblau)
2266 * fixed some bugs revealed by Paul Dietz' test suite:
2267 ** compiler failure in compiling LOGAND expressions including a
2269 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2270 shift greater than 32.
2271 ** FLUSH-DEST did not mark blocks for type check regeneration.
2272 ** HANDLER-CASE failed to accept declarations in handler clauses
2273 in some circumstances.
2275 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2276 * New code in contrib/sb-introspect (still probably not entirely
2277 stable yet) provides some support for smart Lisp development
2278 environments like SLIME.
2279 * The conditions signalled for errors occurring when loading .fasl
2280 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2281 in a way which should help ASDF recover gracefully.
2282 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2283 stuff has been cleaned up. If you code contrib/ stuff, this might
2284 affect you, and you can look at contrib/README, contrib/STANDARDS,
2285 and/or the 0.8.4.27 diff to check.
2286 * In full calls the compiler now does not generate checks for declared
2287 argument types for all arguments.
2288 * various threading fixes
2289 ** and some experimental patches which didn't make it into
2290 the main tree for this release, but which are shipped in
2291 contrib/experimental-thread.patch as a possible fix for some
2292 failures (deadlock, spinning...) in GC-intensive multithreaded
2294 * fixed PPC build problem (source code incompatibility of different
2295 library versions): added offsetof() hackery which attempts to divine
2296 where glibc maintainers put uc_mcontext today
2297 * fixed bug 282: compiler does not trust type assertions while passing
2298 arguments to a full call.
2299 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2300 &OPTIONAL VALUES type parameter.
2301 * fix bug 214: algorithm for noting rejected templates is now more
2302 similar to that of template seletion. (also reported by rydis on
2304 * fixed bug 141b: printing backquoted information readably and prettily
2305 inserts a space where necessary.
2306 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2307 (CHARACTER 10) are now reported as errors, rather than propagated
2308 as unknown types. (reported by piso on #lisp)
2309 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2310 with values NIL and :ERROR. (thanks to Milan Zamazal)
2311 * fixed bug 191c: CLOS now does proper keyword argument checking as
2312 described in CLHS 7.6.5 and 7.6.5.1.
2313 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2314 longer raise an error; further, using a list as a for-as-arithmetic
2315 counter now raises a meaningful error.
2316 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2317 now understood by sequence creation functions such as MAKE-SEQUENCE
2319 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2320 a STRING-INPUT-STREAM.
2321 * compiler enhancement: SIGNUM is now better able to derive the type
2323 * type declarations inside WITH-SLOTS are checked. (reported by
2325 * fixed some bugs revealed by Paul Dietz' test suite:
2326 ** incorrect optimization of TRUNCATE for a positive first
2327 argument and negative second.
2328 ** compiler failure in let-convertion during flushing dead code.
2329 ** compiler failure while deriving type of TRUNCATE on an
2330 interval, containing 0.
2331 ** ASH of a negative bignum by a negative bignum count now returns
2333 ** intersection of CONS types now canonicalizes properly, fixing
2334 inconsistencies in SUBTYPEP.
2336 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2337 * incompatible change: The --disable-debugger command line
2338 option now clobbers the debugger at a more fundamental
2339 level, by redefining #'INVOKE-DEBUGGER instead of by
2340 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2341 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2342 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2343 enter the debugger REPL and then suffer recursive errors
2344 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2345 while under the new system, BREAK in --disable-debugger mode
2346 terminates the system just as an unhandled error would.
2347 * fixed compiler performance when processing loops with a step >1;
2348 * bug fix: DOCUMENTATION now retrieves generic function
2349 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2350 support has been systematized, and now supports the methods
2351 specified by ANSI, along with a default method and a method for
2352 slot documentation. (reported by Nathan Froyd)
2353 * bug fix: effective methods associated with a generic function are
2354 no longer cached over a change of that generic function's method
2355 combination. (reported by Andreas Fuchs)
2356 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2357 names a non-existent directory. (thanks to Andreas Fuchs)
2358 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2359 the PPC platform, lead to stack corruption; this has been fixed.
2360 (reported by Rainer Joswig)
2361 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2362 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2363 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2364 arithmetic in general, is now much more reliable on the PPC
2366 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2367 platform now returns the right answer.
2368 * optimization: restored some effective method precomputation in
2369 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2370 precomputation is now tunable.
2371 * optimization: compiler-internal data structure use has been
2372 reviewed, and changes have been made that should improve the
2373 performance of the compiler by about 20%.
2374 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2375 simple-base-strings and simple-bit-vectors is improved.
2376 * optimization: the optimization of 32-bit logical and arithmetic
2377 functions introduced in version 0.8.3 on the x86 has been
2378 implemented on the mips, ppc and sparc platforms; an
2379 implementation of the same facility, but for 64-bit arithmetic,
2380 has been added for the alpha.
2381 * microoptimization: the compiler is better able to make use of the
2382 x86 LEA instruction for multiplication by constants.
2383 * bug fix: in some situations compiler did not report usage of
2384 generic arithmetic in (SPEED 3) policy.
2385 * bug 145b fix: compiler used wrong type specifier while converting
2386 MEMBER-types to numeric.
2387 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2389 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2390 index argument now works properly on the Alpha platform.
2391 * bug fix: floating point exception treatment on the Alpha platform
2393 * bug fix: FILE-POSITION works much better on string input and
2394 output streams. (thanks to Nikodemus Siivola)
2395 * bug fix: many threading/garbage collection symptoms sorted.
2396 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2397 might be pseudo-atomic.
2398 * internal change: Stopping for GC is now done with signals not
2399 ptrace. GC is now done in whichever thread wanted it, instead of
2401 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2403 * bug fix: result form in DO is not contained in the implicit
2405 * incompatible change: ICR structure is changed; the value part of
2406 CONTINUATION is now called LVAR; corresponding functions are
2407 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2408 * added type deriver for ISQRT (thanks to Robert E. Brown).
2409 * bug fix: better support for loading from the command line when an
2410 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2412 * fixed some bugs revealed by Paul Dietz' test suite:
2413 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2414 ** ROUND and FROUND now give the right answer when given very
2415 small float arguments.
2416 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2418 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2419 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2420 ** LCM should return a non-negative integer.
2421 ** PARSE-INTEGER returned the index of a terminator instead of the
2422 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2423 ** PARSE-INTEGER returned an incorrect index being applied to a
2425 ** LCM with two arguments of 0 returns 0 rather than signalling
2427 ** unsigned addition of a 32-bit constant with the high bit set no
2428 longer causes an internal compiler error on the x86.
2429 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2430 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2431 with negative last argument.
2432 ** byte specifiers with zero size and position no longer cause
2433 an error during type derivation.
2434 ** bignum multiplication on the Alpha platform now returns the
2436 * porting: The system now builds on SuSE AMD64, although it still
2437 generates a 32-bit binary.
2438 * .fasl file incompatibility: The fasl file version number has
2439 been incremented (because of the changes to internal compiler
2440 data structures referred to above).
2442 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2443 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2444 more accurately, on the Darwin kernel running on PowerPC hardware.
2445 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2446 * Compiler code deletion notes now signal a condition of type
2447 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2448 an associated MUFFLE-WARNING restart.
2449 * The compiler now performs limited argument count validation of
2450 constant format strings in FORMAT, and where appropriate in ERROR,
2451 CERROR and WARN. (thanks to Gerd Moellmann)
2452 * New ASDF-INSTALL contrib can be used for automatic download and
2453 installation of third-party Lisp code from CCLAN or other sites
2455 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2456 which forces another thread to execute a function supplied by the
2458 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2459 now accept and act upon their :ELEMENT-TYPE keyword argument.
2460 (reported by Martin Atzmueller, Edi Weitz)
2461 * bug fix: FILE-POSITION now accepts position designators up to
2462 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2463 is the greater. (thanks to Patrik Nordebo)
2464 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2465 Kalle Olavi Niemitalo)
2466 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2467 printable. (reported by Eric Marsden)
2468 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2469 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2470 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2471 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2472 (thanks to Stig Sandoe)
2473 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2474 UNREAD-CHAR now correctly includes the unread character in the
2475 target sequence. (thanks to Gerd Moellmann)
2476 * bug fix in threaded builds: the system can now be suspended and
2477 resumed by shell job control with minimal disruption.
2478 * bug fixes in times and timezones >2038AD
2479 * better handling of "where is GNU make?" problem in build scripts
2480 (thanks to Nikodemus Siivola)
2481 * new optimization: inside a named function any reference to a
2482 function with the same name is considered to be a self-reference;
2483 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2484 optimization quality.
2485 * new optimization on x86: logical functions and + now have
2486 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2487 used when the result is truncated to 32 bits.
2488 * VALUES declaration is partially enabled.
2489 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2490 * fixed some bugs revealed by Paul Dietz' test suite:
2491 ** The system now obeys the constraint imposed by
2492 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2493 types form a lattice under type intersection.
2494 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2495 ** ASSOC now ignores NIL elements in an alist.
2496 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2497 and (1+ MOST-POSITIVE-FIXNUM) answers.
2498 ** The addition of a method with invalid qualifiers to a generic
2499 function does not cause an error to be signalled immediately;
2500 a warning is signalled, and the error is generated only on
2501 calling the generic function.
2502 * changed .fasl file version number, in part to add type codes for
2503 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2504 obscure ANSI requirements
2506 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2507 * fixed bug 148: failure to inline-expand a local function left
2508 garbage, confusing the compiler.
2509 * fixed bugs 3cd: structure slot readers perform type check if the
2510 slot can have an invalid value (i.e. it is either not initialized
2511 or can be written with a less specific slot writer).
2512 * bug fix: the compiler now traps array references to elements off
2513 the end of an array; previously, the bounds checking in some
2514 circumstances could go off-by-one.
2515 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2517 * type declarations for array element types now obey the description
2518 on the CLHS page "Declaration TYPE", as per discussions on
2519 sbcl-help around 2003-05-08. This means that a declaration
2520 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2521 declaration, all references to BAR will be asserted or assumed
2522 (with THE, so dependent on compiler policy) to involve objects of
2523 type FOO. Note that no such declaration is implied in
2524 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2525 * declared types of functions from the "Conditions"
2526 chapter. (reported by Paul Dietz)
2527 * bug fix: CERROR accepts a function as its first argument.
2528 * bug fix: NTH an NTHCDR accept a bignum as index
2529 arguments. (reported by Adam Warner)
2530 * optimization: character compare routines now optimize comparing
2531 against a constant character. (reported by Gilbert Baumann)
2532 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2533 argument now works properly on the MIPS platform.
2534 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2535 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2536 declared to have a functional type, does not cause an error
2538 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2539 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2540 * fixed bug: if last continuation of a deleted block has a
2541 destination, this destination should be deleted too. (reported by
2543 * fixed a bug in the bootstrap process: the host compiler's values
2544 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2545 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2546 test case from Patrik Nordebo)
2547 * improved the ability of the disassembler on the PPC platform to
2548 provide helpful disassembly notes.
2549 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2550 the class in more cases than previously.
2551 * bug fix: FILE-POSITION now understands :START and :END for
2552 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2553 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2554 8) data. (thanks to David Lichteblau)
2555 * bug fix: it is possible to add a method to a generic function
2556 without lambda list.
2557 * bug fix: reader failed to signal END-OF-FILE inside an
2558 object representation. (reported by Nikodemus Siivola)
2559 * fixed some bugs revealed by Paul Dietz' test suite:
2560 ** LAST and [N]BUTLAST should accept a bignum.
2561 ** condition slot accessors are methods.
2562 ** (VECTOR NIL) is a subtype of STRING.
2564 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2565 * minor incompatible change: some nonsensical specialized lambda
2566 lists (used in DEFMETHOD) which were previously ignored now signal
2568 * minor incompatible change: the system is now aware of the types of
2569 variables in the COMMON-LISP package, and will signal errors for
2570 most violations of these type constraints (where previously they
2571 were silently accepted).
2572 * minor incompatible change: COMPILE-FILE now uses the freedom
2573 afforded (ANSI 3.2.2.3) to use derived function types for
2574 functions defined in the same file. This also permits the system
2575 to warn on static type mismatches and function
2576 redefinition. (Currently it does not work with high DEBUG level.)
2577 * minor incompatible change: VALUES declaration is disabled.
2578 * When issuing notes, the compiler now signals a condition of type
2579 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2580 restart for use in user handlers. It is expected that the
2581 COMPILER-NOTE condition will eventually become a condition
2582 supertype to a hierarchy of note types, which will then be
2583 handleable in a similar fashion. However, at the moment, no such
2584 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2585 but it's an implementation detail, not a classification for the
2587 * Changes in type checking closed the following bugs:
2588 ** type checking of unused values (192b, 194d, 203);
2589 ** template selection based on unsafe type assertions (192c, 236);
2590 ** type checking in branches (194bc).
2591 * A short form of VALUES type specifier has ANSI meaning (it has
2592 increased the number of situations when SBCL cannot perform type
2594 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2595 keywords or constants is permissible.
2596 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2597 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2598 classes. (thanks to Antonio Martinez)
2599 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2600 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2601 argument to a pretty-print stream. Instead, it leaves any such
2602 operations to DESCRIBE-OBJECT methods.
2603 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2605 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2606 lambda lists. (thanks to David Lichteblau)
2607 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2608 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2609 (thanks to Gerd Moellmann)
2610 * fixed bug 246: increased compilation speed of long
2611 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2613 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2615 * DEFCONSTANT now throws a condition of type
2616 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2617 constant to a non-EQL value; CONTINUE and ABORT restarts
2618 respectively change and preserve the value.
2619 * fixed bug 63: The code walker, part of the implementation of CLOS,
2620 is now better at handling symbol macros.
2621 * bug fix: There is no longer an internal implementation type named
2622 CL:LENGTH. (reported by Raymond Toy)
2623 * bug fix: In macro-like defining macros/special operators the
2624 implicit block does not enclose the lambda list.
2625 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2626 enclosed in parentheses) are not suitable as type specifiers, and
2627 their use properly signals an error now.
2628 * bug fix: An argument count mismatch for a type specifier in code
2629 being compiled no longer causes an unhandled error at compile
2630 time, but signals a compile-time warning.
2631 * fixed simple vector readable printing
2632 * bug fix: DESCRIBE takes more care over whether the class
2633 precedence list slot of a class is bound before accessing it.
2634 (reported by Markus Krummenacker)
2635 * bug fix: FORMATTER can successfully compile pretty-printer format
2636 strings which use variants of the ~* directive inside.
2637 * bug fix: SEARCH now applies its TEST predicate to the elements of
2638 the arguments in the correct order. (thanks to Wolfhard Buss)
2639 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2640 in type inference. (thanks to Robert E. Brown)
2641 * bug fix: Reading in symbols with an explicit package name of ""
2642 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2643 * fixed some bugs revealed by Paul Dietz' test suite:
2644 ** NIL is now allowed as a structure slot name.
2645 ** Arbitrary numbers, not just REALs, are allowed in certain
2646 circumstances in LOOP for-as-arithmetic clauses.
2647 ** Multiple class redefinitions before slot access no longer
2648 causes a type error.
2649 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2650 association between the name and a class.
2651 ** Generic functions with non-standard method-combination and over
2652 five methods all of which return constants no longer return NIL
2653 after the first few invocations. (thanks to Gerd Moellmann)
2654 ** CALL-NEXT-METHOD with no arguments now passes the original
2655 values of the arguments, even in the presence of assignment.
2656 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2658 ** DISASSEMBLE works with closures and funcallable instances.
2659 ** ADD-METHOD now returns the generic function, not the new method.
2660 ** FIND-METHOD signals an error if the lengths of the specializers
2661 is incompatible with the generic function, even if the ERRORP
2663 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2664 which its argument is a member.
2665 ** DEFCLASS only redefines the class named by its class-name
2666 argument if that name is the proper name of the class;
2667 otherwise, it creates a new class.
2668 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2669 of the UNBOUND-SLOT condition to the name of the slot.
2670 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2671 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2672 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2673 SLOT-MAKUNBOUND in the specified fashion.
2675 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2676 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2677 cross-compilation host. As a consequence, we can now bootstrap our
2678 way up to SBCL starting with a bare gcc toolchain and human-readable
2679 source code (first the source to CLISP, then the source to SBCL).
2680 * A contributed module containing a partial implementation of the
2681 simple-streams interface has been included. (thanks to Rudi
2683 * A contributed module implementing the RFC1321 Message Digest
2684 Algorithm, known as MD5, has been included.
2685 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2686 no longer has any effect, as the code controlled by this feature
2687 has been deleted. (As far as we know, no-one has ever built using
2688 this feature, and its semantics were confused in any case).
2689 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2690 work on conditions (as required by the ANSI specification),
2691 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2692 expected behaviour on conditions. Users should note, however,
2693 that such behaviour is not required by the ANSI specification,
2694 and so use of this behaviour may render their code unportable.
2695 * fixed some bugs revealed by Paul Dietz' test suite:
2696 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2698 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2699 ** SXHASH on condition objects no longer returns NIL.
2700 ** :ALLOCATION :CLASS slots are better treated; their values are
2701 updated on class redefinition, and initforms inherited from
2702 superclasses are applied.
2703 ** REMOVE-METHOD returns its generic function argument even when
2704 no method was removed.
2705 ** SHARED-INITIALIZE now initializes the values of the requested
2706 slots, including those with :ALLOCATION :CLASS.
2707 ** ALLOCATE-INSTANCE now works on structure classes defined via
2708 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2710 ** SLOT-EXISTS-P now works on conditions, as well as structures
2712 ** MAKE-LOAD-FORM now has the required methods on
2713 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2714 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2715 keyword, and now implements the SLOT-NAMES argument.
2716 ** methods with &OPTIONAL arguments no longer allow too many
2717 arguments to be passed in the call without error.
2718 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2719 option is consistent with the required arguments of the generic
2720 function lambda list.
2721 * bug fix: REQUIRE accepts a string designator. (Thanks to
2723 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2724 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2725 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2726 * bug fix: The compiler now checks for duplicated variables in macro
2728 * bug fix: SETQ on globals returns the correct value.
2729 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2730 not a COMPILER-ERROR (followed by some other strange error on
2731 choosing the CONTINUE restart).
2732 * bug fix: make.sh and friends are now more consistent in the way that
2733 they look for GNU "make".
2735 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2736 * experimental native threads support (on x86 Linux >=2.4 only).
2737 This is not compiled in by default: you need to add :SB-THREAD to
2738 the target features. See the "Beyond ANSI" chapter of the manual
2740 * fix for longstanding nonANSIism: The old distinction between
2741 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2742 The return value from CL:FIND-CLASS is now a CLOS class, and
2743 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2744 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2745 * An interface to the MetaObject Protocol, as described in Kiczales,
2746 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2747 MIT Press, 1991, is available from the SB-MOP package.
2748 * incompatible change: the SB-PCL package should now be considered
2749 a private implementation detail, and no longer a semi-private MOP
2751 * minor incompatible change: due to rearrangement for threads, the
2752 control stack and binding stack are now allocated at arbitrary
2753 addresses instead of being hardcoded per-port. Users affected by
2754 this probably have to be doing advanced things with shared
2755 libraries, and will know who they are.
2756 * minor incompatible change: Previously, all --eval forms used were
2757 processed with READ before any of them were processed with EVAL.
2758 Now each --eval form is processed with both READ and EVAL before
2759 the next --eval form is processed. (Thus package operations like
2760 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2761 work as the user might reasonably expect.)
2762 * minor incompatible change: *STANDARD-INPUT* is now only an
2763 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2765 * minor incompatible change: Y-OR-N-P is now character-oriented,
2766 not line oriented. Also, YES-OR-NO-P now works without errors.
2767 (thanks to Antonio Martinez)
2768 * sb-aclrepl module improvements: an integrated inspector, added
2769 repl features, and a bug fix to :trace command.
2770 * Known functions, which cannot be open coded by the backend, are
2771 considered to be able to check types of their arguments. (fixing
2772 a bug report by Nathan J. Froyd)
2773 * fixed a bug in computing method discriminating functions: It is
2774 now possible to define methods specialized on classes which have
2775 forward-referenced superclasses. (thanks to Gerd Moellmann)
2776 * fixed evaluation order in optional entries (reported by Gilbert
2778 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2779 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2780 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2781 required-by-AMOP NAME argument, as well as CLASS and
2782 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2783 * fixed bug 20: DEFMETHOD can define methods using names that are
2784 not the proper names of classes to designate class specializers.
2785 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2786 Unix file descriptors, instead of blowing up. (thanks to Antonio
2788 * Garbage collection refactoring: user-visible change is that a
2789 call to the GC function during WITHOUT-GCING will not do garbage
2790 collection until the end of the WITHOUT-GCING. If you were doing
2791 this you were probably losing anyway.
2792 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2793 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2794 TYPEP the latter but not the former.
2795 * The compiler issues a full WARNING for calls to undefined functions
2796 with names from the CL package.
2797 * MAP-INTO for a vector destination is open coded. (reported by
2798 Brian Downing on c.l.l)
2799 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2800 documentation string.
2801 * fixed some bugs revealed by Paul Dietz' test suite:
2802 ** COPY-ALIST now signals an error if its argument is a dotted
2804 ** Condition slots are now accessed more correctly in the presence
2805 of multiple initargs for a given slot.
2806 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2807 exclude restarts of the same name associated with a different
2809 ** DEFCLASS of forward-referenced classes with another
2810 forward-referenced class in the superclasses list no longer
2812 ** Condition slots are now initialized once each, not multiple
2813 times. (thanks to Gerd Moellmann)
2814 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2816 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2817 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2818 arguments contain duplicated elements.
2819 ** RESTART-CASE understands local macros.
2820 ** RESTART-CASE associates exactly its own restarts with a condition.
2821 ** ENDP in safe mode checks its argument to be of type LIST.
2822 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2823 in question is unbound.
2824 ** Optimized MAKE-INSTANCE functions no longer cause internal
2825 assertion failures in the presence of duplicate initargs.
2826 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2827 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2828 propagated even in the ctor optimized implementation of
2830 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2832 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2833 * a better implementation of SXHASH on (simple) bit vectors,
2834 measured both in execution speed and in distribution of results
2835 over the positive fixnums, has been installed. Likewise, a better
2836 implementation of EQUAL for simple bit vectors is now available.
2837 * fixed CEILING optimization for a divisor of form 2^k.
2838 * fixed bug 240 (emitting extra style warnings "using the lexical
2839 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2840 by Antonio Martinez)
2841 * fixed SXHASH, giving different results for NIL depending on type
2842 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2843 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2844 variable. (found by Rolf Wester)
2845 * SBCL does not ignore type declarations for special
2846 variables. (reported by rif on c.l.l 2003-03-05)
2847 * some bug fixes in contrib/sb-aclrepl/
2848 * fixed some bugs revealed by Paul Dietz' test suite:
2849 ** a bug in the CONS type specifier, whereby the CAR and CDR
2850 types got intertwined, has been fixed;
2851 ** the type system is now able to reason about the interaction
2852 between INTEGER and RATIO types more completely;
2853 ** APPEND, [N]REVERSE and NRECONC check that those their
2854 arguments, which must be proper lists, are really so;
2855 ** An array specialized to be unable to hold elements has been
2856 implemented, as required -- yes, really -- by ANSI;
2857 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2858 on malformed property lists;
2860 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2861 * incompatible packaging change: in line with Unix convention,
2862 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2863 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2864 * REQUIRE and PROVIDE are now optionally capable of doing something
2865 useful. See the documentation string for REQUIRE.
2866 * infrastructure for a managed SBCL contrib system: contributed
2867 modules in this release include:
2868 ** the ASDF system definition facility;
2869 ** an interface to the BSD Sockets API;
2870 ** an ACL-like convenience interface to the repl;
2871 (thanks to Kevin Rosenberg)
2872 ** an implementation of ROTATE-BYTE, with efficient implementation
2874 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2875 gives a closer approximation to the right answer than previously.
2876 (thanks to Raymond Toy)
2877 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2878 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2879 optional environment arguments, as required by ANSI.
2880 * fixed bugs in other functions taking environment objects, allowing
2881 calls with an explicit NIL environment argument to be compiled
2883 * fixed bug 228: primary return values from
2884 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2885 COMPILE or FUNCTION.
2886 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2887 :TYPE will no longer signal errors on innocuous objects.
2888 * fixed bug 231b: SETQ is better at respecting type declarations in
2889 the lexical environment.
2890 * fixed a bug in DEFCLASS: classes named by symbols with no or
2891 unprintable packages can now be defined.
2892 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2893 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2894 * fixed compiler failure related to checking types of functions.
2895 (reported by Robert E. Brown)
2896 * the compiler is now much more consistent in its error-checking
2897 treatment of bounding index arguments to sequence functions: in
2898 (SAFETY 3) code, errors will be signalled in almost all cases if
2899 invalid sequence bounding indices are passed to functions defined
2900 by ANSI to operate on sequences.
2901 * fixed a bug in the build procedure: documentation of SBCL-specific
2902 packages is now preserved and available in the final Lisp image.
2903 * lifted FDEFINITION lookup out of loops in the implementation of
2904 many list operations. (thanks to Robert E. Brown)
2905 * fixed a bug in the reader: the #n# reader macro now works for
2906 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2907 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2908 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2909 E. Brown and Tony Martinez respectively)
2910 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2911 SLOT-BOUNDP: in particular, it is now able to optimize them much
2912 better, and is now not vulnerable to having packages renamed.
2913 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2914 always called when a slot is not present in an instance. (thanks
2916 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2917 longer signal an error on unreading a character following EOF on
2918 the previous constituent stream. (thanks to Tony Martinez)
2919 * fixed some bugs revealed by Paul Dietz' test suite:
2920 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2921 not just nonnegative fixnums;
2922 ** the logical bit-array operators such as BIT-AND now accept an
2923 explicit NIL for their "opt-arg" argument (to indicate a
2924 freshly-consed result bit-array);
2925 ** ELT now signals an error on an invalid sequence index in safe
2927 ** the type system is now cleverer about negations of numeric
2928 types, and consequently understands the BIGNUM and RATIO types
2930 ** the type system is now cleverer about the interaction between
2931 INTEGER and RATIO types: while bugs still remain, many more
2932 cases are accurately computed;
2933 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2934 if it is in the last clause;
2935 ** CONSTANTLY now correctly returns a side-effect-free function in
2937 ** DECLARE is no longer treated as a special-operator; in
2938 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2939 * incremented fasl file version number due to the change in the
2940 DEFSTRUCT-SLOT-DESCRIPTION structure.
2942 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2943 * minor incompatible change: code processed by the "interpreter" or
2944 EVAL now has a compilation optimization policy of (DEBUG 2)
2945 (changed from (DEBUG 1)) to improve debuggability of interactive
2946 development, and to allow the use of the debug RETURN command in
2948 * an experimental implementation of the RETURN command for the
2949 debugger has been included. (thanks to Frederik Kuivinen)
2950 * fixed bug 62: constraints were not propagated into a loop.
2951 * fixed bug in embedded calls of SORT (reported and investigated by
2953 * fixed some bugs revealed by Paul Dietz' test suite:
2954 ** printing and reading of arrays with some dimensions having
2955 length 0 (thanks to Gerd Moellmann);
2956 ** BOA constructor with &AUX argument without a default value does
2957 not cause a type error;
2958 ** CONSTANTP now returns true for all self-evaluating objects.
2960 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2961 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2962 accessors that are related by inheritance, as specified in the
2963 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2964 Valtteri Vuorikoski)
2965 * The compiler is now able to inline functions that were defined in
2966 a complex lexical environment (e.g. inside a MACROLET).
2967 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2969 * Support for the upcoming FreeBSD-5.0 release has been included.
2970 (thanks to Dag-Erling Smorgrav)
2971 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2972 effect when it is not in a toplevel context.
2973 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2974 stand a better chance of being correct. (thanks to Gerd
2976 * fixed bug in COERCE, which now signals an error on coercing a
2977 rational to a bounded real type which excludes the expected
2979 * The compiler is now able to derive types more accurately from the
2980 COERCE and COMPILE functions.
2981 * fixed bug 223: functional binding is considered to be constant
2982 only for symbols in the CL package.
2983 * fixed bug 231: SETQ did not check the type of a variable being set
2984 (reported by Robert E. Brown)
2985 * A new optimization for MAKE-INSTANCE has been included, fixing
2986 various bugs (including relating to :ALLOCATION :CLASS slots and
2987 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2989 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2990 ** As required by ANSI, LOOP now disallows anonymous collection
2991 clauses such as COLLECT I in conjunction with aggregate boolean
2992 clauses such as THEREIS (= I 1);
2993 ** LOOP now signals an error when any variable is reused in the
2994 same loop (including the potentially useful construct analogous
2995 to WITH A = 1 WITH A = (1+ A);
2996 ** IT is only a special loop symbol within the first clause of a
2997 conditional loop clause;
2998 ** LOOP with a typed iteration variable over a hashtable now
2999 signals a type error iff it should.
3000 * fixed some other bugs revealed by Paul Dietz' test suite:
3001 ** FILE-STREAM now names the class previously known as FD-STREAM;
3002 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3003 argument) no longer signals an error;
3004 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3005 of the slot symbol, rather than using the current package
3006 ((:CONC-NAME "") continues to intern the slot's name in the
3008 * incremented fasl file version number, because of the incompatible
3009 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3010 of the new implementation of DEFINE-COMPILER-MACRO.
3012 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3013 * Support for building SBCL for MIPS platforms running in
3014 little-endian mode has now been checked in, and basic
3015 functionality on said platforms verified.
3016 * minor incompatible change: PCL now records the pathname of a file
3017 in which methods and the like are defined, rather than its
3019 * minor incompatible change: TRUENAME now considers the truename of
3020 a file naming a directory to be the pathname with :DIRECTORY
3021 component indicating that directory.
3022 * minor incompatible change: a NAMED clause in the extended form of
3023 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3024 reason for the previous behaviour is unclear.
3025 * more systematization and improvement of CLOS and MOP conformance
3026 in PCL (thanks to Gerd Moellman and Pierre Mai):
3027 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3029 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3030 primary methods with no specializers;
3031 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3033 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3034 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3035 FORWARD-REFERENCED-CLASSes; error reporting on
3036 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3038 ** SXHASH on CLOS instances now uses a slot internal to the
3039 instance to return different numbers on distinct instances,
3040 while preserving the same return value through invocations of
3042 ** DEFMETHOD signals errors when methods with longer incongruent
3043 lambda lists are added to generic functions;
3044 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3045 CLASS, as specified in AMOP;
3046 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3047 on the order returned by the primary method for classes of
3048 class STANDARD-CLASS;
3049 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3050 * fixed some bugs shown by Paul Dietz' test suite:
3051 ** DOLIST puts its body in TAGBODY;
3052 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3054 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3055 value producing form;
3056 ** if more variables are given to PROGV than values, extra
3057 variables are bound and made to have no value;
3058 ** NSUBSTITUTE on list arguments gets the right answer with
3060 ** ELT signals an error of type TYPE-ERROR when the index argument
3061 is not a valid sequence index;
3062 ** LOOP signals (at macroexpansion time) an error of type
3063 PROGRAM-ERROR when duplicate variable names are found;
3064 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3065 ** FUNCALL of special-operators now cause an error of type
3067 ** PSETQ now works as required in the presence of side-effecting
3068 symbol-macro places;
3069 ** NCONC accepts any object as its last argument;
3070 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3072 ** loop-for-as-package does not require a package to be explicitely
3074 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3076 * fixed bug 166: compiler preserves "there is a way to go"
3077 invariant when deleting code.
3078 * fixed bug 172: macro lambda lists with required arguments after
3079 &REST arguments now cause an error to be signalled. (thanks to
3081 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3082 bug. (thanks to Matthew Danish)
3083 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3085 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3086 arguments are lexically rebound. (thanks to Gerd Moellmann and
3088 * fixed bug 194: error messages are now more informative when there
3089 is no primary method applicable in a call to a generic
3090 function. (thanks to Gerd Moellmann)
3091 * fixed bug in command line argument checking (thanks to Julian
3093 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3094 (thanks to Matthew Danish)
3095 * incremented fasl file version number, because of the
3096 SXHASH-related changes in the layout of CLOS data structures
3098 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3099 * minor incompatible change: The runtime (the Unix executable named
3100 "sbcl") is now much pickier about the .core files it will load.
3101 Essentially it now requires .core files to descend from the same
3102 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3103 as the runtime does. (The intent is to prevent the crashes which
3104 can occur, and which can even be reported as mysterious failures,
3105 when people patch the sources or change the build parameters
3106 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3107 sbcl and .core files.)
3108 * fixed bug: VALUES-LIST is no longer optimized away.
3109 * fixed bug 142: The FFI conversion of C string values to Lisp
3110 string values no longer conses excessively. (thanks to Nathan
3111 Froyd porting Raymond Toy's fix to CMU CL)
3112 * began to systematize and improve MOP conformance in PCL (thanks to
3113 Nathan Froyd, Gerd Moellman and Pierre Mai):
3114 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3116 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3117 ** FINALIZE-INHERITANCE is now called on class finalization;
3118 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3119 argument precedence order.
3120 * fixed bug 202: The compiler no longer fails on functions whose
3121 derived types contradict their declared type.
3122 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3123 so it can be non-toplevel.
3124 * The fasl file version number has changed (because of the new
3125 implementation of DEFMACRO).
3126 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3127 safe code, that any length requirement by their type-specifier
3128 argument is valid. The exceptions to this are described in bug
3130 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3131 functions now check the types of their inputs as required by ANSI.
3132 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3133 introduced symbol is DECLAREd to be SPECIAL.
3134 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3135 * fixed bug: :COUNT argument to sequence functions may be negative
3136 * fixed bug: body of DO-SYMBOLS may contain declarations
3137 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3138 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3139 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3140 symbol macro only once
3141 * fixed printing of call frame when argument list is unavailable
3142 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3143 * compiler no longer signals WARNING on unknown keyword
3146 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3147 * A beta-quality port to the mips architecture running Linux,
3148 based on the old CMUCL backend, has been made. It has been tested
3149 on a big-endian kernel, and works sufficiently well to be able to
3150 rebuild itself; it has not been tested in little-endian mode.
3151 * fixed an inconsistency between gencgc.c and purify.c which made
3152 dumping/loading .core files unreliable
3153 * fixed bug 120a: The compiler now deals correctly with IFs where
3154 the consequent is the same as the alternative, instead of
3155 misderiving the return type. (thanks to Alexey Dejneka)
3156 * fixed bug 113: Logical pathnames are now dumpable (the logical
3157 host is resolved at load-time, throwing an error if it is not
3159 * fixed bug 174: FORMAT's error message is slightly clearer when a
3160 non-printing character is used in a format directive.
3161 * fixed several bugs in compiler checking of type declarations, i.e.
3162 violations of the Python "declarations are assertions" principle
3163 (thanks to Alexey Dejneka)
3164 * fixed several bugs in PCL's error checking (thanks to Gerd
3166 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3168 * fixed bug in compilation of functions as first class values
3169 (thanks to Antonio Martinez-Shotton)
3170 * The compiler's handling TYPE-ERRORs which it can prove will
3171 inevitably happen at runtime has been cleaned up and corrected
3172 in several ways. (thanks to Alexey Dejneka)
3173 * improved argument type checking for various basic arithmetic
3174 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3175 lavished on them in the past that they can be compiled in many
3176 ways in different special cases
3177 * fixed bug 181: compiler checks validity of user supplied type
3179 * cleaned up code flushing in optimization: Function calls which
3180 should signal errors for safety purposes (e.g. which ANSI says
3181 should signal errors when their arguments are of incorrect type)
3182 are no longer optimized away.
3183 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3184 * incremented fasl file version number, because changes in the
3185 implementation of sequence functions like COERCE caused
3186 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3187 in old inline expansions) to become undefined. (Actually these
3188 changes were later undone, so we might very well be binary
3189 compatible with 0.7.7 after all, but leaving the version number
3190 incremented seemed like the simplest and most conservative
3193 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3194 * An alpha-quality port to the parisc architecture running Linux,
3195 based on the old CMUCL backend, has been made. This, even more so
3196 than the other backends, should be considered still a work in
3197 progress; known problems include that the Linux kernel in 64-bit
3198 mode does not propagate the correct sigcontext structure to
3199 userspace, and consequently SBCL on a parisc64 kernel will not
3201 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3202 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3203 suppressed.) Also now that the compiler is looking at declarations
3204 in the environment, it checks optimization declarations as well,
3205 and suppresses inlining when (> DEBUG SPEED).
3206 * More fixes have been made to treatment of floating point exception
3207 treatment and other Unix signals. In particular, floating point
3208 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3209 * The detection and handling of control stack exhaustion (infinite
3210 or very deeply nested recursion) has changed. Stack exhaustion
3211 detection is now done by write-protecting pages at the OS level
3212 and applies at all optimization settings; when found, a
3213 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3214 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3215 be caught using IGNORE-ERRORS.
3216 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3217 correctly when transforming recursive LABELS functions to LETs.
3218 (thanks to Alexey Dejneka)
3219 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3220 that are names of constants or global variables.
3221 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3222 alien routines with docstrings.
3223 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3224 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3226 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3227 (thanks to Eric Marsden)
3228 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3229 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3230 * Bug fix: The SPARC backend can now compile functions involving
3231 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3232 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3233 to a functions expecting a PATHNAME-DESIGNATOR.
3234 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3235 lambda lists. (thanks to Alexey Dejneka)
3236 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3237 functions so that EXPORTing the name of the class doesn't cause
3238 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3239 to Antonio Martinez for reporting this)
3240 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3241 constant-folded without failing an assertion. (thanks to Einar
3242 Floystad Dorum for reporting this)
3243 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3244 types are now dealt with more correctly by the compiler.
3245 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3246 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3247 behaviour analogous to RENAME-FILE. This puts its behaviour more
3248 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3250 * The fasl file version number has changed again. (because of the
3251 bug fix involving the names of PCL MAKE-INSTANCE functions)
3253 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3254 * bug fix: Floating point exceptions are treated much more
3255 consistently on the x86/Linux and PPC/Linux platforms.
3256 * Array initialization with :INITIAL-ELEMENT is now much faster for
3257 cases when the compiler cannot open code the array creation, but
3258 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3259 array accesses have also seen a speed increase.
3260 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3261 specified. (This was at the root of some bad interactions between
3262 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3264 * bug fix: Internal error arguments for undefined functions are now
3265 computed correctly on the PPC/Linux platform.
3266 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3267 Raymond Toy's patch for CMU CL)
3268 * Support for the Solaris 9 operating environment has been included
3269 (thanks to Daniel Merritt)
3270 * A very ugly but hopefully complete draft of the missing FFI chapter
3271 of the manual has been created by reformatting the corresponding
3272 CMU CL manual chapter into (currently very ugly and incoherent)
3273 DocBook and bringing it up to date for SBCL behavior. Thus, the
3274 manual is now essentially complete, at least by my extreme
3275 once-and-only-once standards, whereby it's acceptable to refer to
3276 the doc strings of SB-EXT functions as the primary documentation.
3277 * The fasl file version number has changed again, due to cleanup of
3278 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3279 is no longer a static symbol.)
3281 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3282 * SBCL now builds with OpenMCL (version 0.12) as the
3283 cross-compilation host; also, more progress has been made toward
3284 bootstrapping under CLISP.
3285 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3287 * bug 158 fixed: The compiler can now deal with integer loop
3288 increments different from 1; fixing this turned out also to fix
3290 * bug 169 fixed: no more bogus warnings about using lexical bindings
3291 despite the presence of perfectly good SPECIAL declarations (thanks
3292 to David Lichteblau)
3293 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3294 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3295 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3297 * bug 180 fixed: Method combination specifications no longer ignore
3298 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3299 * bug fix: Structure type predicate functions now check their argument
3300 count as they should.
3301 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3302 correctly. (thanks to Pierre Mai)
3303 * minor incompatible change: The --noprogrammer option is deprecated
3304 in favor of the new --disable-debugger option, which is very similar.
3305 (The major difference is that it takes effect at a slightly different
3306 time at startup, causing handling of errors in --sysinit and
3307 --userinit files will be affected differently.) The
3308 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3309 been added to allow this functionality to be controlled from ordinary
3310 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3311 maintainers, who might want to run non-interactive scripts to
3312 build SBCL cores which will later be used interactively.)
3313 * minor incompatible change: The LOAD function no longer, when given
3314 a wild pathname to load, loads all files matching that pathname.
3315 Instead, an error of type FILE-ERROR is signalled.
3317 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3318 * bug 147 fixed: The compiler preserves its block link/count
3319 invariants more correctly now so that it doesn't crash. (thanks
3321 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3323 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3324 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3325 no-value-for-this-character as NIL (as specified by ANSI).
3326 * HOST-NAMESTRING on physical pathnames now returns a string that is
3327 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3328 (thanks to Christophe Rhodes)
3329 * The Alpha port handles icache flushing more correctly. (thanks to
3331 * More progress has been made toward bootstrapping under CLISP. (thanks
3332 to Christophe Rhodes)
3333 * The fasl file format has changed again, because dynamic loading
3334 on OpenBSD (which has non-ELF object files) motivated some cleanups
3335 in the way that foreign symbols are transformed and passed around.
3336 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3337 is no longer treated as whitespace by the reader, but instead as
3338 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3339 |A
\7fB|, instead of A as it used to.
3341 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3342 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3343 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3344 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3345 this as of 0.7.1.45, but was left out of the previous news section
3346 (thanks to Dan Barlow)
3347 * SBCL now runs on the Solaris operating system on SPARC architectures
3348 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3349 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3350 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3351 * SPARC backend cleanups, allowing builds of cores optimized for V8
3352 and V9 SPARCS, and also emission of code targeted to a particular
3353 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3355 * SBCL is closer to bootstrapping under CLISP, thanks to various
3356 fixes by Christophe Rhodes.
3357 * The fasl file format has changed again, to allow the compiler's
3358 INFO database to support symbol macros.
3359 * The user manual (in doc/) is formatted into HTML more nicely.
3360 (thanks to coreythomas)
3361 * The system is smarter about SUBTYPEP relationships, especially
3362 those involving NOT types (including types such as ATOM which are
3363 represented internally using NOT types). Thus SUBTYPEP is less
3364 likely to return (VALUES NIL NIL) in general, and in particular
3365 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3367 * The fasl file format has changed again, because the internal
3368 representation of types now includes a new slot to support the new
3369 SUBTYPEP-of-NOT-types logic.
3370 * (not a change in the main branch of SBCL, but a related prototype
3371 which can hopefully be merged into the main branch of SBCL in the
3372 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3373 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3374 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3375 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3376 dispatch macro character combinations. (thanks to Alexey Dejneka)
3377 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3378 that superficially look like logical namestrings correctly.
3379 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3381 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3382 default supplied-p) for &optional and &key arguments. (thanks to
3385 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3386 * incompatible change: The compiler is now less aggressive about
3387 tail call optimization, doing it only when (> SPACE DEBUG) or
3388 (> SPEED DEBUG). (This is an incompatible change because there are
3389 programs which relied on the old CMU-CL-style behavior to optimize
3390 away their unbounded recursion which will now die of stack overflow.)
3391 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3392 for non-GENCGC systems has been increased to 20M (since that
3393 seems much closer to the likely performance optimum for modern
3394 systems than the old 4M value was)
3395 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3396 and *DEBUG-PRINT-LEVEL*
3397 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3398 of CMU CL's support for SPARC, and various endianness and other
3399 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3400 * new syntactic sugar for the Unix command line: --load foo.bar is now
3401 an alternate notation for --eval '(load "foo.bar")'.
3403 ** The system now detects stack overflow and handles it gracefully,
3404 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3405 optimization settings. (This is a good thing in general, and
3406 its introduction in this version should be particularly timely
3407 for anyone whose code fails because of suppression of tail
3409 ** The system now hunts for the C variable "environ" in a more
3410 devious way, to avoid segfaults when the C library version differs
3411 between compile time and run time. (thanks to Christophe Rhodes)
3412 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3413 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3414 ** The compiler no longer issues bogus style warnings for undefined
3415 classes in the same source file as the DEFCLASSes which defined
3416 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3418 ** fixes in CONDITION class precedence list for undefined function
3419 errors (thanks to Alexei Dejneka)
3420 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3421 correctly. (thanks to Dan Barlow)
3422 ** portability fixes aiming at bootstrapping under CLISP (thanks
3423 to Dave McDonald and Christophe Rhodes)
3424 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3425 ** fixes in type translation and and type inference (thanks to
3427 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3428 ** various fixes in the new ports (thanks to Dan Barlow)
3429 * several changes related to debugging:
3430 ** suppression of tail recursion, as noted above
3431 ** stack overflow detection, as noted above
3432 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3433 is now the default. (For some time encapsulation has been more
3434 reliable than the breakpoint-based :ENCAPSULATE NIL
3435 implementation, at least on X86 systems; and I just noticed that
3436 encapsulation also seems closer to the spirit of the ANSI
3439 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3441 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3442 up properly again. (There was a packaging bug in 0.7.0 which
3443 left their definitions in SB-SYS::LOAD-FOREIGN and
3444 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3445 vital for most things which interface to C-level interfaces,
3446 like extensions working with sockets or databases or
3447 Perl-compatible regexes or whatever, and the need to fix
3448 this bug is the main reason that 0.7.1 was released so
3450 ** DEFGENERIC is now choosier about the methods it redefines, so that
3451 reLOADing a previously-LOADed file containing DEFGENERICs does
3452 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3453 cycle now works as it should. (thanks to Alexey Dejneka)
3454 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3455 fixed by Christophe Rhodes. (He actually submitted this patch
3456 months ago, and I delayed until after 0.7.0.)
3457 ** Bug 111 (internal compiler confusion about runtime checks on
3458 FUNCTION types) was fixed by Alexey Dejneka.
3459 * Some internal cleanups (getting rid of variables which aren't
3460 needed now that the byte interpreter is gone) caused the fasl
3461 file format number to change again.
3463 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3464 * major incompatible change: The default fasl file extension, i.e. the
3465 default extension for files produced by COMPILE-FILE, has changed
3466 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3468 ** There are many changes in the implementation of the compiler.
3469 SBCL is now essentially a compiler-only implementation of ANSI
3470 Common Lisp. EVAL still "interprets" a few special cases, but
3471 almost all the interesting cases are handled by creating
3472 a LAMBDA expression, calling COMPILE on it, then calling
3473 FUNCALL on the result.
3474 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3475 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3476 Since the code is newer, there might still be some new bugs
3477 (though not as many as before Martin Atzmueller's fixes:-). But
3478 the new code is substantially simpler and clearer, and hopefully
3479 any remaining bugs will be simpler, less fundamental, and more
3480 fixable then the bugs in the old code.
3481 ** The revised compiler is still a little unsteady on its feet.
3483 *** The debugging information it produces (particularly the names
3484 of FUNCTION objects) is sometimes much less useful than what
3485 the old compiler produced.
3486 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3487 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3488 within a MACROLET) has been temporarily weakened.
3489 ** There are new compiler optimizations for various functions:
3490 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3491 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3492 *** the math functions TRUNCATE, FLOOR, and CEILING
3493 *** the function-of-all-trades COERCE
3494 Mostly these should be transparent, but there's one
3495 potentially-annoying problem (bug 117): when the compiler
3496 inline-expands a function and does type analysis on the result,
3497 it can create control paths which have type mismatches, and
3498 when it can't prove that those control paths aren't taken,
3499 it will issue WARNINGs about the type mismatches. This is
3500 a particular problem in practice for the new sequence functions.
3501 It's not clear how this should be fixed, and for now, a
3502 workaround is given in the entry for 117 in the BUGS file.
3503 ** (Because of the interaction between the two previous items --
3504 occasional inlining problems and new inline expansions -- some
3505 of the new sequence function optimizations won't really kick in
3506 completely until debugging information, and then inlining, are
3507 straightened out in some future version.)
3508 * minor incompatible changes:
3509 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3510 DIRECTORY no longer implicitly promotes NIL slots of its
3511 pathname argument to :WILD. In particular, when you ask for the
3512 contents of a directory (which you used to be able to do without
3513 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3514 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3515 ** changes in behavior that ANSI explicitly defines to be
3516 implementation dependent:
3517 *** The new compiler-only implementation still conforms with ANSI,
3518 but acts a little different than before. Besides the obvious
3519 changes in performance tradeoffs (that the cost per form passed
3520 to EVAL has gone up, and the cost per form executed by EVAL
3521 has gone down), the behavior of the system changes a little
3522 because there are no longer any interpreted function objects.
3523 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3524 e.g. doing COMPILE on the output of interactive DEFUN is
3526 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3528 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3529 has changed from (:CL) to NIL.
3530 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3531 e.g. #\Newline and #\Tab, don't have names specified in the
3532 ANSI Common Lisp standard, is now based on their ASCII symbolic
3533 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3534 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3535 are no longer used for output.
3536 ** changes in internal implementation constants:
3537 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3538 4 million. (If your application spends a lot of time GCing and
3539 you have a lot of RAM, you might want to experiment with
3540 increasing it even more.)
3541 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3542 However, almost all old code should still continue to work without
3543 immediate update, as SB-C-CALL is now a (deprecated) nickname
3545 ** Old operator names in the style DEF-FOO are now deprecated in
3546 favor of new corresponding names DEFINE-FOO, for consistency with
3547 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3548 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3549 internal symbols, but a few supported extensions like
3550 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3551 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3552 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3553 etc. as you get deeper into recursive calls to the debugger
3554 command loop, instead of the old "5]", "5]]", "5]]]"
3555 sequence. (I was motivated to do this when squabbles between
3556 ILISP and SBCL left me very deeply nested in the debugger. In the
3557 short term, this change will probably provoke more ILISP/SBCL
3558 squabbles, but hopefully it will be an improvement in the long run.)
3559 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3560 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3561 function, since declaiming return types involving aliens is
3562 (1) annoyingly messy to do by hand and (2) vital to efficient
3563 compilation of code which calls such functions.
3564 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3565 longer reexported by the SB-EXT package. They're solely useful
3566 for alien code, so it seems more logical that you should get
3567 them from the SB-ALIEN package, not in SB-EXT.
3568 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3569 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3570 features. Instead, the code that they used to control is always
3571 built into the system.
3572 * many other bug fixes
3573 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3574 advantage of the new EVAL-WHEN stuff and to clean them up in
3575 general, and they are now more ANSI-compliant in a number of
3576 ways. Martin Atzmueller is responsible for a lot of this.
3577 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3579 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3580 *** correct ERROR type for various file operations
3581 *** some fixes for Lisp streams
3582 *** DEFMETHOD syntax checking
3583 *** changing old weird representation of debug information as
3584 strings (which, among their other deficiencies, don't transform
3585 correctly when you rename packages, and don't change their
3586 print representation when you change things like *PACKAGE*
3587 and *PRINT-LENGTH*) to symbols and lists of symbols
3588 He also made several improvements and fixed several bugs in DESCRIBE.
3589 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3591 *** misbehavior of WRITE-STRING/WRITE-LINE
3592 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3593 and several other LOOP problems as well
3594 *** DIRECTORY when similar filenames are present
3595 *** DEFGENERIC with :METHOD options
3596 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3597 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3598 *** argument ordering in FIND with :TEST option
3599 *** mishandled package designator argument in APROPOS-LIST
3600 *** various problems in the backquote readmacro
3601 *** a bug in APROPOS
3602 *** probably some others that I'm not describing very well here,
3603 since the CVS log documents them by reference to sbcl-devel
3604 messages, and the SourceForge archives aren't working well.:-(
3605 ** Dan Barlow improved the Alpha port (and is making progress on the
3606 PPC port, for those of you who think different).
3607 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3608 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3609 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3610 continues to work on the SPARC port (for those of you in a position
3611 to look down upon our little PC-compatible boxes from a great height).
3612 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3613 and uses that copy, rather than the current dynamic value, when
3614 it's trying to decide whether to truncate output. Thus e.g.
3615 (let ((*print-lines* 50))
3616 (pprint-logical-block (stream nil)
3618 (let ((*print-lines* 8))
3619 (print (aref possiblybigthings i) stream)))))
3620 should now truncate the logical block only at 50 lines, instead of
3621 often truncating it at 8 lines, as it did before.
3622 * The doc/cmucl/ directory, containing old CMU CL documentation
3623 from the time of the fork, is no longer part of the base system.
3624 SourceForge has shut down its anonymous FTP service, and with it
3625 my original plan for distributing the old CMU CL documentation
3626 there. For now, if you need these files you can download an old
3627 SBCL source release and extract them from it.
3628 * The fasl file version number changed again, for dozens of reasons,
3629 some of which are apparent above.
3631 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3632 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3633 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3634 patch, so that DISASSEMBLE output is much nicer.
3635 * The code in the SB-PROFILE package now seems reasonably stable.
3636 I still haven't decided what the final interface should look like
3637 (I'd like PROFILE to interact cleanly with TRACE, since both
3638 facilities use function encapsulation) but if you have a need
3639 for profiling now, you can probably use it successfully with
3640 the current CMU-CL-style interface.
3641 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3642 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3643 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3644 link now returns the dangling link itself, and for similar
3645 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3646 link itself. (In these cases the old code signalled an error and
3647 looped endlessly, respectively.) Thus, DIRECTORY now works even
3648 in the presence of dangling and cyclic symbolic links.
3649 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3650 is now a supported extension again, since the consensus on
3651 sbcl-devel was that it can be useful for ordinary development
3652 work, not just for debugging SBCL itself.
3653 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3654 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3655 that currently-defined functions might be redefined later with
3656 different return types.
3657 * Hash tables can be printed readably, as inspired by CMU CL code
3658 of Eric Marsden and SBCL code of Martin Atzmueller.
3659 * better error handling in CLOS method combination, thanks to
3660 Martin Atzmueller porting Pierre Mai's CMU CL patches
3661 * more overflow fixes for >16Mbyte I/O buffers
3662 * A bug in READ has been fixed, so that now a single Ctrl-D
3663 character suffices to cause end-of-file on character streams.
3664 In particular, now you only need one Ctrl-D at the command
3665 line (not two) to exit SBCL.
3666 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3668 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3669 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3670 does the right thing.
3671 * The implementation of some type tests, especially for CONDITION
3672 types, is now tidier and maybe faster, due to CMU CL code
3673 originally by Douglas Crosher, ported by Martin Atzmueller.
3674 * Some math functions have been fixed, and there are new
3675 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3676 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3677 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3678 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3679 likely to be built into sbcl-0.7.0.)
3680 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3681 CL, ported to SBCL by Martin Atzmueller)
3682 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3683 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3684 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3685 * There's a new slam.sh hack to shorten the edit/compile/debug
3686 cycle for low-level changes to SBCL itself, and a new
3687 :SB-AFTER-XC-CORE target feature to control the generation of
3688 the after-xc.core file needed by slam.sh.
3689 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3690 COMPILE-FILE is no longer supported, so that now every function
3691 gets an entry point, so that block compilation looks a little
3692 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3693 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3694 returns the number of bytes consed since the system started,
3695 rather than the number consed since the first time the function
3696 was called. (The new definition parallels ANSI functions like
3697 CL:GET-INTERNAL-RUN-TIME.)
3698 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3699 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3700 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3701 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3702 * Fasl file version numbers are now independent of the target CPU,
3703 since historically most system changes which required version
3704 number changes have affected all CPUs equally. Similarly,
3705 the byte fasl file version is now equal to the ordinary
3708 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3709 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3710 is no longer recognized. I apologize for this, because it was
3711 listed in SB-EXT as a supported extension, but I found that
3712 its existing behavior was poorly specified, as well as incorrectly
3713 specified, and it looked like too much of a mess to straighten it
3714 out. I have enough on my hands trying to get ANSI stuff to work..
3715 * many patches ported from CMU CL by Martin Atzmueller, with
3716 half a dozen bug fixes in pretty-printing and the debugger, and
3717 half a dozen others elsewhere
3718 * fixed bug 13: Floating point infinities are now supported again.
3719 They might still be a little bit flaky, but thanks to bug reports
3720 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3721 as flaky as they were.
3722 * The --noprogrammer command line option is now supported. (Its
3723 behavior is slightly different in detail from what the old man
3724 page claimed it would do, but it's still appropriate under the
3725 same circumstances that the man page talks about.)
3726 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3727 are now supported, and enabled by default. Thus, the compiler can
3728 handle many floating point and complex operations much less
3729 inefficiently. (Thus e.g. you can implement a complex FFT
3731 * The compiler now detects type mismatches between DECLAIM FTYPE
3732 and DEFUN better, and implements CHECK-TYPE more correctly, and
3733 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3734 thanks to patches from Martin Atzmueller.
3735 * various fixes to make the cross-compiler more portable to
3736 ANSI-conforming-but-different cross-compilation hosts (notably
3737 Lispworks for Windows, following bug reports from Arthur Lemmens)
3738 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3739 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3740 fixed thanks to Pierre Mai's CMU CL patches.
3741 * fixes to keep the system from overflowing internal counters when
3742 it tries to use i/o buffers larger than 16M bytes
3743 * fixed bug 45a: Various internal functions required to support
3744 complex special functions have been merged from CMU CL sources.
3745 (When I was first setting up SBCL, I misunderstood a compile-time
3746 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3747 * improved support for type intersection and union, fixing bug 12
3748 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3749 more obscure bugs as well
3750 * some steps toward byte-compiling non-performance-critical
3751 parts of the system, courtesy of patches from Martin Atzmueller
3752 * Christophe Rhodes has made some debian packages of sbcl at
3753 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3754 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3755 stable, but are nonetheless usable. When he's ready, I'd be happy
3756 to add them to the SourceForge "File Releases" section. (And if
3757 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3758 * new fasl file format version number (because of changes in
3759 internal representation of (OR ..) types to accommodate the new
3760 support for (AND ..) types, among other things)
3762 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3763 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3765 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3767 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3768 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3769 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3770 * DESCRIBE now works on structure objects again.
3771 * Most function call argument type mismatches are now handled as
3772 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3773 know whether the function will be redefined before the call is
3774 executed. (The compiler could flag local calls with full WARNINGs,
3775 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3776 it doesn't keep track of enough information to know whether calls
3777 are local in this sense.)
3778 * Compiler output is now more verbose, with messages truncated
3779 later than before. (There should be some supported way for users
3780 to override the default verbosity, but I haven't decided how to
3781 provide it yet, so this behavior is still controlled by the internal
3782 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3783 src/compiler/ir1util.lisp.)
3784 * Fasl file format version numbers have increased again, because
3785 support for the Gray streams extension changes the layout of the
3786 system's STREAM objects.
3787 * The Gray subclassable streams extension now works, thanks to a
3788 patch from Martin Atzmueller.
3789 * The full LOAD-FOREIGN extension (not just the primitive
3790 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3791 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3792 but like most other programs, it defaults to copying the Unix
3793 environment from the original process instead of starting the
3794 new process in an empty environment.
3795 * Extensions which manipulate the Unix environment now support
3796 an :ENVIRONMENT keyword option which doesn't smash case or
3797 do other bad things. The CMU-CL-style :ENV option is retained
3798 for porting convenience.
3799 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3800 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3802 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3804 * A patch from Martin Atzmueller seems to have solved the SIGINT
3805 problem, and as far as we know, signal-handling now works cleanly.
3806 (If you find any new bugs, please report them!)
3807 * The system no longer defaults Lisp source file names to types
3808 ".l", ".cl", or ".lsp", but only to ".lisp".
3809 * The compiler no longer uses special default file extensions for
3810 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3811 seems to expect a single default extension for all compiled code,
3812 and there's no compelling reason to try to stretch the standard
3813 to allow two different extensions.) Instead, byte-compiled files
3814 default to the same extension as native-compiled files.
3815 * Fasl file format version numbers have increased again, because
3816 a rearrangement of internal implementation packages made some
3817 dumped symbols in old fasl files unreadable in new cores.
3818 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3819 many fewer weird special cases.
3820 * Bug #17 (differing COMPILE-FILE behavior between logical and
3821 physical pathnames) has been fixed, and some related misbehavior too,
3822 thanks to a patch from Martin Atzmueller.
3823 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3824 by Tim Moore, ported to SBCL by Martin Atzmueller.
3825 * Martin Atzmueller fixed several filesystem-related problems,
3826 including bug #36, in part by porting CMU CL patches, which were
3827 written in part by Paul Werkowski.
3828 * More compiler warnings in src/runtime/ are gone, thanks to
3829 more patches from Martin Atzmueller.
3830 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3833 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3835 * DESCRIBE now works on CONDITION objects.
3836 * The debugger now handles errors which arise when trying to print
3837 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3839 * The build system now uses an additional file, customize-target-features.lisp,
3840 to allow local modifications to the target *FEATURES* list. (The point of
3841 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3842 debugging features enabled, without having to worry about propagating it
3843 into everyone's system when I do a "cvs update".) When no
3844 customize-target-features.lisp file exists, the target *FEATURES* list
3845 should be constructed the same way as before.
3846 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3847 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3848 warn about behavior which is undefined under the ANSI spec. Note: This
3849 is specified by ANSI, but it's not very popular with programmers.
3850 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3851 macro in the SBCL sources for an example of a workaround which you
3852 might use to make portable ANSI-standard code which does what you want.
3853 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3854 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3855 when it's used as a non-toplevel form. (This is required in order
3856 to implement the DEFCONSTANT-EQX macro.)
3857 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3858 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3859 funny things, due to bugs in the implementation of EVAL-WHEN
3860 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3861 non-toplevel DEFCONSTANTs are uncommon.)
3862 * The core file version number and fasl file version number have been
3863 incremented, because the old noncompliant DEFCONSTANT behavior involved
3864 calling functions which no longer exist, and because I also took the
3865 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3866 * fixed bug 1 (error handling before read-eval-print loop starts), and
3867 redid debugger restarts and related debugger commands somewhat while
3869 ** The QUIT debugger command is gone, since it did something
3870 rather different than the SB-EXT:QUIT command, and since it never
3871 worked properly outside the main toplevel read/eval/print loop.
3872 Invoking the new TOPLEVEL restart provides the same functionality.
3873 ** The GO debugger command is also gone, since you can just invoke
3874 the CONTINUE restart directly instead.
3875 ** The TOP debugger command is also gone, since it's redundant with the
3876 FRAME 0 command, and since it interfered with abbreviations for the
3878 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3880 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3881 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3882 some porting work by Martin Atzmueller)
3883 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3884 with a boatload of patches which he ported to SBCL. Now that those
3886 ** The system tries to make sure that its low-priority messages
3887 are prefixed by semicolons, to help people who like to use
3888 syntax highlighting in their ILISP buffer. (This patch
3889 was originally due to Raymond Toy.)
3890 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3891 patches originally written by Raymond Toy.
3892 ** The compiler understands coercion between single-value and
3893 multiple-VALUES type expressions better, getting rid of some very
3894 weird behavior, thanks to patches originally by Robert MacLachlan
3895 and Douglas Crosher.
3896 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3897 lambda lists, thanks to a patch originally by Pierre Mai.
3898 ** The system no longer bogusly warns about "abbreviated type
3900 ** The compiler gets less confused by inlining and RETURN-FROM,
3901 thanks to some patches originally by Tim Moore.
3902 ** The system no longer hangs when dumping circular lists to fasl
3903 files, thanks to a patch originally from Douglas Crosher.
3904 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3905 undefined function error.
3906 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3907 in structure slot initforms) for now, documented workaround instead:-|
3908 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3909 * fixed bug 5: added stubs for various Gray stream functions called
3910 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3911 installed, at least appropriate type errors are generated
3912 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3913 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3914 consistently in DEFMETHOD forms.
3915 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3916 it doesn't seem to affect SBCL after all
3917 * The C runtime system now builds with better optimization and many
3918 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3920 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3922 * The system is now under CVS at SourceForge (instead of the
3923 CVS repository on my home machine).
3924 * The new signal handling code has been tweaked to treat register
3925 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3926 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3927 downstream, e.g. in debug-int.lisp, has implicit dependencies
3928 on the unsignedness of integer representation of machine words,
3929 and that caused the system to bomb out with infinite regress
3930 when trying to recover from type errors involving signed values,
3931 e.g. (BUTLAST '(1 2 3) -1).)
3932 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3933 (This was one of the bugs Peter Van Eynde reported back in July.)
3934 * The system now uses code inspired by Colin Walters' O(N)
3935 implementation of MAP (from the cmucl-imp@cons.org mailing
3936 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3937 inline the MAP operation, and there is more than one
3938 sequence argument to the MAP call (so that it can't just
3939 do ETYPECASE once and for all based on the type of the
3940 single sequence argument). (The old non-inline implementation
3941 of the general M-argument sequence-of-length-N case required
3942 O(M*N*N) time when any of the sequence arguments were LISTs.)
3943 * The QUIT :UNIX-CODE keyword argument has been renamed to
3944 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3945 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3947 * Raymond Wiker's patches to port dynamic loading from Linux to
3948 FreeBSD have been added.
3949 * The BUGS file is now more nearly up to date, thanks in large part
3950 to Martin Atzmueller's review of it.
3951 * The debugger now flushes standard output streams before it begins
3952 its output ("debugger invoked" and so forth).
3953 * The core version number and fasl file version number have both
3954 been incremented, because of incompatible changes in the layout
3956 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3957 used to not be called for a saved Lisp image.)
3958 * Martin Atzmueller's version of a patch to fix a compiler crash,
3959 as posted on sbcl-devel 13 September 2000, has been installed.
3960 * Instead of installing Martin Atzmueller's patch for the
3961 compiler transform for SUBSEQ, I deleted the compiler transform,
3962 and transforms for some similar consing operations.
3963 * A bug in signal handling which kept TRACE from working on OpenBSD
3965 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3966 away by constant folding
3967 * The system now defines its address space constants in one place
3968 (in the Lisp sources), and propagates them automatically elsewhere
3969 (through GENESIS and the sbcl.h file). Therefore, patching the
3970 address map is less unnecessarily tedious and error-prone. The
3971 Lisp names of address space constants have also been systematized.
3972 * CVS tags like dollar-Header-dollar have been removed from
3973 the sources, because they have never saved me trouble and
3974 they've been source of trouble working with patches and other
3975 diff-related operations.
3976 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3977 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3979 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3981 * The system has been ported to OpenBSD.
3982 * The system now compiles with a simple "sh make.sh" on the systems
3983 that it's supported on. I.e., now you no longer need to tweak
3984 text in the target-features.lisp-expr and symlinks in src/runtime/
3985 by hand, the make.sh takes care of it for you.
3986 * The system is no longer so grossly inefficient when compiling code
3987 involving vectors implemented as general (not simple) vectors (VECTOR T),
3988 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3989 which dares to use the various sequence functions on non-simple
3990 vectors, takes less of a performance hit.
3991 * There is now a primitive type predicate VECTOR-T-P
3992 to test for the (VECTOR T) type, so that e.g.
3993 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3994 can now be compiled with some semblance of efficiency. (The old code
3995 turned the type declaration into a full call to %TYPEP at runtime!)
3996 * AREF on (VECTOR T) is still not fast, since it's still compiled
3997 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3998 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3999 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4000 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4001 without making full calls to SUBTYPEP at runtime.
4002 (Some analogous efficiency issues for non-simple vectors specialized to
4003 element types other than T, or for non-simple multidimensional arrays,
4004 have not been addressed. They could almost certainly be handled the
4005 same way if anyone is motivated to do so.)
4006 * The changes in array handling break binary compatibility, so
4007 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4008 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4009 instead of (VALUES T T).
4010 * By following the instructions that Dan Barlow posted to sbcl-devel
4011 on 2 July 2000, I was able to enable primitive dynamic object
4012 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4013 functionality is not implemented (since it calls ld to resolve
4014 library references automatically, requiring RUN-PROGRAM for its
4015 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4016 try to resolve library references) is now supported.
4017 * The system now flushes the standard output streams when it terminates,
4018 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4019 them at several other probably-convenient times, e.g. in each pass of
4020 the toplevel read-eval-print loop, and after evaluating a form given
4021 as an "--eval" command-line option. (These changes were motivated by a
4022 discussion of stream flushing issues on cmucl-imp in August 2000.)
4023 * The source transform for TYPEP of array types no longer assumes
4024 that an array whose element type is a not-yet-defined type
4025 is implemented as an array of T, but instead punts, so that the
4026 type will be interpreted at runtime.
4027 * There is now some support for cross-compiling in make.sh: each of
4028 the phases of make.sh has its own script. (This should be transparent
4029 to people doing ordinary, non-cross-compile builds.)
4030 * Since my laptop doesn't have hundreds of megabytes of memory like
4031 my desktop machine, I became more motivated to do some items on
4032 my to-do list in order to reduce the size of the system a little:
4033 ** Arrange for various needed-only-at-cold-init things to be
4034 uninterned after cold init. To support this, those things have
4035 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4036 symbols with such names are now uninterned after cold init).
4037 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4038 fasl files for cold load.
4039 ** Remove the old compiler structure pooling code (which used to
4040 be conditional on the target feature :SB-ALLOC) completely.
4041 ** Redo the representation of some data in cold init to be more compact.
4042 (I also looked into supporting byte compiled code at bootstrap time,
4043 which would probably reduce the size of the system a lot, but that
4044 looked too complicated, so I punted for now.)
4045 * The maximum signal nesting depth in the src/runtime/ support code has
4046 been reduced from 4096 to 256. (I don't know any reason for the very
4047 large old value. If the new smaller value turns out to break something,
4048 I'll probably just bump it back up.)
4049 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4051 * Many, many bugs reported by Peter Van Eynde have been added to
4052 the BUGS list; some have even been fixed.
4053 * While enabling dynamic object file loading, I tried to make the
4054 code easier to understand, renaming various functions and variables
4055 with less ambiguous names, and changing some function calling
4056 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4057 * While trying to figure out how to do the OpenBSD port, I tried to
4058 clean up some of the code in src/runtime/. In particular, I dropped
4059 support for non-POSIX signal handling, added various comments,
4060 tweaked the code to reduce the number of compilation warnings, and
4061 renamed some files to increase consistency.
4062 * To support the new automatic configuration functionality in make.sh,
4063 the source file target-features.lisp-expr has been replaced with the
4064 source file base-target-features.lisp-expr and the machine-generated
4065 file local-target-features.lisp-expr.
4066 * fixed a stupid quoting error in make.sh so that using CMU CL
4067 "lisp -batch" as cross-compilation host works again
4069 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4071 * DESCRIBE no longer tries to call itself recursively to describe
4072 bound/fbound values, so that it no longer fails on symbols which are
4073 bound to themselves (like keywords, T, and NIL).
4074 * DESCRIBE now works on generic functions.
4075 * The printer now prints less-screwed-up representations of closures
4076 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4077 * A private symbol is used instead of the :EMPTY keyword previously
4078 used to mark empty slots in hash tables. Thus
4079 (DEFVAR *HT* (MAKE-HASH-TABLE))
4080 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4081 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4082 now does what ANSI says that it should. (You can still get
4083 similar noncompliant behavior if bang on the hash table
4084 implementation with all the symbols you get back from
4085 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4086 This breaks binary compatibility, since tests for equality to
4087 :EMPTY are wired into things like the macroexpansion of
4088 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4090 * There's now a minimal placeholder implementation for CL:STEP,
4091 as required by ANSI.
4092 * An obscure bug in the interaction of the normal compiler, the byte
4093 compiler, inlining, and structure predicates has been patched
4094 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4095 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4097 * Missing ordinary arguments in a macro call are now detected even
4098 when the macro lambda list contains &KEY or &REST.
4099 * The debugger no longer complains about encountering the top of the
4100 stack when you type "FRAME 0" to explicitly instruct it to go to
4101 the top of the stack. And it now prints the frame you request even
4102 if it's the current frame (instead of saying "You are here.").
4103 * As specified by ANSI, the system now always prints keywords
4104 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4106 * The default initial SIZE of HASH-TABLEs is now smaller.
4107 * Type information from CLOS class dispatch is now propagated
4108 into DEFMETHOD bodies, so that e.g.
4109 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4111 is now basically equivalent to
4112 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4113 (DECLARE (TYPE SINGLE-FLOAT X))
4115 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4116 operation, without having to do run-time type dispatch.
4117 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4118 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4119 * Since I don't seem to be making big file reorganizations very often
4120 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4121 is overflowing my ability to conveniently back them up), I've finally
4122 checked the system into CVS. (The CVS repository is on my home system,
4123 not at SourceForge -- putting it on SourceForge might come later.)
4124 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4125 high-level GC-NOTIFY-FOO functions send their output. (There's
4126 still very little control of where low-level verbose GC functions
4127 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4128 less than it used to -- the GC-NOTIFY-FOO functions are now under
4129 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4130 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4131 in only one place in the source code, and propagates it automatically
4132 everywhere that it's needed. Thus e.g. when I bump the version from
4133 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4134 * The C source files now include boilerplate legalese and documentation
4135 at the head of each file (just as the Lisp source files already did).
4136 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4137 to his page will be replaced with a link to his new CLiki service.
4139 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4141 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4142 * The build process now looks for GNU make under the default name "gmake",
4143 instead of "make" as it used to. If GNU make is not available as "gmake"
4144 on your system, you can change this default behavior by setting the
4145 GNUMAKE environment variable.
4146 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4147 can build without error under CMU CL.
4149 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4151 * There is now a partial SBCL user manual (with some new text and some
4152 text cribbed from the CMU CL manual).
4153 * The beginnings of a profiler have been added (starting with the
4154 CMU CL profiler and simplifying and cleaning up). Eventually the
4155 main interface should be through the TRACE macro, but for now,
4156 it's still accessed through vaguely CMU-CL-style functions and macros
4157 exported from the package SB-PROFILE.
4158 * Some problems left over from porting CMU CL to the new
4159 cross-compilation bootstrap process have been cleaned up:
4160 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4161 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4162 part when getting rid of the compiler *BACKEND* stuff.)
4163 ** The value of *NULL-TYPE* was screwed up, because it was
4164 being initialized before the type system knew the final
4165 definition of the 'NULL type. This screwed up several key
4166 optimizations in the compiler, causing inefficiency in all sorts
4167 of places. (I found it because I wanted to understand why
4168 GET-INTERNAL-RUN-TIME was consing.)
4169 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4170 PROCLAIM FTYPE information. Unfortunately this broke binary
4171 compatibility again, since now the forms output by DEFGENERIC
4172 to refer to functions which didn't exist in 0.6.3.
4173 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4174 can use the new (as of 0.6.3) transform for SOME into MAP into
4176 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4177 values to (INTEGER 0 1000000), so that the time code will no longer
4178 occasionally get blown up by Linux returning 1000000 microseconds
4179 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4180 its output conform to the ANSI spec. (Alas, this makes its output
4181 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4182 * A full call to MAP NIL with a single sequence argument no longer conses.
4183 * fixes to problems pointed out by Martin Atzmueller:
4184 * The manual page no longer talks about multiprocessing as though
4185 it were currently supported.
4186 * The ILISP support patches have been removed from the distribution,
4187 because as of version 5.10.1, ILISP now supports SBCL without us
4188 having to maintain patches.
4189 * added a modified version of Raymond Toy's recent CMU CL patch for
4190 EQUALP comparison of HASH-TABLE
4192 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4194 * The system still can't cross-compile itself with
4195 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4196 entails), but at least it can compile more of itself that way
4197 than it used to be able to, and various buglets which were uncovered
4198 by trying to cross-compile itself that way have now been fixed.
4199 * This release breaks binary compatibility again. This time
4200 at least I've incremented the FASL file format version to 2, so that the
4201 problem can be detected reliably instead of just causing weird errors.
4202 * various new style warnings:
4203 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4204 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4205 ** using the lexical binding of a variable named in the *FOO* style
4206 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4207 as specified by ANSI.
4208 * *RANDOM-STATE* is no longer automatically initialized from
4209 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4210 default behavior of the system is to repeat its behavior every time
4211 it's run. If you'd like to change this behavior, you can always
4212 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4213 old convention there was no comparably easy way to get the system to
4214 repeat its behavior every time it was run.
4215 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4216 FTYPE declarations has been removed, in favor of their ANSI
4217 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4218 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4219 the types of their sequence arguments can be determined at compile time.
4220 This is done through a new open code expansion for MAP which eliminates
4221 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4222 argument types can be determined at compile time.
4223 * The optimizer now transforms COERCE into an identity operation when it
4224 can prove that the coerced object is already of the correct type. (This
4225 can be a win for machine generated code, including the output of other
4226 optimization transforms, such as the MAP transform above.)
4227 * Credit information has been moved from source file headers into CREDITS.
4228 * Source file headers have been made more standard.
4229 * The CASE macro now compiles without complaining even when it has
4232 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4234 * (Note that the way that the PCL macroexpansions were rewritten
4235 to accommodate the change in DEFGENERIC below breaks binary
4236 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4237 not run under sbcl-0.6.2. Once we get out of alpha releases,
4238 i.e. hit release 1.0.0, we'll probably try to maintain binary
4239 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4240 and sbcl-1.4.4. Until then, however, it might be fairly common
4241 for maintenance releases to break binary compatibility.)
4242 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4244 * The reporting of unhandled signals has been changed to print some
4245 explanatory text as well as the report form. (Previously only
4246 the report form was printed.)
4247 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4248 it defines, so that the compiler no longer issues undefined function
4249 warnings for compiled-but-not-yet-loaded generic functions.
4250 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4251 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4252 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4253 section of the ANSI spec are supported.
4254 * The "" nickname for the "KEYWORD" package has been removed.
4255 The reader still handles symbol tokens which begin with a package marker
4256 as keywords, but it doesn't expose its mechanism for doing so in the
4257 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4258 * The system now issues STYLE-WARNINGs for contradictory TYPE
4259 proclamations. (Warnings for contradictory FTYPE proclamations would
4260 be nice too, but those can't be done usefully unless the type system
4261 is made smarter about FUNCTION types.)
4262 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4263 systematized, so that "*target-*.lisp is supposed to exist only on the
4264 target and imply that there's a related file which exists on the
4265 host, and *host-*.lisp is supposed to exist only on the host and imply
4266 that there's a related file which exists on the target. This involves a
4267 lot of renaming. Hopefully the acute confusion caused by the renaming
4268 will be justified by the reduction in chronic confusion..
4269 ** runtime-type.lisp -> early-target-type.lisp
4270 ** target-type.lisp -> late-target-type.lisp
4271 ** early-host-format.lisp -> early-format.lisp
4272 ** late-host-format.lisp -> late-format.lisp
4273 ** host-error.lisp -> misc-error.lisp
4274 ** early-error.lisp -> early-target-error.lisp
4275 ** late-error.lisp -> late-target-error.lisp
4276 ** host-defboot.lisp -> early-defboot.lisp
4277 ** code/misc.lisp -> code/target-misc.lisp
4278 ** code/host-misc.lisp -> code/misc.lisp
4279 ** code/numbers.lisp -> code/target-numbers.lisp
4280 ** code/early-numbers.lisp -> numbers.lisp
4281 ** early-host-type.lisp -> early-type.lisp
4282 ** late-host-type.lisp -> late-type.lisp
4283 ** host-typep.lisp -> typep.lisp
4284 ** load.lisp -> target-load.lisp
4285 ** host-load.lisp -> load.lisp
4286 ** host-disassem.lisp -> disassem.lisp
4287 ** host-insts.lisp -> insts.lisp
4288 ** byte-comp.lisp -> target-byte-comp.lisp
4289 ** host-byte-comp.lisp -> byte-comp.lisp
4290 ** host-signal.lisp -> signal.lisp
4291 ** host-defstruct.lisp -> defstruct.lisp
4292 ** late-target-type.lisp -> deftypes-for-target.lisp
4293 Furthermore, several other previously target-only files foo.lisp (e.g.
4294 hash-table.lisp and random.lisp) have been split into a target-and-host
4295 foo.lisp file and a target-only target-foo.lisp file, with their key type
4296 definitions in the target-and-host part, so that the cross-compiler will
4297 know more about target types.
4298 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4299 gone away. In their place are various *BACKEND-FOO* variables
4300 corresponding to the slots of the old structure.
4301 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4302 into the target SBCL has been fixed.
4303 * The chill.lisp system for loading cold code into a running SBCL
4305 * Support for the CMU CL "scavenger hook" extension has been removed.
4306 (It was undocumented and unused in the CMU CL sources that SBCL was
4307 derived from, and stale in sbcl-0.6.1.)
4308 * Various errors in the cross-compiler type system were detected
4309 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4310 (enabling various consistency checks). Many of them were fixed,
4311 but some hard problems remain, so the compiler is back to
4312 running without *TYPE-SYSTEM-INITIALIZED* for now.
4313 * As part of the cross-compiler type system cleanup, I implemented
4314 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4315 * I have started adding UNCROSS calls throughout the type system
4316 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4317 on cross-compiler input files will be able to go away, and various
4319 * CONSTANTP now returns true for quoted forms (as explicitly required
4322 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4324 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4325 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4326 (MAKE-LIST -1). In the longer term, it ought to become true
4327 that declarations are assertions even at SAFETY 1. For now, it's not
4328 quite true even at SAFETY 3, but it's at least more nearly true..
4329 (Note that this change seems to increases the size of the system by
4330 O(5%) and to decrease the speed of the compiler by 20% or more.)
4331 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4332 for the problem of printing dozens of lines of distracting information
4333 about low-level system machinery as part of the top stack frame
4334 on entry to the debugger when an undefined function was called.
4335 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4336 is not reset to COMMON-LISP-USER.
4337 * Compilation of stuff related to dyncount.lisp has been made conditional
4338 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4339 smaller. The various dyncount-related symbols have been moved into
4340 a new "SB-DYNCOUNT" package.
4341 * tty-inspect.lisp has been renamed to inspect.lisp.
4342 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4343 feature has gone away. (When we eventually port to other flavors of
4344 libc and/or Unix, we'll try to make the differences between flavors
4345 invisible at the user level.)
4346 * Various other *FEATURES* tags, and/or their associated conditionals,
4347 have been removed if obsolescent, or given better documentation, or
4348 sometimes given more-mnemonic names.
4350 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4352 * tidied up "make.sh" script
4353 * tidied up system directory structure
4354 * better "clean.sh" behavior
4355 * added doc/FOR-CMUCL-DEVELOPERS
4356 * many many small tweaks to output format, e.g. removing possibly-confusing
4357 trailing #\. character in DESCRIBE-INSTANCE
4358 * (EQUALP #\A 'A) no longer signals an error.
4359 * new hashing code, including EQUALP hashing
4360 * tidied up Lisp initialization and toplevel
4361 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4362 * command line argument processing
4363 * added POSIX-GETENV function to deal with Unix-ish environment variables
4364 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4365 terminating SBCL on EOF
4366 * non-verbose GC by default
4367 * There is no more "sbcl" shell script; the sbcl file is now the C
4368 runtime executable (just like CMU CL).
4369 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4371 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4372 symbol and package information as Lisp native symbol and package objects
4373 instead of strings naming symbols and strings naming packages. This way,
4374 whenever packages are renamed (as in warm init), debug information is
4375 transformed along with everything else.
4376 * tweaked the optimization policy declarations which control the building
4377 of SBCL itself. Now, among other things, the system no longer saves
4378 source location debugging information. (This helps two problems at once
4379 by reducing SBCL size and by keeping SBCL from trying to look for its
4380 sources -- which may not exist -- when reporting errors.)
4381 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4382 debugging and testing purposes
4383 * cleaned up printing, making the printer call PRINT-OBJECT for
4384 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4385 methods, giving nearly-ANSI behavior
4386 * converted almost all special variables to use *FOO* naming convention
4387 * deleted PARSE-TIME functionality, since it can be done portably
4388 * moved some files out of cold init into warm init
4389 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4391 * regularized formatting of source files
4392 * added an install.sh script
4393 * fixed ridiculous memory usage of cross-compiler by making
4394 compiler/alloc.lisp not try to do pooling unless it can hook
4395 itself into the GC of the cross-compilation host. Now the system
4396 builds nicely on my old laptop.
4397 * added :SB-ALLOC in target-features.lisp-expr
4398 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4399 was not implemented)
4400 * re-did condition handling and note reporting in the compiler. Notes
4401 are no longer handled by signalling conditions. Style warnings
4402 and warnings are handled more correctly and reported in such a way
4403 that it's easy to find one or the other in your output (so that you
4404 can e.g. figure out which of many problems caused COMPILE-FILE to
4406 * changed the severity of several compiler warnings from full WARNING
4407 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4408 compiler note reporting so that it doesn't use the condition system
4409 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4410 COMPILE-FILE command)
4411 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4412 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4413 no longer supported, since their implementation was deeply intertwingled
4414 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4415 * removed bogus "support" for compiler macros named (SETF FOO), and
4416 removed the compiler macro for SETF INFO (but only after making a fool
4417 of myself on the cmucl-imp mailing list by posting a bogus patch for
4418 DEFINE-COMPILER-MACRO..)
4419 * Compiled files containing forms which have side effects on the Lisp
4420 reader (such as DEFPACKAGE forms) are now handled more correctly.
4421 (Compiler queuing of top level lambdas has been suppressed by setting
4422 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4423 * deleted various currently-unused source files, e.g. gengc.lisp. They
4424 may be added back at some point e.g. when porting to other architectures,
4425 but until they are it's distracting to distribute them and to try to
4427 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4428 there were so many of them they're just distractions, and UNCROSS is
4429 known to be able to handle the current sources
4430 * moved PROFILE functionality into TRACE, so that it will be clear
4431 how the wrapping and unwrapping of functions when you profile them
4432 interacts with the wrapping and unwrapping of functions when you
4433 trace them. (Actually, the functionality isn't there yet, but at least
4434 the interface specification is there. Hopefully, the functionality will
4435 arrive with some future maintenance release.)
4436 * removed host-oops.lisp
4437 * changed signature of QUIT function to allow UNIX-CODE argument
4438 * fixed READ-SEQUENCE bug
4439 * tweaked verbose GC output so that it looks more like the progress
4440 output that ANSI specifies for functions like LOAD
4441 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4442 * added <http://sbcl.sourceforge.com> to the banner information printed by