1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
3 * incompatible change: removed writer methods for host-ent-name,
4 host-ent-addresses -- changing the values did not update the DNS
6 * minor incompatible change: changed experimental JOIN-THREAD interface
7 * documentation: the manual now lists reader and writer methods
8 in class slot documentation sections. (thanks to Richard M Kreuter)
9 * documentation: unwinding from asyncronous events has been
11 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
12 in multithreaded application code.
13 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
14 platforms. (thanks to James Anderson for the optimization hint)
15 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
17 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
18 variants no longer cons.
19 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
20 their NOT- variants no longer cons.
21 * optimization: Stack allocation of arrays containing unboxed
22 elements is slightly more efficient on x86 and x86-64.
23 * enhancement: XREF information is now collected to references made
24 to global variables using SYMBOL-VALUE and a constant argument.
25 * enhancement: SIGINT now causes a specific condition
26 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
27 * bug fix: inlined calls to C now ensure 16byte stack alignment on
29 * bug fix: bad type declaration in the CLOS implementation has
30 been fixed. (reported by James Anderson)
31 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
32 (thanks to Lutz Euler)
33 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
34 required by the CLHS. (reported by Samium Gromoff)
35 * bug fix: dead unbound variable references now signal an error.
36 (reported by Marco Monteiro)
37 * bug fix: / with an unused value was being deleted in safe code.
38 (thanks to Marco Monteiro and Kevin Reid)
39 * bug fix: number of characters that can be written onto a single
40 line in a file is unlimited.
41 * bug fix: some GC deadlocks caused by asynchronous interrupts have
42 been fixed by inhibiting interrupts for when GC is disbled.
43 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
45 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
46 specifier no longer causes infinite recursion.
47 * improvement: the x86-64/darwin port now passes all tests and
48 should be considered non-experimental.
50 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
51 * new platform: experimental support for x86-64/darwin (MacOS).
52 * incompatible change: the thread-safe (on most platforms) getaddrinfo
53 and getnameinfo sockets functions are used instead of gethostbyaddr
54 and gethostbyname, on platforms where the newer functions are available.
55 As a result, the ALIASES field of HOST-ENT will always be NIL on these
57 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
58 don't print the startup banner, but behave as if --noinform was passed
59 as a command line argument. (thanks to Kevin Reid)
60 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
61 * optimization: code using alien values with undeclared types is much faster.
62 * optimization: the compiler is now able to open code SEARCH in more cases.
63 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
64 * bug fix: using standardized COMMON-LISP special variables as loop
65 variables no longer signals bogus package lock violations. (reported
67 * bug fix: declaring local loop variables to be of a range-limited type
68 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
69 (reported by Andras Simon)
70 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
71 bugs remain on x86-64.)
72 * bug fix: the #= and ## reader macros now interact reasonably with
73 funcallable instances.
74 * bug fix: type-checks for function arguments were compiled using the
75 compiler policy settings of the wrong lexical environment (in 1.0.2
77 * bug fix: SHADOW accepts characters as string designators, as required
78 by the spec (thanks to Eric Marsden)
79 * bug fix: fixed GC safety issues when foreign functions are called with
80 non-base strings as arguments
81 * bug fix: more consistent error output for fatal-compiler-warnings, like
83 * bug fix: functions with proper names were showing up as NIL in
84 backtraces (reported by Edi Weitz, regression in 1.0.2)
86 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
87 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
88 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
89 produce more efficient code on the x86 if the compiler can determine
90 that (+ INDEX <constant>) does not require a bounds check and FOO
91 has an element type at least 8 bits wide.
92 * bug fix: references to undefined variables are handled the same way
93 in toplevel forms as in the normal compiler
94 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
96 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
97 (thanks to Jon Buller)
98 * improvement: faster compilation times for complex functions
99 * improvement: added readlink support to SB-POSIX (thanks to Richard
102 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
103 * improvement: experimental support for mach exception handling on
104 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
106 * improvement: support for GBK external format.
107 (thanks to Chun Tian (binghe))
108 * improvement: the debugger now displays variables that have been closed
109 over, in code compiled with (DEBUG 2) or higher
110 * improvement: support for executable cores on NetBSD (thanks to
112 * new feature: added a RESTART-FRAME debugger command
113 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
114 be used to provide an implementation for FILE-POSITION on Gray streams
115 (thanks to Eric Marsden)
116 * new feature: add syslog, openlog and closelog support to SB-POSIX
117 (thanks to Richard Kreuter)
118 * optimization: the function call overhead in code compiled with
119 a high DEBUG optimization setting is significantly
120 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
121 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
122 (thanks to Tony Martinez)
123 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
124 evaluated once (reported by Kevin Reid)
125 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
126 works again (thanks to Kilian Sprotte)
127 * bug fix: an error is signaled for tagbodies with duplicate tags
128 (thanks to Stephen Wilson)
129 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
130 * bug fix: Win32 port can now handle foreign code unwinding Lisp
131 stack frames from alien callbacks.
132 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
133 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
134 65k pages (thanks to David Woodhouse)
135 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
137 changes in sbcl-1.0.1 relative to sbcl-1.0:
138 * new platform: FreeBSD/x86-64, including support for threading.
139 * new feature: the compiler stores cross-referencing information
140 abount function calls (who-calls), macroexpansion (who-macroexpands)
141 and special variables (who-binds, who-sets, who-references) for code
142 compiled with (< SPACE 3). This information is available through the
143 sb-introspect contrib.
144 * new feature: users may subclass SEQUENCE, and have instances of
145 these classes interoperate with standard Common Lisp functions if
146 a number of methods are defined. (This feature is experimental
147 and the interface subject to change based on feedback from SBCL
148 users and the general community)
149 * improvement: sb-sprof traces call stacks to an arbitrary depth on
150 x86/x86-64, rather than the previous fixed depth of 8
151 * improvement: another pthread back-end of mutex "pthread-futex".
152 * bug fix: non-ascii command-line arguments are processed correctly
153 (thanks to Yaroslav Kavenchuk)
154 * bug fix: non-required arguments were not passed correctly when a method
155 defined using DEFMETHOD was called from a mop-generated method using
156 CALL-NEXT-METHOD (reported by Pascal Costanza)
157 * bug fix: recursion is now permitted in accessors through
158 SLOT-UNBOUND. (reported by Pascal Costanza)
159 * bug fix: an error was signaled at startup if the HOME environment
160 variable was defined, but had an empty value (reported by Peter Van Eynde)
161 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
162 signal a PROGRAM-ERROR, not a BUG.
163 * bug fix: SB-POSIX:READDIR works when built with large file support.
164 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
166 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
167 for better automatic indentation support. (thanks to Matt Pillsbury)
168 * optimization: loading generic functions no longer takes O(n^2) time,
169 proportional to the amount of methods in the generic function
170 (reported by Todd Sabin and Jeremy Brown)
171 * optimization: the FIND and POSITION family of sequence functions
172 are significantly faster on arrays whose element types have been
174 * improvements to the Windows port:
175 ** Intermittent heap corruption problems have been fixed. (thanks
176 to Alastair Bridgewater)
177 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
179 ** Lisp is able to unwind foreign exception frames from alien
180 callbacks. (thanks to Alastair Bridgewater)
182 changes in sbcl-1.0 relative to sbcl-0.9.18:
183 * improvement: experimental support for threading on FreeBSD/x86.
184 (thanks to NIIMI Satoshi)
185 * improvement: runtime option --dynamic-space-size can be used
186 to set the size of the dynamic space reserved on startup.
187 * improvement: floating point modes in effect are now saved in
188 core, and restored on startup.
189 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
190 startup, not time since first call to GET-INTERNAL-REAL-TIME.
191 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
192 threads are not running after *SAVE-HOOKS* have run.
193 * improvement: writes to CLOS instance slots are type-checked in code
194 compiled with (SAFETY 3)
195 * improvement: floating-point exception handling on FreeBSD (thanks to
197 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
198 (thanks to Zach Beane)
199 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
201 * improvement: added support for the Shift-JIS external format.
202 (contributed by NIIMI Satoshi)
203 * improvement: callbacks are supported on Linux/PPC. (thanks to
205 * bug fix: compiler bug triggered by a (non-standard) VALUES
206 declaration in a LET* was fixed. (reported by Kaersten Poeck)
207 * bug fix: file compiler no longer confuses validated and already
208 dumped structurres. (reported by Kaersten Poeck)
209 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
210 fill-pointer signals a type-error as required. (thanks to
212 * bug fix: disassemly of funcallable instances works.
213 * bug fix: single stepping on PPC.
214 * bug fix: fix thread-safety problems in the type system (generally
215 manifesting as nonsensical errors like "STRING is a bad type specifier
216 for sequences" or "The value 1 is not of type FIXNUM")
217 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
218 (reported by Josip Gracin)
219 * bug fix: an error is signaled for attempts to displace arrays with
220 incompatible element types (thanks to Mario Mommer)
221 * bug fix: more correct handling of wide characters in debug info
222 (bug reported by Attila Lendvai and fixed by Juho Snellman)
223 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
224 and don't cause extra consing
225 * optimization: MAP and MAP-INTO are significantly faster on vectors
226 whose elements types have been declared.
227 * Improvements to SB-SPROF:
228 ** Support for allocation profiling
229 ** Reduced profiling overhead, especially for long profiling runs
230 * Improvements to the Windows port:
231 ** floating point exceptions are now reported correctly.
232 ** stack exhaustion detection works partially.
233 ** more accurate GET-INTERNAL-REAL-TIME.
234 ** better RUN-PROGRAM behaviour in case of errors in the child
236 ** PROBE-FILE now simplifies pathnames correctly.
237 ** DIRECTORY now works correctly with :WILD-INFERIORS.
238 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
239 1st 1970 (although time timezone data will be unreliable).
240 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
242 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
243 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
244 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
246 * bug fix: two potential GC deadlocks affecting threaded builds.
247 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
248 returns T (reported by Anton Kazennikov)
249 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
250 with non-variable places
251 * bug fix: remove a race condition in the setting of
252 funcallable-instance functions, this should make threaded CLOS
253 code more stable against memory faults.
254 * bug fix: corruption of specials when unbinding is interrupted by an
255 asynchronous unwind (reported by Hannu Koivisto)
256 * improvement: the debugger will now also display local variables that
257 are only used once, for code compiled with a DEBUG optimization quality
260 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
261 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
262 * incompatible change: External-format support for FFI calls. The
263 SB-ALIEN:C-STRING no longer implies an ASCII
264 external-format. Instead the string is subject to external-format
265 conversion. Additionally return values of type SB-ALIEN:C-STRING
266 are by default of type (SIMPLE-ARRAY CHARACTER), not
267 SIMPLE-BASE-STRING. For an alien type that behaves like the old
268 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
269 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
270 * incompatible change: SB-EXT package no longer contains the
271 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
272 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
274 * incompatible change: the single-stepper is no longer available
275 on Alpha, Mips and Sparc platforms.
276 * minor incompatible change: the direct superclasses of
277 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
278 not (STANDARD-OBJECT FUNCTION). This makes the
279 class-precedence-lists of GENERIC-FUNCTION and
280 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
282 * minor incompatible change: the default stream external format on
283 non-unicode SBCL is no longer detected from the locale, but is always
285 * new feature: Add a version of evaluator that uses an interpreter instead
286 of the compiler. EVAL still uses the compiler by default, to switch it
287 to use the interpreter, set the value of the variable
288 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
289 * minor incompatible change: the single-stepper REPL has been merged
290 with the normal debugger (see the "Stepping" heading of the debugger help
291 for more details). The debugger command STEP will no longer switch
292 to the single-stepper REPL.
293 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
294 for a type now works.
295 * bug fix: loading of default sysinit file works. (thanks to Leonid
297 * bug fix: better detection of circularities in the file-compiler.
298 (reported by Marco Monteiro)
299 * bug fix: the CL pathname functions now work with files that have
300 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
301 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
302 systems with Unix98 pty semantics.
303 * bug fix: ASDF-INSTALL will now work with bsd tar.
304 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
306 * bug fix: timers expiring in dead threads no longer cause a
307 type-error (reported by Paul "Nonny Mouse").
308 * bug fix: thanks to more lightweight single-stepper instrumentation,
309 code compiled with (DEBUG 3) will compile and execute significantly faster,
310 and will have more accurate type-inferencing than before
311 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
312 whose bindings are modified
313 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
314 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
315 ** replace dummy implementations of CL:MACHINE-INSTANCE and
316 CL:SOFTWARE-VERSION with proper version
318 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
319 * feature: implemented the READER-METHOD-CLASS and
320 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
321 as specified by AMOP.
322 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
324 * optimization: faster LOGCOUNT implementation on x86 and x86-64
325 (thanks to Lutz Euler)
326 * optimization: hashing of general arrays and vectors has been
327 improved. (reported by Any Fingerhut)
328 * enhancement: SB-INTROSPECT is now able to find definitions of
329 profiled functions. (thanks to Troels Henriksen)
330 * enhancement: compiler-macro expansion applies now to FUNCALL forms
331 as well, allowing compiler-macros for SETF-functions to expand.
332 * enhancement: step-instrumentation no longer wraps known
333 single-value functions with multiple-value context, allowing
334 better type inference.
335 * fixed bug #337: use of MAKE-METHOD in method combination now works
336 even in the presence of user-defined method classes. (reported by
337 Bruno Haible and Pascal Costanza)
338 * fixed bug #339(c): if there are applicable methods not part of any
339 long-form method-combination group, call INVALID-METHOD-ERROR.
340 (reported by Bruno Haible)
341 * fixed bug #361: the :FUNCTION initarg in the protocol for
342 initialization of methods can now be used to override
343 internally-produced optimized functions. (reported by Bruno
345 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
346 system-provided lambda expression no longer cause warnings about
347 unbound #:|pv-table| symbols.
348 * bug fix: improved the handling of type declarations and the
349 detection of violations for keyword arguments with non-constant
351 * bug fix: potentially erronous calls to PATHNAME and
352 MERGE-PATHNAMES were being flushed in some cases.
353 (reported by Richard Kreuter)
354 * bug fix: compiled calls to TYPEP were mishandling obsolete
355 instances. (reported by James Bielman and Attila Lendvai)
356 * bug fix: format strings with ~> without matching ~< no longer
357 trigger an AVER, but signal an understandable error instead.
358 (reported by Antonio Martinez)
359 * bug fix: specifying an output-file without a pathname-type for
360 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
361 of input-file instead of "fasl". (reported by Robert Dodier)
362 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
363 (reported by James Y Knight).
364 * bug fix: compiler-macros-function did not consider the environment
365 argument for shadowing by local functions.
366 * bug fix: compiler-macros expansion was inhibited by local INLINE
368 * bug fix: inline expansions of known functions were subject to
369 step-instrumentation in high debug policies, leading to problems
371 * bug fix: compiler failed to differentiate between different CONS
373 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
374 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
375 * bug fix: #368: incorrect use of expressed vs. upgraded array
377 * bug fix: #306a: more precise unions of array types.
378 * thread-safety improvements:
379 ** CONDITION-WAIT could return early on Linux, if the thread was
380 interrupted and subsequently continued with SIGCONT.
381 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
383 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
384 * added support for the ucs-2 external format. (contributed by Ivan
386 * minor incompatible change: pretty printing of objects of type
387 (cons symbol) is, in the default pprint-dispatch-table, now
388 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
390 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
391 called later in a class's lifetime, possibly as late as when the
392 first instance of the class is created. Previously,
393 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
394 class became finalizeable.
395 * fixed bug: FILE-POSITION sometimes returned inconsistent results
396 for multibyte external-format streams. (thanks to "vbzoli")
397 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
398 with :ALLOCATION :CLASS inherited from superclasses of the
400 * fixed bug: anonymous classes can now be created using the :NAME
401 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
402 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
403 * fixed bug: core-files saved with :EXECUTABLE T can again be
404 executed when SBCL_HOME isn't set. (reported by James Knight)
405 * fixed bug: toplevel LOCALLY forms with declarations could
406 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
407 * fixed bug: printing from several different threads using different
408 values of *print-case* could cause invalid output, due to
409 some internal special variables of the printer not being bound
410 thread-locally (reported by Max Mikhanosha)
411 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
412 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
413 (reported by Pascal Costanza)
414 * fixed bug: FINALIZE-INHERITANCE is called from
415 REINITIALIZE-INSTANCE on classes when the class has previously
416 been finalized, as required by AMOP.
417 * minor code generation optimizations:
418 ** better register allocation in CLOS dispatching functions
419 ** overflow detection when coercing signed bytes to fixnums on x86-64
420 ** is now implemented with one IMUL instruction instead of three shifts
421 ** more efficient bit-vector access on x86 and x86-64
422 ** more efficient access to raw structure slots on x86 and x86-64
423 * fixed some bugs revealed by Paul Dietz' test suite:
424 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
427 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
428 * feature: thread support on Solaris/x86, and experimental thread support
430 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
432 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
433 default method for SLOT-UNBOUND.
434 * minor incompatible change: prevent the user from specializing the
435 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
436 somewhat counter to the intent of the protocol, I (CSR) think, and
437 additionally it just doesn't work in SBCL as currently
438 implemented, thanks to optimizations (that are always valid for
439 the other three SLOT-VALUEish functions, but not for the setter).
440 * minor incompatible change: the :SB-LDB feature is now enabled by
441 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
442 the low-level debugger.
443 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
444 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
445 down to an SB-SYS:FD-STREAM, allowing direct communication in
446 more cases. Temporary files and pipes are still used as a fallback
448 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
449 on the non-reentrant C functions localtime(3) and gmtime(3).
450 * bug fix: global optimization policy was not visible in LOCALLY and
452 * bug fix: class objects can be used as specializers in methods.
453 (reported by Pascal Costanza)
454 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
455 with a type error. (reported by blitz_ on #lisp)
456 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
457 failed with a type error. (reported by Pascal Bourguignon)
458 * bug fix: merging pathnames against defaults with :DIRECTORY
459 starting with '(:RELATIVE :BACK) should preserve the :BACK.
460 (reported by James Y Knight)
461 * bug fix: saving large (>2GB) cores on x86-64 now works
462 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
463 constant index and a (simple-array (signed-byte 32)) array
464 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
465 error (patch by Robert J. Macomber)
466 * feature: TIME now displays the amount of run-time spent in GC
467 * bug fix: The debugger now does a better job of respecting
468 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
469 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
470 workaround for bug 403.)
471 * bug fix: floating point exception handling now works on FreeBSD
472 (thanks to NIIMI Satoshi)
473 * fixed some bugs revealed by Paul Dietz' test suite:
474 ** MISC.641: LET-conversion were not supposed to work in late
476 * improvements to the Win32/x86 port:
477 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
478 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
480 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
481 * new feature: source path information is generated for macro-expansion
482 errors for use in IDE's like Slime (thanks to Helmut Eller)
483 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
484 * bug fix: compiler does not loop forever on an invalid type in
486 * improvement: compilation of most CLOS applications is significantly
488 * optimization: added a limited bytecode compiler for simple toplevel
489 forms, speeding up compilation and FASL loading
490 * bug fix: the statistical profiler now properly distinguishes anonymous
493 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
494 * minor incompatible change: in sbcl-0.9.11 (but not earlier
495 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
496 into subclasses of the finalized class. Now user calls to
497 FINALIZE-INHERITANCE finalize just the one class, and calls by the
498 system return before any subclasses are finalized.
499 * minor incompatible change: The reader no longer ignores errors
500 regarding non-existent packages in #+ and #- feature tests.
501 * new feature: command line options --no-sysinit, --no-userinit to
502 inhibit loading the corresponding init files
503 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
504 for sending data through UDP sockets (thanks to François-René Rideau)
505 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
506 error is signalled instead (thanks to François-René Rideau)
507 * minor incompatible change: Error signalling behaviour of lexical
508 operations violating package locks has changed slightly. Refer to
509 documentation on package locks for details.
510 * enhancement: EVAL can process IF-expressions without resorting to the
512 * bug fix: Some operations involving SIN, COS, or TAN weren't being
513 constant-folded properly due to missing out-of-line functions.
514 (reported by Mika Pihlajamäki)
515 * bug fix: LISTEN sometimes returned T even in cases where no data was
516 immediately available from the stream
517 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
518 were reversed. (reported by Levente Mészáros)
519 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
520 when given a vector displaced to an adjustable array. (reported
522 * bug fix: derivation of float boundaries from numbers outside the
523 appropriate float range (reported by John Wiseman)
524 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
526 * bug fix: precomputing cache entries for generic functions with
527 some subclasses of specializers as yet invalid does not attempt to
528 fill a cache line with a negative offset. (reported by Levente
530 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
531 allow condition class objects as arguments to DOCUMENTATION and
532 (SETF DOCUMENTATION); only find and set documentation for
533 structure names for the STRUCTURE doc-type. (suggested by Gary
535 * improvements to the Win32/x86 port:
536 ** (user-homedir-pathname) and default initialization file
537 locations now know about the user's "Documents and Settings"
538 directory (thanks to Yaroslav Kavenchuk)
539 ** run-program is implemented (thanks to Mike Thomas)
540 ** sockets support (thanks to Timothy Ritchey)
541 ** better backtrace support (thanks to Alastair Bridgewater)
542 ** sb-grovel supported
543 ** asdf-install and sb-posix work somewhat
544 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
545 * improvements to the Solaris/x86 port:
546 ** works on Solaris 11/Solaris Express
547 ** floating-point exception handling support
548 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
549 * fixed some bugs revealed by Paul Dietz' test suite:
550 ** REMOVE-METHOD returns its generic function argument even when
551 the method is not one of the generic functions' methods.
552 ** objects declared as MEMBER types can be admissible arguments to
554 ** printing characters should simply be printed by the FORMAT ~:C
556 ** compiler failure when compiling functions with hairy constant
557 defaults for optional parameters.
558 ** compiler produces wrong code when MAYBE-INLINE-expanding a
559 function, which is already optimized.
561 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
562 * new platform: experimental support for SBCL x86/Darwin, including
563 MacOS X 10.4.5 on Intel.
564 * new feature: Unicode character names are now known to the system
565 (through CHAR-NAME and NAME-CHAR).
566 * new feature: the filesystem location of SBCL's core file is
567 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
568 * minor incompatible change: the contrib modules SB-POSIX and
569 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
570 this change is to make it easier to distribute
571 location-independent binaries.
572 * bug fix: as implied by AMOP, standardized classes no longer have
573 slots named by external symbols of public packages. (reported by
575 * optimization: calling functions via constant symbols -- as in
576 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
577 via the function object as in (FUNCALL #'FOO).
578 * optimization: on x86 and x86-64, the calling convention for the
579 case of dealing with an unknown number of values has been altered
580 to be friendlier to the prediction heuristics implemented,
581 particularly with reference to CALL and RET pairing. (thanks to
582 Alastair Bridgewater)
583 * optimization: on x86, the code for access to thread-local symbol
584 values and binding/unbinding of thread-local symbols is smaller.
585 (thanks to Alastair Bridgewater)
586 * enhancement: CONSTANTP is now able to determine constantness of
587 more complex forms, including calls to constant-foldable standardized
588 functions and some special forms beyond QUOTE.
589 * fixed bug: occasional GC crashes on Solaris/x86
590 * optimization: x86-64 supports stack allocation of results of simple
591 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
592 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
593 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
594 (thanks to James Knight)
595 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
596 by the standard (thanks to Stephen Compall)
598 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
599 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
600 be used for bundling the runtime and the core file into one
601 executable binary. This feature is not currently supported on all SBCL
602 platforms. (thanks to James Bielman and NIIMI Satoshi)
603 * new feature: a generational or ephemeral garbage collector is now
604 the default on the PowerPC platform (both Linux and Darwin). The
605 old Cheney (stop and copy) collector is a build-time option.
606 (thanks to Cyrus Harmon, after Raymond Toy)
607 * minor incompatible change: the method by which SBCL finds its
608 contributed modules has changed; it no longer relies on symbolic
609 links from an $SBCL_HOME/systems directory, but searches directly
610 in the subdirectories of $SBCL_HOME.
611 * enhancement: the dynamic heap size on the Linux/PPC platform is
612 markedly larger, even using the older Cheney garbage collector.
613 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
614 certain complicated string types. (reported by Gary King)
615 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
616 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
617 * fixed bug: DOCUMENTATION on structures no longer signals an
618 error if no structure type of the right name exists. (reported by
620 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
621 applicable non-standard primary or :AROUND method on
622 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
623 no longer cause unbound variable errors. (reported by Kevin Reid)
624 * fixed bug: metacircle resolution in cases where methods have slots
625 added before the slots from STANDARD-METHOD. (reported by Jean
627 * fixed bug: the Power PC assembly code for calling into lisp
628 sometimes computed its return address wrongly (depending on the
629 15th bit of the address, fixed at link time). This used to
630 manifest itself in a segmentation violation while building PCL.
631 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
632 many others over the years)
633 * fixed bug: ROOM no longer reports silly numbers for stack usage if
634 the stack pointer (treated as a signed integer) is negative.
635 (thanks to Peter van Eynde)
637 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
638 * new platform: experimental support for the Windows operating
639 system has been added. (thanks to Alastair Bridgewater)
640 * fixed several bugs in and robustified the PPC FFI (including
641 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
642 * bug fix: allow non-simple string symbol names (reported by Paul
644 * bug fix: interrupt handling on NetBSD (thanks to Richard M
646 * bug fix: saving a core corrupted callbacks on x86/x86-64
647 * bug fix: closed a loophole in metacircularity detection and
648 grounding in the PCL implementation of CLOS.
649 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
650 (reported by tomppa on #lisp)
651 * bug fix: clear the direction flag on Lisp -> C transitions, as
652 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
653 (reported by Andrej Grozin and Hendrik Maryns)
654 * optimization: major improvements to GC efficiency on GENCGC platforms
655 * optimization: faster implementation of EQUAL
656 * optimization: emit more efficient opcodes for some common
657 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
658 * optimization: several other minor code-generation improvements on x86-64
659 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
661 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
662 * minor incompatible change: (SETF CLASS-NAME) and (SETF
663 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
664 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
665 by ANSI as a generic function, and for consistency (SETF
666 GENERIC-FUNCTION-NAME) is treated likewise.
667 * fixed bug #233.b: make constraint propagation notice when a variable
668 value is changed after it is referenced but before it is used
669 * fixed bug #296: no more arbitrary behaviour from filenames with
670 odd characters as --load arguments. (reported by Adam Warner)
671 * fixed bug #390: :CHARACTER-SET pathname components now work as
672 expected. (reported by Tim Daly Jr)
673 * fixed bug #391: complicated :TYPE intersections in slot
674 definitions no longer cause an error in PCL internals.
675 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
676 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
677 returning the number of octets which would be written to the
678 file-stream. (thanks to Robert J. Macomber)
679 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
680 arguments to the internal stream functions. (thanks to David
682 * bug fix: stack exhaustion is now handled on the x86/SunOS
683 platform. (thanks to Daisuke Homma)
684 * bug fix: more accurate ROOM results on GENCGC platforms
685 * optimization: improved type inference for arithmetic-for
686 index variables in LOOP
687 * optimization: faster floating-point SQRT on x86-64
688 * fixed some bugs revealed by Paul Dietz' test suite:
689 ** DOCUMENTATION returns NIL instead of "" for method combinations
690 that don't have a docstring
692 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
693 * minor incompatible change: (SETF CLASS-NAME) and (SETF
694 GENERIC-FUNCTION-NAME) are no longer generic functions, and
695 therefore are not customizeable by user code (as seems to be at
696 least permitted and maybe required by AMOP). As a consolation,
697 however, the SBCL implementation of these functions now calls
698 REINITIALIZE-INSTANCE as specified by AMOP.
699 * bug fix: slot-definition documentation is propagated as per ANSI
700 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
701 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
702 argument, as do eslotd initialization methods. (from Pascal
703 Costanza's "Closer" project)
704 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
705 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
707 * bug fix: it is now possible to have more than one subclass of
708 STANDARD-GENERIC-FUNCTION without causing stack overflow.
709 (reported by Bruno Haible, Pascal Costanza and others)
710 * bug fix: the dependent update protocol now works for generic
711 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
713 * bug fix: condition-class instances corresponding to
714 DEFINE-CONDITION forms are now created eagerly. (reported by
715 Kalle Olavi Niemitalo on comp.lang.lisp)
716 * bug fix: floating point printing is more accurate in some
717 circumstances. (thanks to Simon Alexander)
718 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
719 merged with *DEFAULT-PATHNAME-DEFAULTS*.
720 * bug fix: callbacks on OS X now preserve stack-alignment.
721 (thanks to Cyrus Harmon)
722 * enhancement: the x86-64 disassembler is much better at
723 disassembling SSE instructions. (thanks to Lutz Euler)
724 * enhancement: CHECK-TYPE on a variable now allows type-inference
725 (thanks to Vincent Arkesteijn)
726 * optimization: improved performance of EUC-JP external format.
727 (thanks to NIIMI Satoshi)
728 * optimization: performance improvements to IO on file streams of
729 :ELEMENT-TYPE CHARACTER
730 * optimization: much faster memory allocation on x86-64
731 * optimization: faster garbage collections (latency and throughput)
733 * optimization: faster bignum subtraction and division on x86 and x86-64
734 (thanks to Lutz Euler)
735 * optimization: more accurate type-inference for loops that use a
736 floating point index variable or a negative step.
738 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
739 * bug fix: add a workaround to SBCL looping infinitely at startup on
740 Linux kernels with apparently buggy implementations of personality().
741 (thanks to Svein Ove Aas)
742 * bug fix: Unicode symbols are correctly printed in LDB backtraces
743 (thanks to David Lichteblau)
744 * bug fix: local bindings shadow global symbol macros for
745 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
746 * optimization: non-open coded uses of numeric comparison operators
747 (e.g. >) no longer cons when called with more than one parameter
748 on platforms supporting dynamic-extent allocation.
749 * enhancement: saving cores with foreign code loaded is now supported
750 on MIPS/Linux in addition to the previously supported platforms.
751 * bug fix: FIND-RESTART now tests for activity, not applicability when given
752 a restart object as identifier. (reported by Helmut Eller for CMUCL)
753 * bug fix: division by zero in sb-sprof when no samples were collected
754 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
755 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
756 (thanks to Alexander Kjeldaas)
757 * new feature: ignoring the return values of destructive functions
758 like NREVERSE signals a compile-time style-warning.
759 (thanks to Kevin Reid)
760 * new feature: the alignment of alien structure fields can be
761 explicitly specified. (thanks to Cyrus Harmon)
762 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
763 and dump core on SIGQUIT
765 ** incompatible change: threads do not inherit values of specials
766 from their parents (see manual)
767 ** bug fix: threads stacks belonging to dead threads are freed by the
768 next exiting thread, no need to gc to collect thread stacks anymore
769 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
770 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
771 ** bug fix: dynamic variable and thread start related gc lossage
772 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
774 * fixed some bugs revealed by Paul Dietz' test suite:
775 ** SUBTYPEP is slightly more accurate on heinously complicated
776 CONS types where some of the members have uncertain (in the
777 NIL, NIL sense) type relationships to each other.
779 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
780 no longer purified unless :PURIFY T is explicitly specified.
781 ** Non-purified cores are significantly smaller than before
783 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
784 * new feature: timers based on Zach Beane's excellent timer package
785 * added support for the following external formats: koi8-u,
786 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
787 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
788 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
789 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
790 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
791 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
792 * incompatible change: a threaded SBCL will no longer revert to
793 non-threaded mode on non-NPTL systems, but refuse to start entirely.
794 * bug fix: interrupts are disabled until startup is complete; no
795 more sigsegvs when receiving a signal to soon
796 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
798 * bug fix: add a workaround for the memory randomization features in
799 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
800 workaround will only be in effect on systems with the proc filesystem
802 * bug fix: printing objects of type HASH-TABLE signals a
803 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
805 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
806 have been running for over 50 days. (reported by Gilbert Baumann)
807 * bug fix: the logic for getting names of functions gets less
808 confused when confronded with alternate-metaclass
809 funcallable-instances. (reported by Cyrus Harmon)
810 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
811 consistent, even on internal alternate-metaclass objects.
812 * bug fix: flush closure information collected by physical
813 environment analysis prepass before the main pass. (bug reported
815 * bug fix: compiler pack phase does not modify a hashtable, which is
816 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
817 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
818 may be expanded. (reported by Alexander <malishev> on c.l.l)
819 * bug fix: no more style-warnings when compiling code using LAST
820 under high-SPEED low-SPACE settings. (reported by David Wallin
822 * fixed some incompatibilities between SBCL's MOP and the MOP
824 ** the METAOBJECT class is now implemented;
825 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
826 STANDARD-OBJECT, as required;
827 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
828 now compatible (as required by VALIDATE-SUPERCLASS); there
829 remains a consistency requirement over the presence or absence
830 of the FUNCTION class in the superclass list of finalized
831 classes; see the manual for more details;
832 ** the :AROUND method for COMPUTE-SLOTS on
833 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
834 requested slot ordering.
836 ** bug fix: parent thread now can be gc'ed even with a live
838 ** bug fix: binding a special with PROGV to no value is not
839 the same as that symbol not having been bound (thanks to
841 ** bug fix: binding specials is thread safe (thanks to
843 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
844 and SB-SPROF do not die with 'no handler for signal XX in
845 interrupt_handle_now(..)' anymore
846 ** bug fix: WITH-TIMEOUT works with multiple running threads
847 * fixed some bugs revealed by Paul Dietz' test suite:
848 ** ENSURE-GENERIC-FUNCTION should take a method class object for
849 the :method-class keyword argument.
851 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
852 * new port: the Solaris operating system on x86 processors is now
853 mostly supported, though some rough edges in the environment
854 remain. (thanks to Daisuke Homma)
855 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
856 space for its heap. (thanks to Thiemo Seufer)
857 * optimization: SBCL on MIPS platforms now supports dynamic-extent
858 closures. (thanks to Thiemo Seufer)
859 * minor incompatible change: eof selects abort in the debugger.
860 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
861 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
863 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
864 deprecated, and will go away in a future revision of SBCL.
865 * minor incompatible change: GC-ON and GC-OFF are no longer
866 implemented with a counter, it does not matter how many times gc
867 is switched on or off
868 * bug fix: discriminating functions for generic function classes
869 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
870 make invalid assumptions about method precedence order. (reported
872 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
873 platforms. (thanks to Thiemo Seufer)
874 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
875 * bug fix: correct pathname printing with printer escaping is on.
876 (thanks to Kevin Reid)
877 * bug fix: complex VOP definitions in "user-space" no longer trigger
878 package locks. (reported by Zach Beane)
879 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
880 for funcallable-instance functions, and is no different from
882 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
883 characters in the current readtable. (reported by Nicholas Neuss)
884 * bug fix: SUBTYPEP on various CONS types returns more a more
885 accurate acknowledgment of its certainty.
886 * optimizations: REMOVE-DUPLICATES now runs in linear time on
887 lists in some cases. This partially fixes bug 384.
888 * flush all standard streams before prompting in the REPL and the
890 * bug fix: signal handling and triggering gc do not conflict
891 directly anymore, in particular a high frequency sb-sprof does
892 not prevent gc from running
893 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
894 approximation for timezone and DST information between the
895 universal time epoch and the smallest negative 32-bit time_t.
896 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
897 year 1899 so long as the encoded time is non-negative.
898 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
899 the correct answer for non-constant rotate amounts when performing
900 an inline 32-bit rotation.
902 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
903 there is only one thread in the session
904 ** bug fix: memory leak for streams created in one thread and
905 written to in another
906 ** bug fix: lockup when compiled with gcc4
907 ** bug fix: race that allows the gc to be triggered when gc is
909 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
910 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
912 ** bug fix: finished threads are gc'ed properly
913 * fixed some bugs revealed by Paul Dietz' test suite:
914 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
915 has arguments with a different set of applicable methods from
916 the orignal arguments.
917 ** The type error thrown by MAP now has a correct expected-type
919 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
920 name a compiled function.
921 ** Three MISC tests where a large bignum was improperly coerced to
922 a float that couldn't represent that bignum during type
923 derivation were fixed.
924 ** SUBTYPEP can now handle types involving the negation of a
925 list-form FUNCTION type.
926 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
927 as (COMPLEX (AND RATIO (NOT FIXNUM))).
928 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
930 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
931 * New feature: Experimental support for bivalent streams: streams
932 opened with :element-type :default now allow character and binary
933 (unsigned-byte 8) I/O
934 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
935 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
937 * bug fix: no more highly sporadic "couldn't check whether ~S is
938 readable" when reading a stream and an interrupt hits in the middle
939 of a select system call
940 * compiler better recognizes complex arrays (reported by Eduardo
942 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
943 for variables on SBCL built with linkage-tables. (reported by Luis
945 * various error reporting improvements.
946 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
947 (Thanks to Thiemo Seufer)
948 * bug fix: nested reader invokations work correctly
949 * bug fix: it is possible to have simultaneous references to foreign
950 code and foreign data with the same name.
952 ** added x86-64 support
953 ** incompatible change: the threading api now works with thread
954 objects instead of thread ids
955 ** bug fix: threads are protected from signals and interruption when
956 starting up or going down
957 ** bug fix: a race where an exiting thread could lose its stack to gc
958 ** bug fix: don't halt on infinite error in threads if possible
959 ** fixed numerous gc deadlocks introduced in the pthread merge
960 ** bug fix: fixed thread safety issues in read and print
961 ** bug fix: debugger doesn't hang on session lock if interrupted at
962 an inappropriate moment
963 ** bug fix: run-program is now thread safe(r)
964 ** bug fix: inner with-recursive-lock no longer releases the mutex
965 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
966 * fixed some bugs revealed by Paul Dietz' test suite:
967 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
968 EXPECTED-TYPE slots filled.
969 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
971 ** MAKE-CONDITION accepts classes as type-designators.
972 ** COMPILE may never return NIL.
973 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
974 range before calling Unix time functions
976 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
977 * numerous signal handling fixes to increase stability
978 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
979 * minor incompatible change: we now correctly canonize default
980 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
981 as per the MOP, rather than the historical (INITARG INITFUNCTION
982 INITFORM). (reported by Bruno Haible)
983 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
984 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
985 for more information.
986 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
987 pathname is a directory pathname.
988 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
989 * dynamic space size on PPC has been increased to 768Mb. (thanks to
991 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
992 :METACLASS argument in addition to a class name. (reported by
993 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
994 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
995 (reported by Bruno Haible; patch from Gerd Moellmann)
996 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
998 * bug fix: more cleanups to the floating point exception handling on
999 x86-64 (thanks to James Knight)
1000 * bug fix: the compiler does not try to constant fold calls of
1001 COERCE and BIT-* functions when they return freshly constructed
1002 objects. (reported by Nikodemus Siivola and Paul Dietz)
1003 * optimization: improved the MIPS versions of generic (in the
1004 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1005 * optimization: direct conversion of (unsigned-byte 32) to floats on
1006 the PowerPC platform.
1007 * optimization: structure instances with raw slots now use less
1008 memory, and probably show better memory locality. (thanks to
1010 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1011 are treated as requests for stack allocation on the x86-64,
1012 Alpha-32, PPC and SPARC platforms.
1013 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1014 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1015 (thanks to Tony Martinez)
1017 ** gcing a dead thread can no longer lead to lockups
1018 ** threads block signals until they are set up properly
1019 ** errno is no longer shared by threads
1020 ** interrupt-thread restores the eflags register on x86
1021 ** fixed some lockups due to gc/thread interaction
1022 * fixed some bugs revealed by Paul Dietz' test suite:
1023 ** invalid dotted lists no longer raise a read error when
1024 *READ-SUPPRESS* is T
1025 ** signal an error if a symbol that names a declaration is used
1026 as the name of a type, or vice versa
1027 ** allow using the (declare (typespec var*)) abbreviation for
1028 (declare (type typespec var*)) with all type specifiers
1029 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1030 ** always return NIL from PPRINT-POP when OBJECT is NIL
1031 ** don't signal errors when pretty-printing malformed LABELS,
1032 FLET or MACROLET forms
1033 ** declarations in a DOLIST body are also in scope for the
1035 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1037 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1040 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1041 * fixed cross-compiler leakages that prevented building a 32-bit
1042 target with a 64-bit host compiler.
1043 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1044 opened with :IF-EXISTS :OVERWRITE.
1045 * fixed bug 281: error for an invalid qualifier in a short-form method
1046 combination method is not signalled until the faulty method is called.
1047 * bug fix: iteration variable type inferrer failed to deal with open
1048 intervals. (reported by Alan Shields)
1049 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1050 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1051 or more runtime options were provided to the sbcl binary.
1052 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1053 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1054 * optimization: there's now a fast-path for fixnum arguments in the
1055 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1056 * optimization: the code generated on x86-64 is more compact thanks
1057 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1058 * bug fix: floating-point exception handling now partly works on
1059 x86-64. (Thanks to James Knight)
1060 * improvement to the MIPS backend from Thiemo Seufer: C-style
1061 64-bit long long arguments and return values to alien functions
1063 * fixed some bugs revealed by Paul Dietz' test suite:
1064 ** the type-error signalled from WARN has a filled-in DATUM slot.
1065 ** the type-error required when a stream is not associated with
1066 a file has the stream as its datum.
1067 ** type-errors on single-floats on x86-64 no longer have
1068 :INVALID-OBJECT as the datum
1069 ** the type-errors signalled for invalid function names now have
1070 a correct expected type
1071 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1072 for structures defined with a :TYPE. Documentation strings for
1073 typed structures are no longer immediately discarded
1074 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1075 broadcast streams and synonym streams. FILE-LENGTH now also works
1076 on broadcast streams.
1078 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1079 * incompatible change: the --noprogrammer option, deprecated since
1080 version 0.7.5, has been removed. Please use the equivalent
1081 --disable-debugger option instead.
1082 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1084 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1085 inoperational for a while now) has been completely removed.
1086 * null lexical environments are now printed as #<NULL-LEXENV>,
1087 significantly reducing the amount of clutter in typical backtraces.
1088 * documentation on weak pointers, finalization, and after GC hooks
1089 has been added to the manual.
1090 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1091 as other specialized array types, is much faster. SUBSEQ and
1092 COPY-SEQ on such arrays have also been sped up.
1093 * optimization: EQL is now more efficient when at least other argument
1094 is known to be of type (OR FIXNUM (NOT NUMBER)).
1095 * fixed bug: the runtime is now less vulnerable to changes in the
1096 size of the SBCL object on OS X, and virtual memory is reserved for
1097 all spaces that need to be at a fixed address.
1098 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1099 * fixed bug: finalizers and after GC hooks that cause consing are now
1101 * fixed bug: compiler error messages and summaries are now printed to
1102 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1103 * fixed inference of the upper bound of an iteration variable.
1104 (reported by Rajat Datta).
1105 * fixed bug 211e: calling local functions with duplicated constant
1106 keyword argument no longer causes a bogus style warning about an
1108 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1109 ftype declaration to be disregarded. (reported by Dave Roberts)
1110 * fixed bug 373: caused by erronous compilation of references to alien
1111 variables in the runtime on ppc/darwin.
1112 * fixed bug 376: CONJUGATE type deriver.
1113 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1114 (reported by Baughn on #lisp)
1115 * a cross-compiler bug on non-x86 platforms has been identified and
1116 fixed. (thanks to Bruno Haible)
1117 * improvements to the MIPS runtime code for increased stability.
1118 (thanks to Thiemo Seufer)
1119 * increased the maximimum compact environment size to allow
1120 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1121 * partial workaround for bug 135: don't name the function we're
1122 calling for hairy cases of EVAL, so as not to accumulate one environment
1123 entry per call to EVAL. (reported by Kevin Reid)
1124 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1125 * optimization: type testing for non-vector arrays should be faster.
1126 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1128 * bug fix: setting 31st element of a bit vector to zero did not work
1130 * bug fix: redefining a class definition which failed due to a
1131 previous accessor / function clash now works (but see BUGS entry
1132 #380 for more problems in this area). (thanks to Zach Beane)
1133 * the long form of DEFINE-METHOD-COMBINATION disables method group
1134 checking when given a single method group with pattern *. (thanks
1136 * on x86 compiler supports stack allocation of results of simple
1137 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1138 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1139 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1140 * fixed some bugs related to Unicode integration:
1141 ** the restarts for recovering from input and output encoding
1142 errors only appear when there is in fact such an error to
1144 * fixed some bugs revealed by Paul Dietz' test suite:
1145 ** MISC.549 and similar: late transformation of unsafe type
1146 assertions into derived types caused unexpected code
1148 ** SCALE-FLOAT type deriver is less wrong.
1149 ** type derivers for EXP, LOG and similar functions compute result
1150 types for complex arguments better.
1151 ** (MISC.563) CONJUGATE type deriver works for very restricted
1153 ** out-of-line type testers for character strings are available.
1154 ** EQUAL compiler transform understands specialness of objects
1156 ** accessing double-floats stored on the stack now works on x86-64.
1157 ** debugger internals could sometimes create invalid lispobjs,
1158 resulting in GC crashes.
1159 ** MISC.548: type check weakening can convert required type into
1161 ** initialization forms for bindings are not in scope of free special
1164 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1165 * incompatible change: thread support for non-NPTL systems has
1166 been removed - locking is buggy and unreliable. A threaded
1167 SBCL build will now warn at startup and refuse to create
1168 new threads, unless futex support is detected in the kernel
1169 * incompatible change: the top level REPL now has only an ABORT
1170 restart associated with it, not TOPLEVEL and ABORT as it used to.
1171 TOP and TOPLEVEL are now available as debugger commands for
1172 returning to the top level.
1173 * incompatible change: forms evaluated in the REPL now use the
1174 global optimization policy.
1175 * incompatible change: user- and system-initialization files are
1176 no longer processed with LOAD, but by READ and EVAL; hence the
1177 global optimization policy, startup package, readtable, etc,
1179 * The .fasl file format number has been incremented because of
1180 various incompatible changes.
1181 * internal entry point details and argument counts no longer appear
1182 in backtraces unless explicitly requested by setting
1183 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1184 * built-in and standard functions no longer have names like "top
1185 level local call to FOO".
1186 * fixed bug 32: functions defined in non-null lexical environments
1187 now have more legible printed representation
1188 * fixed bug 33: functions defined in non-null lexical environemnts
1189 are now more amenable to inspection by INSPECT.
1190 * workaround for bug 354: XEPs no longer appear in backtraces unless
1191 explicitly requested.
1192 * fixed bug: receiving the signal which results from attempting to
1193 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1194 SIGBUS on other *BSDs) on architectures where the C stack is also
1195 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1196 to debugger confusion or wild instability. (reported by Cheuksan
1198 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1199 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1200 (reported by Lutz Euler)
1201 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1202 compiled code. (reported by Paul Dietz)
1203 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1204 specialized parameter to a method to something that is not TYPEP
1205 the specializer is now possible.
1206 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1207 face of package deletion.
1208 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1209 pathname. (reported by David Tolpin on c.l.l)
1210 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1211 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1212 STRING-STREAM and FILE-STREAM, respectively.
1213 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1214 than previously; in addition, its test suite is now run on build.
1215 (thanks to Luke Gorrie)
1216 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1217 * optimization: added a immediate representation for single-floats
1219 * fixed some bugs revealed by Paul Dietz' test suite:
1220 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1221 ** attempting to create a package with a colliding nickname causes
1222 correctable errors to be signalled.
1223 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1224 ** several x86-64 backend bugs related to sign-extension of immediate
1227 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1228 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1229 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1231 * fixed loading of multiply forward-referenced layouts.
1232 (thanks to Cheuksan Wang)
1233 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1234 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1235 (see COMPILE-FILE documentation for details of :PRINT option)
1236 * fixed bugs 19 and 317: fixed-format floating point printing is
1237 more accurate. This also fixes a bug reported by Adam Warner
1238 related to the ~@F format directive.
1239 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1241 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1242 dispatch table if the from-char is a dispatch macro character.
1243 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1244 by the wordsize no longer ignore the last word. (reported by Lutz
1246 * fixed bug in type checking of dynamic-extent variables. (reported
1248 * optimization: sequence traversal functions use their freedom to
1249 coerce function designators to functions.
1250 * optimization: code with many calls to CLOS methods specialized on
1251 CLOS classes has had redundant type checks removed.
1252 * optimization: added declarations to speed up operations that access
1253 the internal character database (for example STRING-UPCASE)
1254 * optimization: comparison operations between floats and sufficiently small
1255 fixnums no longer create extra rationals
1256 * fixed some bugs related to Unicode integration:
1257 ** portions of multibyte characters at the end of buffers for
1258 character-based file input are correctly transferred to the
1259 start of the buffer at the next read.
1260 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1261 passing it through to OPEN.
1262 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1263 argument given to internal calls to OPEN.
1264 ** the built-in comment readers (introduced by character sequences
1265 ";" and "#|") are more forgiving to encoding errors; they will
1266 STYLE-WARN and then attempt to resync the stream at a character
1267 boundary. (thanks to Teemu Kalvas)
1268 * fixed some bugs in the x86-64 port:
1269 ** Negative short int return values from c-calls are sign-extended
1271 ** The stack is aligned to 16-bytes for c-calls, as required by
1272 the ABI. (thanks to Cheuksan Wang)
1273 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1274 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1275 for enums. (thanks to Vincent Arkesteijn)
1276 ** Multiple small optimizations and bugfixes for floating point
1278 * fixed some bugs revealed by Paul Dietz' test suite:
1279 ** Space, Tab, Linefeed, Return and Page have the invalid
1280 secondary constituent character trait.
1281 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1283 ** WITH-INPUT-FROM-STRING should only update the index place on
1285 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1286 works more reliably.
1287 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1288 with one-dimensional arrays with fill pointers.
1289 ** TYPE-OF failed on a complex with an integer realpart and a
1291 ** compiler failure during type inference for the code of form
1292 (IF (EQL X (THE ...)) ...) (MISC.535).
1294 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1295 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1296 platform. The port supports 61-bit fixnums, large memory spaces
1297 and reloading shared object files.
1298 * enhancement: saving cores with foreign code loaded is now
1299 supported on ppc/Darwin in addition to the previously supported
1301 * enhancement: the statistical profiler now walks deeper into the
1302 call stack for more meaningful call-graphs and accrued time
1303 reports (x86/x86-64 only). It also now reports time spent in
1305 * enhancement: it is now possible to trace most individual methods
1306 of a generic function in addition to tracing the generic function
1308 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1310 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1311 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1312 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1313 produces an error. (thanks to Vincent Arkesteijn)
1314 * bug fix: NAMESTRING on pathnames with :WILD components in their
1315 directories works correctly. (thanks to Artem V. Andreev)
1316 * fixed bug 125: compiler preserves identity of closures. (reported
1318 * bug fixed: functions with &REST arguments sometimes failed with
1319 "Undefined function" when compiled with (DEBUG 3). (reported by
1321 * bug fix: overflow during compiling of setting element of a bit
1322 vector with constant index and value. (reported by Timmy Douglas)
1323 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1324 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1325 * on x86 compiler partially supports stack allocation of dynamic-extent
1327 * GO and RETURN-FROM do not check the extent of their exit points
1328 when compiled with SAFETY 0.
1329 * fixed some bugs related to Unicode integration:
1330 ** encoding and decoding errors are now much more robustly
1331 handled; it should now be possible to recover even from invalid
1332 input or output to the terminal. (thanks to Teemu Kalvas)
1333 ** provided a first cut at implementing STRING-TO-OCTETS and
1334 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1335 ** altered the SB-MD5 contributed module slightly, changing the
1336 interface just enough for it to be supportable for builds where
1337 lisp characters are not eight bits.
1338 * fixed some bugs revealed by Paul Dietz' test suite:
1339 ** the FORMATTER-generated functions for ~V[ conditionals require
1340 the correct number of arguments.
1341 ** READ-FROM-STRING returns the mandated second value when applied
1342 to displaced strings.
1343 ** the #\Rubout and #\Backspace characters are treated as invalid
1344 constituent characters by the tokenizer.
1346 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1347 * new feature: reloading changed shared object files with
1348 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1349 * new feature: references to foreign variables and functions
1350 can now be compiled and loaded before the corresponding shared
1351 object file is loaded, as long as the foreign definitions are
1352 available at runtime.
1353 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1354 supported. (thanks to Dan Debertin)
1355 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1356 just keywords. (thanks to Vincent Arkesteijn)
1357 * fixed bug #331: structure-class instances corresponding to
1358 DEFSTRUCT forms are now created eagerly.
1359 * fixed bug #345: backtraces from calls to undefined functions work
1360 on x86 as well. Related bug #61 is now also partially fixed on x86
1361 (backtraces from throws to unknown catch tags.)
1362 * bug fix: lambda-list parsing is now stricter vrt. order and number
1363 of lambda-list keywords.
1364 * bug fix: as specified by AMOP, an error is signalled if a
1365 class-option appears multiple times in a DEFCLASS form. (reported
1367 * bug fix: RANDOM can be compiled when the compiler derives the type
1368 of its numeric argument as a disjoint set of small integers.
1369 (reported by Paul Dietz)
1370 * bug fix: starting a core saved with shared objects loaded when
1371 those objects are not available no longer causes threaded SBCL to
1372 hang. (reported by Sean Ross)
1373 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1375 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1376 argument, which must always be NIL. (reported by Kalle Niemitalo)
1377 * bug fix: printing 1.0d+23 no longer results in an error.
1378 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1379 * bug fix: structure slot setters preserve evaluation order. (thanks
1381 * fixed some bugs related to Unicode integration:
1382 ** RUN-PROGRAM can allow its child to take input from a Lisp
1383 stream. (reported by Stefan Scholl)
1384 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1385 locales. (reported by Ken Causey)
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** INCF, DECF and REMF evaluate their place form as specified in
1389 ** FORMATTER expands ~{ ~} iteration directives with V or #
1390 parameters correctly.
1391 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1392 consequent uses no arguments correctly.
1393 ** the system has a partial understanding of the (COMPLEX RATIO)
1395 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1398 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1399 * new feature: a build-time option (controlled by the :SB-UNICODE
1400 keyword feature, enabled by default) for building the system with
1401 support for the entire 21-bit character space defined by the
1403 * new feature: the system now has rudimentary external-format
1404 support; the primary user-visible change at this time is that
1405 characters with the high bit set (such as lower-case-e-acute) will
1406 print correctly to a terminal in a UTF-8 environment.
1407 * minor incompatible change: BASE-CHAR no longer names a class;
1408 however, CHARACTER continues to do so, as required by ANSI.
1409 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1410 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1411 has been moved to the SB-EXT package (temporarily re-exported from
1413 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1414 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1416 * fixed bug #351: better error-handlind and reporting for malformed
1418 * fixed bug #350: bignum-printing is now more memory-efficient,
1419 allowing printing of very large bignums, eg. (expt 2 10000000).
1420 (reported by Bruno Haible)
1421 * fixed bug #302: better primitive-type selection for intersection
1423 * fixed bug #308: non-graphic characters now all have names, as
1424 required. (reported by Bruno Haible)
1425 * bug fix: redefining a class with different superclasses now correctly
1426 removes it from the direct-subclasses of its previous superclasses.
1427 (reported by David Morse)
1428 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1429 new value now works. (reported by Bruno Haible)
1430 * bug fix: correct canonicalization of multiple non-standard slot
1431 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1432 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1433 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1434 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1435 now exists, an signals an error.
1436 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1437 during macroexpansion for non-iteration variables as well. (reported
1438 by Bruno Haible for CMUCL)
1439 * bug fix: Cyclic structures and unprintable objects in compiler
1440 messages no longer cause errors. (reported by Bruno Haible)
1441 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1442 and READ-FROM-STRING all now return a primary value of NIL if
1443 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1444 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1445 (reported by Bruno Haible for CMUCL)
1446 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1447 do not designate a condition. (reported by Bruno Haible for
1449 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1450 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1451 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1452 permitting resolution in favour of any of the conflicting symbols.
1453 (reported by Bruno Haible for CMUCL)
1454 * bug fix: EQUAL compiler optimizations is less aggressive on
1455 strings which can potentially compare true despite having distinct
1456 specialized array element types.
1457 * bug fix: unit enumerations can be defined without dividing by
1458 zero. (thanks to Vincent Arkesteijn)
1459 * FORMAT compile-time argument count checking has been enhanced.
1460 (report from Bruno Haible for CMUCL)
1461 * a partial workaround for the bug 262: the compiler does not try to
1462 inline-expand a local function doing RETURN-FROM from a deleted
1463 BLOCK. (thanks to Peter Denno for the bug report and to David
1464 Wragg for the simple test case)
1465 * fixed some bugs revealed by Paul Dietz' test suite:
1466 ** RENAME-PACKAGE allows all package designators as new package
1468 ** constraint propagation and lambda variable substitution are
1469 more cautious in dealing with partially deleted code.
1470 ** compiler sometimes lost reoptimization passes.
1471 ** CERROR, given a condition as condition designator, passes its
1472 remaining arguments to the continue format control without
1474 ** Case-altering FORMAT directives work correctly on non-ASCII
1476 ** The REAL type specifier handles bounds outside the floating
1477 point ranges without signalling FLOATING-POINT-OVERFLOW.
1478 ** Functions with IR1-transformations can create intercomponent
1479 references to global functions.
1480 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1482 ** FORMAT ~R treats a nil value for its first parameter correctly.
1484 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1485 * enhancement: saving cores with foreign code loaded is now
1486 supported on x86/NetBSD and sparc/Linux in addition to the previously
1487 supported platforms.
1488 * bug fix: on some platforms repeated installations caused multiple
1489 copies of HTML documentation to be installed -- should not happen
1490 any more. (reported by Stefan Scholl)
1491 * bug fix: parsing self-recursive alien record types multiple times
1492 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1493 original patch by Helmut Eller for CMUCL)
1494 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1495 (thanks to Richard Kreuter)
1496 * bug fix: defining classes whose accessors are methods on existing
1497 generic functions in other (locked) packages no longer signals
1498 bogus package lock violations. (reported by François-René Rideau)
1499 * bug fix: special variables as DEFMETHOD parameters no longer have
1500 associated bogus type declarations. (reported by David Wragg and
1502 * bug fix: read-write consistency on streams of element-type
1503 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1504 * bug fix: redefiniton of the only method of a generic function with
1505 no DEFGENERIC no longer emits a full WARNING. In addition,
1506 redefinition of generic functions with no DEFGENERIC to an
1507 incompatible lambda list now signals an error. (thanks to Zach
1509 * bug fix: DEFGENERIC now works even when there's a function of the
1510 same name in an enclosing lexical environment. (thanks to Zach
1512 * fixed compiler failure, caused by instrumenting code during
1513 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1514 * optimization: added loop analysis and improved register allocation
1515 * fixed some bugs revealed by Paul Dietz' test suite:
1516 ** POSITION on displaced vectors with non-zero displacement
1517 returns the right answer.
1518 ** (SIMPLE-STRING) is a valid type specifier for sequence
1520 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1522 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1523 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1525 ** The justification version of the FORMAT ~< directive treats
1526 non-zero minpad parameter correctly.
1528 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1529 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1530 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1531 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1532 the supported interface.
1533 * new feature: Single-stepping of code compiled with DEBUG 2 or
1534 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1535 * new feature: saving cores with foreign code loaded is now
1536 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1537 Timothy Moore's work for CMUCL)
1538 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1539 parameters to * instead of NIL if no initform is supplied.
1540 (reported by Johan Bockgaard)
1541 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1542 correctly when FUNCALL appears as the car of the form. Note:
1543 despite this FUNCALL forms are not currently subject to
1544 compiler-macro expansion. (port of Raymond Toy's fix for the
1545 same from CMUCL, reported by Johan Bockgaard)
1546 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1547 (thanks for Teemu Kalvas)
1548 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1549 case instead of terminating the iteration (thanks for Julian
1550 Squires, Sean Champ and Raymond Toy)
1551 * bug fix: incorrect expansion of defgeneric that caused a style
1552 warning. (thanks for Zach Beane)
1553 * bug fix: slot accessor effective method computation works properly
1554 for classes with multiple non-standard applicable methods on
1555 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1556 * on x86 compiler supports stack allocation of results of LIST and
1557 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1558 CMUCL implementation by Gerd Moellmann)
1559 * fixed some bugs revealed by Paul Dietz' test suite:
1560 ** FORMAT strings with both the ~<~:;~> form of the justification
1561 directive and pretty-printing directives cause an error.
1563 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1564 * incompatible change: the internal functions
1565 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1566 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1567 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1568 instead of the old functions.
1569 * new feature: on platforms where "dladdr" is available foreign
1570 function names now appear in backtraces. (based on Helmut Eller's
1572 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1573 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1575 * documentation: ASDF manual is now distributed with SBCL.
1576 * bug fix: SBCL can now load its contributed modules with REQUIRE
1577 even if the system-provided entries have been removed from
1578 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1580 * bug fix: backtraces involving undefined functions or assembly
1581 routines are more informative. (thanks to Brian Downing)
1582 * bug fix: mutually referent alien structures now work correctly.
1583 (reported by Rick Taube)
1584 * bug fix: structures defined by WITH-ALIEN can be referred to
1585 within other definitions in the same WITH-ALIEN.
1586 * bug fix: division operators (MOD, TRUNCATE and the like) with
1587 constant zero divisors and integer dividends no longer generate
1589 * bug fix: provide default methods for INPUT-STREAM-P and
1590 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1591 * bug fix: improve the detection and resolution of MOP metacycles
1592 (where effective method computation depends on the generic function
1593 having its effective method computed). (reported by Bruno Haible)
1594 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1595 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1596 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1597 * bug fix: pathnames with relative directory components can be
1598 represented relative to default pathnames.
1599 * optimization: in taking the GCD of bignums, reduce the two bignums
1600 to approximately the same size (using Euclid's algorithm) before
1601 applying the more sophisticated binary GCD. (thanks to Juho
1603 * optimization: COUNT on bitvectors now operates word-at-a-time.
1604 * optimization: ASH with a positive, but not necessarily constant,
1605 (leftwards) shift, when in a modular context, is compiled to a
1607 * fixed some bugs revealed by Paul Dietz' test suite:
1608 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1609 if the corresponding argument is NIL.
1610 ** FORMAT directives accepting a minpad parameter treat negative
1612 ** literal commas as character parameters to FORMAT directives are
1614 ** literal spaces directly after ~<Newline> directives within a
1615 format-logical-block (~:< ~@:>) do not induce :FILL-style
1616 conditional newlines.
1617 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1618 as specified: it no longer includes conditional newlines).
1619 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1621 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1622 :PER-LINE-PREFIX argument does not evaluate to a string.
1624 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1625 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1626 the manual for details; for now, package locks can be disabled by
1627 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1628 if no major problems are found then it is likely that they will be
1629 compiled in unconditionally.
1630 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1631 unsupported operators on all platforms. To load a shared library
1632 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1633 object file, link it into a shared library outside of SBCL and
1634 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1636 * bug workaround: ROOM T and the GC were not working together
1637 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1638 reporting facility aren't preserved. That reporting has been
1639 disabled, so now until and unless someone figures out how to make it
1640 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1641 * minor incompatible change: as threatened around sbcl-0.8.0, the
1642 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1643 The symbols are also exported from SB-PCL for backwards
1644 compatibility, but more so than before SB-PCL should be treated as
1645 an implementation-internal package.
1646 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1648 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1649 It is known as of this release not to work on the Alpha, however.
1650 * fixed bug #167: errors signalled due to illegal syntax in method
1651 bodies are now more legible.
1652 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1653 designators and can hence be used with TYPEP.
1654 * fixed bug #333: CHECK-TYPE now ensures that the type error
1655 signalled, if any, has the right object to be accessed by
1656 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1657 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1658 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1660 * fixed bug #334: programmatic addition of slots using specialized
1661 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1662 and :ALLOCATION :CLASS slots.
1663 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1664 float exponents. (rereported by Peter Seibel)
1665 * fixed a bug: #\Space (and other whitespace characters) are no
1666 longer considered to be macro characters in standard syntax by
1667 GET-MACRO-CHARACTER.
1668 * fixed bug: initialization of condition class metaobjects no longer
1669 causes an instance of the condition to be created. (reported by
1671 * fixed bug: it is now possible to have slots such that
1672 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1673 description is neither :INSTANCE nor :CLASS.
1674 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1675 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1676 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1677 non-standard SLOT-DEFINITION-ALLOCATION values.
1678 * fixed bugs: various race conditions handling exiting threads.
1679 CL-PPCRE's thread torture test now passes.
1680 * fixed arguably-a-bug: GC time proportional to number of threads in
1681 system even when most of them are idle
1682 * optimization: improved performance of BIT and SBIT on bit-vectors.
1683 * .fasl file incompatibility: The fasl file version number has
1684 been incremented because of changes associated with package locks.
1686 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1687 * minor incompatible change: the system no longer provides
1688 optimization hints (conditions which are TYPEP
1689 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1690 mode; these hints are still emitted when higher SPEED optimization
1692 * new contrib module: a sampling profiler (profiling by statistical
1693 sampling, rather than by instrumenting functions) is available as
1694 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1695 * the behaviour of the standard function ED is now customizeable by
1696 third parties through a hook variable: see ED's documentation
1697 string for information on the protocol.
1698 * the compiler no longer emits efficiency notes for (FUNCALL X)
1699 when the type of X is uncertain under default optimization
1701 * fixed bug 276: mutating a binding of a specialized parameter to a
1702 method to something that is not TYPEP the specializer is now
1704 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1705 exhibit the required behaviour on the x86 platform. (thanks to
1706 Peter van Eynde, Eric Marsden and Bruno Haible)
1707 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1708 even for difficult arguments. (reported by Peter Graves)
1709 * fixed bug 141a: the backquote printer now descends quoted
1711 * fixed another bug in backquote printing: no more destructive
1712 modification of the form's list structure. (reported by Brian
1714 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1715 move between its address being taken and the call to
1716 interrupt_thread, fixing a crashing race condition.
1717 * the SB-POSIX contrib implementation has been adjusted so that it
1718 no longer exhibits ridiculously poor performance when constructing
1719 instances corresponding to C structs.
1721 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1722 * minor incompatible change: the sb-grovel contrib now treats C
1723 structures as alien (in the sense of SB-ALIEN) objects rather than
1724 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1725 has implications for memory management of client code
1726 (sb-grovel-returned objects must now be manually managed) and for
1727 type safety (alien objects now have full types).
1728 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1729 used to control emission of compiler diagnostics, rather than the
1730 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1731 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1732 quality should be considered deprecated.
1733 * install.sh now installs the user manual as well
1734 * (not quite a new documentable feature, but worth considering in
1735 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1736 beginnings of a semantically meaningful condition hierarchy is
1737 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1739 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1740 designator as the defaults argument.
1741 * fixed bug: Displaced arrays whose displaced-to array has become
1742 too small now cause ARRAY-DIMENSION to signal an error, providing
1743 for safer bounds-checking. (reported by Bruno Haible)
1744 * fixed bug: DEFCLASS slot definitions with identical :READER and
1745 :WRITER names now signal a reasonable error. (reported by Thomas
1747 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1749 * fixed bug: Invalid slot specification errors now print correctly.
1750 (thanks to Zach Beane)
1751 * fixed bug 320: Shared to local slot value transfers in class
1752 redefinitions now happen corrently from superclasses as
1753 well. (reported by Bruno Haible)
1754 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1756 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1757 improper lists correctly. (reported by Bruno Haible)
1758 * fixed bug 313: source-transform for <fun-name> was erroneously
1759 applied to a call of a value of a variable with name <fun-name>.
1760 (reported by Antonio Menezes Leitao)
1761 * fixed bug 307: The obsolete instance protocol ensures that
1762 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1763 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1764 remove dead unknown-values globs from the middle of the stack.
1765 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1766 easier to resume long computations after using *BREAK-ON-SIGNALS*
1767 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1768 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1769 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1770 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1771 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1772 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1774 * fixed a (fairly theoretical) bug in string printing: if
1775 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1776 does not have array-element-type equal to the most general string
1778 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1779 function name. (thanks to Nikodemus Siivola)
1780 * fixed bug: compile-time format string checker failed on
1781 non-closed ~{. (reported by Thomas F Burdick)
1782 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1783 don't warn on legal loop code involving more than one aggregate
1784 boolean. (thanks to Nikodemus Siivola)
1785 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1786 in keyword position to MAKE-PACKAGE was not being reported as
1788 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1789 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1791 * fixed bugs 280 and 312: the checking for multiple definitions in a
1792 file is less likely to become confused by uses of inline
1794 * fixed bug: the #S reader macro performs the keyword coercion
1795 specified for slot names. (reported by Kalle Niemitalo)
1796 * fixed bug: lambda lists may contain symbols whose names start with
1797 & but are not lambda-list-keywords; their occurrence triggers a
1799 * fixed bug 321: define-method-combination argument lambda lists do
1800 not cause invalid code to be generated when &optional or &aux
1801 variables have default values. (reported by Bruno Haible)
1802 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1803 have CLOS classes; this appears also to have fixed reported
1804 instability in stack exhaustion detection.
1805 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1806 MUFFLE-WARNING when no associated restart is present is now
1808 * optimization: rearranged the expansion of various defining macros
1809 so that each expands into only one top-level form in a
1810 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1812 * optimization: used a previously little-used slot in symbols to
1813 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1815 * fixed some bugs revealed by Paul Dietz' test suite:
1816 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1818 ** Fixed an optimization bug related to inheritance of initforms
1819 from local to shared slots.
1820 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1821 ** CAST optimizer forgot to flush argument derived type.
1822 ** print/read consistency on floats is now orders of magnitude
1823 more likely. (thanks also to Bruno Haible for a similar report
1825 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1826 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1827 for objects of type REAL. Make it so.
1828 ** SXHASH is specified (infelicitously) to respect similarity,
1829 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1830 it so. (thanks to Markus Ziegler)
1831 ** on the Alpha, the compiler succeeds in compiling functions
1832 returning a known number of arguments greater than 63.
1833 ** fixed handling of invalid NIL arguments in keyword position
1835 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1837 ** provided a readably-printable representation for RANDOM-STATE
1839 ** ensured that pathnames, where they have a namestring, always
1840 print using #P"..." syntax.
1842 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1843 * Support for the forthcoming 2.0 version of the NetBSD kernel
1844 running on x86 hardware has been added. (thanks to Perry
1845 E. Metzger most immediately, and others for their past work)
1846 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1847 current release version; SBCL's OpenBSD support had been broken
1848 since about the time of OpenBSD's switch to ELF binary format.)
1849 * [placeholder for DX summary]
1850 ** user code with &REST lists declared dynamic-extent, under high
1851 speed or space and low safety and debug optimization policy.
1852 * The manual has been converted to Texinfo format and the debugger
1853 chapter from the cmucl manual has been added.
1854 * A facility has been added to extract documentation strings from
1855 sbcl and store them as Texinfo-formatted snippets for inclusion in
1856 the manual (via Texinfo's @include directive)
1857 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1858 the test case to Dave Roberts)
1859 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1860 pointers. (reported by Sean Ross)
1861 * bug fix: PROFILE output is printed nicely even for large numerical
1862 values. (thanks to Zach Beane)
1863 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1864 greater than 32 handle EOF correctly.
1865 * bug fix: on X86 an immediate argument of the IMUL instruction is
1866 correctly printed in disassembly. (thanks to Lutz Euler)
1867 * bug fix: class slots in redefined classes preserve their old
1868 values. (thanks to Bruno Haible and Nikodemus Siivola)
1869 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1870 succeeds. (reported by Marco Baringer)
1871 * fixed some bugs revealed by Paul Dietz' test suite:
1872 ** READ-SEQUENCE now works on ECHO-STREAMs.
1873 ** RATIONALIZE works more according to its specification. (thanks
1876 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1877 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1878 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1879 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1880 should matter to you only if you rebind the printer control
1881 variables and then find you want different bindings in the
1882 debugger than in the ordinary execution of your program.)
1883 * The runtime build system has been tweaked to support building
1884 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1885 assembler and linker. (thanks to Nikodemus Siivola)
1886 * Unbound, undefined, undeclared variables now trigger full
1887 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1888 more often programmer error than deliberate exploitation of undefined
1890 * optimization: the hash algorithm for strings has changed to one
1891 that is less vulnerable to spurious collisions. (thanks to Juho
1893 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1894 needless bounds checking. (thanks to Juho Snellman)
1895 * optimization: implemented multiplication as a modular
1896 (UNSIGNED-BYTE 32) operation on the PPC backend.
1897 * fixed some bugs revealed by Paul Dietz' test suite:
1898 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1900 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1901 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1902 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1904 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1906 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1908 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1909 their output stream on EOF from read.
1910 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1911 have been read to end-of-file.
1912 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1914 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1915 description of determination of which consecutive characters
1917 ** Printing the "Space" character with escaping on now yields "#\\ ",
1918 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1919 ** Reading floating-point numbers with *READ-BASE* set to a number
1920 less than 10 works correctly.
1921 ** Reading floating-point numbers with *READ-BASE* set to a number
1922 more than 10 works correctly.
1923 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1924 the readtable currently in effect.
1926 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1927 * minor incompatible change: parsing of namestrings on a physical
1928 (Unix) host has changed; numbers after the final #\. in a
1929 namestring are no longer interpreted as a version field. This is
1930 intented to be largely invisible to the user, except that the
1931 meaning of the namestring "*.*.*" has changed: it now refers to a
1932 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1933 should usually be replaced by
1934 (make-pathname :name :wild :type :wild :version :wild)
1935 with the added benefit that this is more likely to be portable.
1936 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1937 option to OPEN now signals an error if the file being opened
1938 exists; this may have an impact on existing code.
1939 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1940 signals received, on the PowerPC platforms (both Linux and
1941 Darwin). (thanks to Pierre Mai for pointing out the location of
1943 * several fixes on OS X: The system now builds and runs cleanly on
1944 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1945 (thanks to Brian Mastenbrook)
1946 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1947 second-resolution: integer multiples of 1/3600 between -24 and 24.
1948 (thanks to Vincent Arkesteijn)
1949 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1950 type when called with 1 argument; PEEK-CHAR checked type of
1951 PEEK-TYPE only after having read first character from a
1952 stream. (reported by Peter Graves)
1953 * bug fix: the garbage collector now has much better locality
1954 behaviour, and in particular no longer treats all memory as being
1955 exhausted when many small objects point to each other in a deeply
1957 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1958 recognized as being TYPEP their class.
1959 * bug fix: the PUSHNEW documentation string has been corrected.
1960 (thanks to Vincent Arkesteijn)
1961 * bug fix: defaulting of the value for the last of an atypically
1962 large number of multiple values being bound was not being
1963 performed correctly on the Alpha or PPC platforms
1964 * optimization: implemented multiplication as a modular
1965 (UNSIGNED-BYTE 32) operation on the x86 backend.
1966 * optimization: SEARCH on simple-base-strings can now be open-coded.
1967 (see also contrib/compiler-extras.lisp for inspiration for
1968 teaching the compiler about the Boyer-Moore algorithm).
1969 * value, returned by MAX (and MIN) called with several EQUALP, but
1970 not EQL, arguments now does not depend on compiler settings.
1971 * fixed some bugs revealed by Paul Dietz' test suite:
1972 ** in stack analysis liveness information is propagated from
1973 non-local entry points.
1974 ** pathwise CAST removing failed when the CAST node did not start
1976 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1977 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1979 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1980 host is already defined.
1981 ** RENAME-FILE works on streams instead of signalling an internal
1983 ** PEEK-CHAR uses the current readtable when determining whether
1984 or not a character is whitespace.
1985 ** MERGE-PATHNAMES handles the case when the pathname does not
1986 specify a name while the default-pathname specifies a version.
1987 ** Pathnames now stand a better chance of respecting print/read
1989 ** Attempting to use standardized file system operators with a
1990 pathname with invalid :DIRECTORY components signals a
1992 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1994 ** DIRECTORY on logical pathnames is more correct.
1995 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1996 signal a TYPE-ERROR if their argument is not a stream.
1997 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1998 designator argument does not designate a stream.
1999 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2000 examining the synonym.
2001 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2003 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2004 element-type larger than ([UN]SIGNED-BYTE 32).
2006 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2007 * When built with the :SB-FUTEX feature, threaded builds now take
2008 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2009 for faster/more reliable mutex and condition variable support.
2010 * Incompatible change (but one you probably shouldn't have been using
2011 anyway): the interface and code for arbitrating between multiple
2012 threads in the same user session has been redesigned.
2013 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2014 error when called without an explicit environment argument.
2015 (thanks to Nikodemus Siivola)
2016 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2017 confused about their position. (thanks to Adam Warner and Gerd
2019 * bug fix: Pretty printing backquoted forms with unquotations in the
2020 argument list position of various code constructs such as LAMBDA
2021 now works correctly. (reported by Paul Dietz)
2022 * bug fix: Pretty printing unquotations no longer loses all
2023 stream position information.
2024 * optimization: performance of string output streams is now less
2025 poor for multiple small sequence writes.
2026 * optimization: performance of CSUBTYPEP in the presence of complex
2027 expressions involving CONS and NOT many times has been improved.
2028 (reported by Paul Dietz)
2029 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2031 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2032 streams for the REPL, for the debugger, and so on. (thanks to
2034 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2035 * fixed some bugs revealed by Paul Dietz' test suite:
2036 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2037 is ignored if it is too small, rather than propagating through
2038 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2039 ** extremely complex negations of CONS types were not being
2040 sufficiently canonicalized, leading to inconsistencies in
2042 ** VALUES tranformer lost derived type.
2044 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2045 * fixed a bootstrapping bug: the build process no longer assumes
2046 that the various BOOLE-related constants have the same value in
2047 host and target lisps. (noted by Paul Dietz' test suite on an
2048 SBCL binary built from CLISP)
2049 * The system can now be dynamically linked on the MIPS platform,
2050 which enables dynamic loading of foreign code from Lisp. (thanks
2051 to Ralf Baechle for discussions on the MIPS ABI)
2052 * The system now records debugging information for its own source
2053 files in a filesystem-position-independent manner, relative to
2054 the "SYS" logical host.
2055 * fixed a compiler bug: MV-LET convertion did not check references
2056 to the "max args" entry point. (reported by Brian Downing)
2057 * tweaked disassembly notes to be less confident about proclaiming
2058 some instruction as an LRA. (thanks to Brian Downing)
2059 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2060 now each have their own history, command character, and other
2061 characteristics. (thanks to David Lichteblau)
2062 * fixed some bugs revealed by Paul Dietz' test suite:
2063 ** compiler failure in compiling LOGAND expressions including a
2065 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2066 shift greater than 32.
2067 ** FLUSH-DEST did not mark blocks for type check regeneration.
2068 ** HANDLER-CASE failed to accept declarations in handler clauses
2069 in some circumstances.
2071 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2072 * New code in contrib/sb-introspect (still probably not entirely
2073 stable yet) provides some support for smart Lisp development
2074 environments like SLIME.
2075 * The conditions signalled for errors occurring when loading .fasl
2076 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2077 in a way which should help ASDF recover gracefully.
2078 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2079 stuff has been cleaned up. If you code contrib/ stuff, this might
2080 affect you, and you can look at contrib/README, contrib/STANDARDS,
2081 and/or the 0.8.4.27 diff to check.
2082 * In full calls the compiler now does not generate checks for declared
2083 argument types for all arguments.
2084 * various threading fixes
2085 ** and some experimental patches which didn't make it into
2086 the main tree for this release, but which are shipped in
2087 contrib/experimental-thread.patch as a possible fix for some
2088 failures (deadlock, spinning...) in GC-intensive multithreaded
2090 * fixed PPC build problem (source code incompatibility of different
2091 library versions): added offsetof() hackery which attempts to divine
2092 where glibc maintainers put uc_mcontext today
2093 * fixed bug 282: compiler does not trust type assertions while passing
2094 arguments to a full call.
2095 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2096 &OPTIONAL VALUES type parameter.
2097 * fix bug 214: algorithm for noting rejected templates is now more
2098 similar to that of template seletion. (also reported by rydis on
2100 * fixed bug 141b: printing backquoted information readably and prettily
2101 inserts a space where necessary.
2102 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2103 (CHARACTER 10) are now reported as errors, rather than propagated
2104 as unknown types. (reported by piso on #lisp)
2105 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2106 with values NIL and :ERROR. (thanks to Milan Zamazal)
2107 * fixed bug 191c: CLOS now does proper keyword argument checking as
2108 described in CLHS 7.6.5 and 7.6.5.1.
2109 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2110 longer raise an error; further, using a list as a for-as-arithmetic
2111 counter now raises a meaningful error.
2112 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2113 now understood by sequence creation functions such as MAKE-SEQUENCE
2115 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2116 a STRING-INPUT-STREAM.
2117 * compiler enhancement: SIGNUM is now better able to derive the type
2119 * type declarations inside WITH-SLOTS are checked. (reported by
2121 * fixed some bugs revealed by Paul Dietz' test suite:
2122 ** incorrect optimization of TRUNCATE for a positive first
2123 argument and negative second.
2124 ** compiler failure in let-convertion during flushing dead code.
2125 ** compiler failure while deriving type of TRUNCATE on an
2126 interval, containing 0.
2127 ** ASH of a negative bignum by a negative bignum count now returns
2129 ** intersection of CONS types now canonicalizes properly, fixing
2130 inconsistencies in SUBTYPEP.
2132 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2133 * incompatible change: The --disable-debugger command line
2134 option now clobbers the debugger at a more fundamental
2135 level, by redefining #'INVOKE-DEBUGGER instead of by
2136 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2137 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2138 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2139 enter the debugger REPL and then suffer recursive errors
2140 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2141 while under the new system, BREAK in --disable-debugger mode
2142 terminates the system just as an unhandled error would.
2143 * fixed compiler performance when processing loops with a step >1;
2144 * bug fix: DOCUMENTATION now retrieves generic function
2145 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2146 support has been systematized, and now supports the methods
2147 specified by ANSI, along with a default method and a method for
2148 slot documentation. (reported by Nathan Froyd)
2149 * bug fix: effective methods associated with a generic function are
2150 no longer cached over a change of that generic function's method
2151 combination. (reported by Andreas Fuchs)
2152 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2153 names a non-existent directory. (thanks to Andreas Fuchs)
2154 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2155 the PPC platform, lead to stack corruption; this has been fixed.
2156 (reported by Rainer Joswig)
2157 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2158 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2159 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2160 arithmetic in general, is now much more reliable on the PPC
2162 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2163 platform now returns the right answer.
2164 * optimization: restored some effective method precomputation in
2165 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2166 precomputation is now tunable.
2167 * optimization: compiler-internal data structure use has been
2168 reviewed, and changes have been made that should improve the
2169 performance of the compiler by about 20%.
2170 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2171 simple-base-strings and simple-bit-vectors is improved.
2172 * optimization: the optimization of 32-bit logical and arithmetic
2173 functions introduced in version 0.8.3 on the x86 has been
2174 implemented on the mips, ppc and sparc platforms; an
2175 implementation of the same facility, but for 64-bit arithmetic,
2176 has been added for the alpha.
2177 * microoptimization: the compiler is better able to make use of the
2178 x86 LEA instruction for multiplication by constants.
2179 * bug fix: in some situations compiler did not report usage of
2180 generic arithmetic in (SPEED 3) policy.
2181 * bug 145b fix: compiler used wrong type specifier while converting
2182 MEMBER-types to numeric.
2183 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2185 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2186 index argument now works properly on the Alpha platform.
2187 * bug fix: floating point exception treatment on the Alpha platform
2189 * bug fix: FILE-POSITION works much better on string input and
2190 output streams. (thanks to Nikodemus Siivola)
2191 * bug fix: many threading/garbage collection symptoms sorted.
2192 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2193 might be pseudo-atomic.
2194 * internal change: Stopping for GC is now done with signals not
2195 ptrace. GC is now done in whichever thread wanted it, instead of
2197 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2199 * bug fix: result form in DO is not contained in the implicit
2201 * incompatible change: ICR structure is changed; the value part of
2202 CONTINUATION is now called LVAR; corresponding functions are
2203 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2204 * added type deriver for ISQRT (thanks to Robert E. Brown).
2205 * bug fix: better support for loading from the command line when an
2206 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2208 * fixed some bugs revealed by Paul Dietz' test suite:
2209 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2210 ** ROUND and FROUND now give the right answer when given very
2211 small float arguments.
2212 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2214 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2215 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2216 ** LCM should return a non-negative integer.
2217 ** PARSE-INTEGER returned the index of a terminator instead of the
2218 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2219 ** PARSE-INTEGER returned an incorrect index being applied to a
2221 ** LCM with two arguments of 0 returns 0 rather than signalling
2223 ** unsigned addition of a 32-bit constant with the high bit set no
2224 longer causes an internal compiler error on the x86.
2225 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2226 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2227 with negative last argument.
2228 ** byte specifiers with zero size and position no longer cause
2229 an error during type derivation.
2230 ** bignum multiplication on the Alpha platform now returns the
2232 * porting: The system now builds on SuSE AMD64, although it still
2233 generates a 32-bit binary.
2234 * .fasl file incompatibility: The fasl file version number has
2235 been incremented (because of the changes to internal compiler
2236 data structures referred to above).
2238 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2239 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2240 more accurately, on the Darwin kernel running on PowerPC hardware.
2241 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2242 * Compiler code deletion notes now signal a condition of type
2243 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2244 an associated MUFFLE-WARNING restart.
2245 * The compiler now performs limited argument count validation of
2246 constant format strings in FORMAT, and where appropriate in ERROR,
2247 CERROR and WARN. (thanks to Gerd Moellmann)
2248 * New ASDF-INSTALL contrib can be used for automatic download and
2249 installation of third-party Lisp code from CCLAN or other sites
2251 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2252 which forces another thread to execute a function supplied by the
2254 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2255 now accept and act upon their :ELEMENT-TYPE keyword argument.
2256 (reported by Martin Atzmueller, Edi Weitz)
2257 * bug fix: FILE-POSITION now accepts position designators up to
2258 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2259 is the greater. (thanks to Patrik Nordebo)
2260 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2261 Kalle Olavi Niemitalo)
2262 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2263 printable. (reported by Eric Marsden)
2264 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2265 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2266 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2267 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2268 (thanks to Stig Sandoe)
2269 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2270 UNREAD-CHAR now correctly includes the unread character in the
2271 target sequence. (thanks to Gerd Moellmann)
2272 * bug fix in threaded builds: the system can now be suspended and
2273 resumed by shell job control with minimal disruption.
2274 * bug fixes in times and timezones >2038AD
2275 * better handling of "where is GNU make?" problem in build scripts
2276 (thanks to Nikodemus Siivola)
2277 * new optimization: inside a named function any reference to a
2278 function with the same name is considered to be a self-reference;
2279 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2280 optimization quality.
2281 * new optimization on x86: logical functions and + now have
2282 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2283 used when the result is truncated to 32 bits.
2284 * VALUES declaration is partially enabled.
2285 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2286 * fixed some bugs revealed by Paul Dietz' test suite:
2287 ** The system now obeys the constraint imposed by
2288 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2289 types form a lattice under type intersection.
2290 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2291 ** ASSOC now ignores NIL elements in an alist.
2292 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2293 and (1+ MOST-POSITIVE-FIXNUM) answers.
2294 ** The addition of a method with invalid qualifiers to a generic
2295 function does not cause an error to be signalled immediately;
2296 a warning is signalled, and the error is generated only on
2297 calling the generic function.
2298 * changed .fasl file version number, in part to add type codes for
2299 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2300 obscure ANSI requirements
2302 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2303 * fixed bug 148: failure to inline-expand a local function left
2304 garbage, confusing the compiler.
2305 * fixed bugs 3cd: structure slot readers perform type check if the
2306 slot can have an invalid value (i.e. it is either not initialized
2307 or can be written with a less specific slot writer).
2308 * bug fix: the compiler now traps array references to elements off
2309 the end of an array; previously, the bounds checking in some
2310 circumstances could go off-by-one.
2311 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2313 * type declarations for array element types now obey the description
2314 on the CLHS page "Declaration TYPE", as per discussions on
2315 sbcl-help around 2003-05-08. This means that a declaration
2316 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2317 declaration, all references to BAR will be asserted or assumed
2318 (with THE, so dependent on compiler policy) to involve objects of
2319 type FOO. Note that no such declaration is implied in
2320 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2321 * declared types of functions from the "Conditions"
2322 chapter. (reported by Paul Dietz)
2323 * bug fix: CERROR accepts a function as its first argument.
2324 * bug fix: NTH an NTHCDR accept a bignum as index
2325 arguments. (reported by Adam Warner)
2326 * optimization: character compare routines now optimize comparing
2327 against a constant character. (reported by Gilbert Baumann)
2328 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2329 argument now works properly on the MIPS platform.
2330 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2331 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2332 declared to have a functional type, does not cause an error
2334 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2335 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2336 * fixed bug: if last continuation of a deleted block has a
2337 destination, this destination should be deleted too. (reported by
2339 * fixed a bug in the bootstrap process: the host compiler's values
2340 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2341 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2342 test case from Patrik Nordebo)
2343 * improved the ability of the disassembler on the PPC platform to
2344 provide helpful disassembly notes.
2345 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2346 the class in more cases than previously.
2347 * bug fix: FILE-POSITION now understands :START and :END for
2348 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2349 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2350 8) data. (thanks to David Lichteblau)
2351 * bug fix: it is possible to add a method to a generic function
2352 without lambda list.
2353 * bug fix: reader failed to signal END-OF-FILE inside an
2354 object representation. (reported by Nikodemus Siivola)
2355 * fixed some bugs revealed by Paul Dietz' test suite:
2356 ** LAST and [N]BUTLAST should accept a bignum.
2357 ** condition slot accessors are methods.
2358 ** (VECTOR NIL) is a subtype of STRING.
2360 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2361 * minor incompatible change: some nonsensical specialized lambda
2362 lists (used in DEFMETHOD) which were previously ignored now signal
2364 * minor incompatible change: the system is now aware of the types of
2365 variables in the COMMON-LISP package, and will signal errors for
2366 most violations of these type constraints (where previously they
2367 were silently accepted).
2368 * minor incompatible change: COMPILE-FILE now uses the freedom
2369 afforded (ANSI 3.2.2.3) to use derived function types for
2370 functions defined in the same file. This also permits the system
2371 to warn on static type mismatches and function
2372 redefinition. (Currently it does not work with high DEBUG level.)
2373 * minor incompatible change: VALUES declaration is disabled.
2374 * When issuing notes, the compiler now signals a condition of type
2375 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2376 restart for use in user handlers. It is expected that the
2377 COMPILER-NOTE condition will eventually become a condition
2378 supertype to a hierarchy of note types, which will then be
2379 handleable in a similar fashion. However, at the moment, no such
2380 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2381 but it's an implementation detail, not a classification for the
2383 * Changes in type checking closed the following bugs:
2384 ** type checking of unused values (192b, 194d, 203);
2385 ** template selection based on unsafe type assertions (192c, 236);
2386 ** type checking in branches (194bc).
2387 * A short form of VALUES type specifier has ANSI meaning (it has
2388 increased the number of situations when SBCL cannot perform type
2390 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2391 keywords or constants is permissible.
2392 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2393 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2394 classes. (thanks to Antonio Martinez)
2395 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2396 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2397 argument to a pretty-print stream. Instead, it leaves any such
2398 operations to DESCRIBE-OBJECT methods.
2399 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2401 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2402 lambda lists. (thanks to David Lichteblau)
2403 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2404 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2405 (thanks to Gerd Moellmann)
2406 * fixed bug 246: increased compilation speed of long
2407 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2409 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2411 * DEFCONSTANT now throws a condition of type
2412 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2413 constant to a non-EQL value; CONTINUE and ABORT restarts
2414 respectively change and preserve the value.
2415 * fixed bug 63: The code walker, part of the implementation of CLOS,
2416 is now better at handling symbol macros.
2417 * bug fix: There is no longer an internal implementation type named
2418 CL:LENGTH. (reported by Raymond Toy)
2419 * bug fix: In macro-like defining macros/special operators the
2420 implicit block does not enclose the lambda list.
2421 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2422 enclosed in parentheses) are not suitable as type specifiers, and
2423 their use properly signals an error now.
2424 * bug fix: An argument count mismatch for a type specifier in code
2425 being compiled no longer causes an unhandled error at compile
2426 time, but signals a compile-time warning.
2427 * fixed simple vector readable printing
2428 * bug fix: DESCRIBE takes more care over whether the class
2429 precedence list slot of a class is bound before accessing it.
2430 (reported by Markus Krummenacker)
2431 * bug fix: FORMATTER can successfully compile pretty-printer format
2432 strings which use variants of the ~* directive inside.
2433 * bug fix: SEARCH now applies its TEST predicate to the elements of
2434 the arguments in the correct order. (thanks to Wolfhard Buss)
2435 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2436 in type inference. (thanks to Robert E. Brown)
2437 * bug fix: Reading in symbols with an explicit package name of ""
2438 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2439 * fixed some bugs revealed by Paul Dietz' test suite:
2440 ** NIL is now allowed as a structure slot name.
2441 ** Arbitrary numbers, not just REALs, are allowed in certain
2442 circumstances in LOOP for-as-arithmetic clauses.
2443 ** Multiple class redefinitions before slot access no longer
2444 causes a type error.
2445 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2446 association between the name and a class.
2447 ** Generic functions with non-standard method-combination and over
2448 five methods all of which return constants no longer return NIL
2449 after the first few invocations. (thanks to Gerd Moellmann)
2450 ** CALL-NEXT-METHOD with no arguments now passes the original
2451 values of the arguments, even in the presence of assignment.
2452 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2454 ** DISASSEMBLE works with closures and funcallable instances.
2455 ** ADD-METHOD now returns the generic function, not the new method.
2456 ** FIND-METHOD signals an error if the lengths of the specializers
2457 is incompatible with the generic function, even if the ERRORP
2459 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2460 which its argument is a member.
2461 ** DEFCLASS only redefines the class named by its class-name
2462 argument if that name is the proper name of the class;
2463 otherwise, it creates a new class.
2464 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2465 of the UNBOUND-SLOT condition to the name of the slot.
2466 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2467 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2468 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2469 SLOT-MAKUNBOUND in the specified fashion.
2471 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2472 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2473 cross-compilation host. As a consequence, we can now bootstrap our
2474 way up to SBCL starting with a bare gcc toolchain and human-readable
2475 source code (first the source to CLISP, then the source to SBCL).
2476 * A contributed module containing a partial implementation of the
2477 simple-streams interface has been included. (thanks to Rudi
2479 * A contributed module implementing the RFC1321 Message Digest
2480 Algorithm, known as MD5, has been included.
2481 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2482 no longer has any effect, as the code controlled by this feature
2483 has been deleted. (As far as we know, no-one has ever built using
2484 this feature, and its semantics were confused in any case).
2485 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2486 work on conditions (as required by the ANSI specification),
2487 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2488 expected behaviour on conditions. Users should note, however,
2489 that such behaviour is not required by the ANSI specification,
2490 and so use of this behaviour may render their code unportable.
2491 * fixed some bugs revealed by Paul Dietz' test suite:
2492 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2494 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2495 ** SXHASH on condition objects no longer returns NIL.
2496 ** :ALLOCATION :CLASS slots are better treated; their values are
2497 updated on class redefinition, and initforms inherited from
2498 superclasses are applied.
2499 ** REMOVE-METHOD returns its generic function argument even when
2500 no method was removed.
2501 ** SHARED-INITIALIZE now initializes the values of the requested
2502 slots, including those with :ALLOCATION :CLASS.
2503 ** ALLOCATE-INSTANCE now works on structure classes defined via
2504 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2506 ** SLOT-EXISTS-P now works on conditions, as well as structures
2508 ** MAKE-LOAD-FORM now has the required methods on
2509 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2510 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2511 keyword, and now implements the SLOT-NAMES argument.
2512 ** methods with &OPTIONAL arguments no longer allow too many
2513 arguments to be passed in the call without error.
2514 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2515 option is consistent with the required arguments of the generic
2516 function lambda list.
2517 * bug fix: REQUIRE accepts a string designator. (Thanks to
2519 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2520 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2521 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2522 * bug fix: The compiler now checks for duplicated variables in macro
2524 * bug fix: SETQ on globals returns the correct value.
2525 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2526 not a COMPILER-ERROR (followed by some other strange error on
2527 choosing the CONTINUE restart).
2528 * bug fix: make.sh and friends are now more consistent in the way that
2529 they look for GNU "make".
2531 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2532 * experimental native threads support (on x86 Linux >=2.4 only).
2533 This is not compiled in by default: you need to add :SB-THREAD to
2534 the target features. See the "Beyond ANSI" chapter of the manual
2536 * fix for longstanding nonANSIism: The old distinction between
2537 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2538 The return value from CL:FIND-CLASS is now a CLOS class, and
2539 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2540 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2541 * An interface to the MetaObject Protocol, as described in Kiczales,
2542 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2543 MIT Press, 1991, is available from the SB-MOP package.
2544 * incompatible change: the SB-PCL package should now be considered
2545 a private implementation detail, and no longer a semi-private MOP
2547 * minor incompatible change: due to rearrangement for threads, the
2548 control stack and binding stack are now allocated at arbitrary
2549 addresses instead of being hardcoded per-port. Users affected by
2550 this probably have to be doing advanced things with shared
2551 libraries, and will know who they are.
2552 * minor incompatible change: Previously, all --eval forms used were
2553 processed with READ before any of them were processed with EVAL.
2554 Now each --eval form is processed with both READ and EVAL before
2555 the next --eval form is processed. (Thus package operations like
2556 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2557 work as the user might reasonably expect.)
2558 * minor incompatible change: *STANDARD-INPUT* is now only an
2559 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2561 * minor incompatible change: Y-OR-N-P is now character-oriented,
2562 not line oriented. Also, YES-OR-NO-P now works without errors.
2563 (thanks to Antonio Martinez)
2564 * sb-aclrepl module improvements: an integrated inspector, added
2565 repl features, and a bug fix to :trace command.
2566 * Known functions, which cannot be open coded by the backend, are
2567 considered to be able to check types of their arguments. (fixing
2568 a bug report by Nathan J. Froyd)
2569 * fixed a bug in computing method discriminating functions: It is
2570 now possible to define methods specialized on classes which have
2571 forward-referenced superclasses. (thanks to Gerd Moellmann)
2572 * fixed evaluation order in optional entries (reported by Gilbert
2574 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2575 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2576 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2577 required-by-AMOP NAME argument, as well as CLASS and
2578 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2579 * fixed bug 20: DEFMETHOD can define methods using names that are
2580 not the proper names of classes to designate class specializers.
2581 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2582 Unix file descriptors, instead of blowing up. (thanks to Antonio
2584 * Garbage collection refactoring: user-visible change is that a
2585 call to the GC function during WITHOUT-GCING will not do garbage
2586 collection until the end of the WITHOUT-GCING. If you were doing
2587 this you were probably losing anyway.
2588 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2589 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2590 TYPEP the latter but not the former.
2591 * The compiler issues a full WARNING for calls to undefined functions
2592 with names from the CL package.
2593 * MAP-INTO for a vector destination is open coded. (reported by
2594 Brian Downing on c.l.l)
2595 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2596 documentation string.
2597 * fixed some bugs revealed by Paul Dietz' test suite:
2598 ** COPY-ALIST now signals an error if its argument is a dotted
2600 ** Condition slots are now accessed more correctly in the presence
2601 of multiple initargs for a given slot.
2602 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2603 exclude restarts of the same name associated with a different
2605 ** DEFCLASS of forward-referenced classes with another
2606 forward-referenced class in the superclasses list no longer
2608 ** Condition slots are now initialized once each, not multiple
2609 times. (thanks to Gerd Moellmann)
2610 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2612 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2613 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2614 arguments contain duplicated elements.
2615 ** RESTART-CASE understands local macros.
2616 ** RESTART-CASE associates exactly its own restarts with a condition.
2617 ** ENDP in safe mode checks its argument to be of type LIST.
2618 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2619 in question is unbound.
2620 ** Optimized MAKE-INSTANCE functions no longer cause internal
2621 assertion failures in the presence of duplicate initargs.
2622 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2623 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2624 propagated even in the ctor optimized implementation of
2626 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2628 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2629 * a better implementation of SXHASH on (simple) bit vectors,
2630 measured both in execution speed and in distribution of results
2631 over the positive fixnums, has been installed. Likewise, a better
2632 implementation of EQUAL for simple bit vectors is now available.
2633 * fixed CEILING optimization for a divisor of form 2^k.
2634 * fixed bug 240 (emitting extra style warnings "using the lexical
2635 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2636 by Antonio Martinez)
2637 * fixed SXHASH, giving different results for NIL depending on type
2638 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2639 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2640 variable. (found by Rolf Wester)
2641 * SBCL does not ignore type declarations for special
2642 variables. (reported by rif on c.l.l 2003-03-05)
2643 * some bug fixes in contrib/sb-aclrepl/
2644 * fixed some bugs revealed by Paul Dietz' test suite:
2645 ** a bug in the CONS type specifier, whereby the CAR and CDR
2646 types got intertwined, has been fixed;
2647 ** the type system is now able to reason about the interaction
2648 between INTEGER and RATIO types more completely;
2649 ** APPEND, [N]REVERSE and NRECONC check that those their
2650 arguments, which must be proper lists, are really so;
2651 ** An array specialized to be unable to hold elements has been
2652 implemented, as required -- yes, really -- by ANSI;
2653 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2654 on malformed property lists;
2656 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2657 * incompatible packaging change: in line with Unix convention,
2658 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2659 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2660 * REQUIRE and PROVIDE are now optionally capable of doing something
2661 useful. See the documentation string for REQUIRE.
2662 * infrastructure for a managed SBCL contrib system: contributed
2663 modules in this release include:
2664 ** the ASDF system definition facility;
2665 ** an interface to the BSD Sockets API;
2666 ** an ACL-like convenience interface to the repl;
2667 (thanks to Kevin Rosenberg)
2668 ** an implementation of ROTATE-BYTE, with efficient implementation
2670 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2671 gives a closer approximation to the right answer than previously.
2672 (thanks to Raymond Toy)
2673 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2674 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2675 optional environment arguments, as required by ANSI.
2676 * fixed bugs in other functions taking environment objects, allowing
2677 calls with an explicit NIL environment argument to be compiled
2679 * fixed bug 228: primary return values from
2680 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2681 COMPILE or FUNCTION.
2682 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2683 :TYPE will no longer signal errors on innocuous objects.
2684 * fixed bug 231b: SETQ is better at respecting type declarations in
2685 the lexical environment.
2686 * fixed a bug in DEFCLASS: classes named by symbols with no or
2687 unprintable packages can now be defined.
2688 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2689 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2690 * fixed compiler failure related to checking types of functions.
2691 (reported by Robert E. Brown)
2692 * the compiler is now much more consistent in its error-checking
2693 treatment of bounding index arguments to sequence functions: in
2694 (SAFETY 3) code, errors will be signalled in almost all cases if
2695 invalid sequence bounding indices are passed to functions defined
2696 by ANSI to operate on sequences.
2697 * fixed a bug in the build procedure: documentation of SBCL-specific
2698 packages is now preserved and available in the final Lisp image.
2699 * lifted FDEFINITION lookup out of loops in the implementation of
2700 many list operations. (thanks to Robert E. Brown)
2701 * fixed a bug in the reader: the #n# reader macro now works for
2702 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2703 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2704 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2705 E. Brown and Tony Martinez respectively)
2706 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2707 SLOT-BOUNDP: in particular, it is now able to optimize them much
2708 better, and is now not vulnerable to having packages renamed.
2709 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2710 always called when a slot is not present in an instance. (thanks
2712 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2713 longer signal an error on unreading a character following EOF on
2714 the previous constituent stream. (thanks to Tony Martinez)
2715 * fixed some bugs revealed by Paul Dietz' test suite:
2716 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2717 not just nonnegative fixnums;
2718 ** the logical bit-array operators such as BIT-AND now accept an
2719 explicit NIL for their "opt-arg" argument (to indicate a
2720 freshly-consed result bit-array);
2721 ** ELT now signals an error on an invalid sequence index in safe
2723 ** the type system is now cleverer about negations of numeric
2724 types, and consequently understands the BIGNUM and RATIO types
2726 ** the type system is now cleverer about the interaction between
2727 INTEGER and RATIO types: while bugs still remain, many more
2728 cases are accurately computed;
2729 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2730 if it is in the last clause;
2731 ** CONSTANTLY now correctly returns a side-effect-free function in
2733 ** DECLARE is no longer treated as a special-operator; in
2734 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2735 * incremented fasl file version number due to the change in the
2736 DEFSTRUCT-SLOT-DESCRIPTION structure.
2738 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2739 * minor incompatible change: code processed by the "interpreter" or
2740 EVAL now has a compilation optimization policy of (DEBUG 2)
2741 (changed from (DEBUG 1)) to improve debuggability of interactive
2742 development, and to allow the use of the debug RETURN command in
2744 * an experimental implementation of the RETURN command for the
2745 debugger has been included. (thanks to Frederik Kuivinen)
2746 * fixed bug 62: constraints were not propagated into a loop.
2747 * fixed bug in embedded calls of SORT (reported and investigated by
2749 * fixed some bugs revealed by Paul Dietz' test suite:
2750 ** printing and reading of arrays with some dimensions having
2751 length 0 (thanks to Gerd Moellmann);
2752 ** BOA constructor with &AUX argument without a default value does
2753 not cause a type error;
2754 ** CONSTANTP now returns true for all self-evaluating objects.
2756 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2757 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2758 accessors that are related by inheritance, as specified in the
2759 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2760 Valtteri Vuorikoski)
2761 * The compiler is now able to inline functions that were defined in
2762 a complex lexical environment (e.g. inside a MACROLET).
2763 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2765 * Support for the upcoming FreeBSD-5.0 release has been included.
2766 (thanks to Dag-Erling Smorgrav)
2767 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2768 effect when it is not in a toplevel context.
2769 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2770 stand a better chance of being correct. (thanks to Gerd
2772 * fixed bug in COERCE, which now signals an error on coercing a
2773 rational to a bounded real type which excludes the expected
2775 * The compiler is now able to derive types more accurately from the
2776 COERCE and COMPILE functions.
2777 * fixed bug 223: functional binding is considered to be constant
2778 only for symbols in the CL package.
2779 * fixed bug 231: SETQ did not check the type of a variable being set
2780 (reported by Robert E. Brown)
2781 * A new optimization for MAKE-INSTANCE has been included, fixing
2782 various bugs (including relating to :ALLOCATION :CLASS slots and
2783 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2785 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2786 ** As required by ANSI, LOOP now disallows anonymous collection
2787 clauses such as COLLECT I in conjunction with aggregate boolean
2788 clauses such as THEREIS (= I 1);
2789 ** LOOP now signals an error when any variable is reused in the
2790 same loop (including the potentially useful construct analogous
2791 to WITH A = 1 WITH A = (1+ A);
2792 ** IT is only a special loop symbol within the first clause of a
2793 conditional loop clause;
2794 ** LOOP with a typed iteration variable over a hashtable now
2795 signals a type error iff it should.
2796 * fixed some other bugs revealed by Paul Dietz' test suite:
2797 ** FILE-STREAM now names the class previously known as FD-STREAM;
2798 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2799 argument) no longer signals an error;
2800 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2801 of the slot symbol, rather than using the current package
2802 ((:CONC-NAME "") continues to intern the slot's name in the
2804 * incremented fasl file version number, because of the incompatible
2805 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2806 of the new implementation of DEFINE-COMPILER-MACRO.
2808 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2809 * Support for building SBCL for MIPS platforms running in
2810 little-endian mode has now been checked in, and basic
2811 functionality on said platforms verified.
2812 * minor incompatible change: PCL now records the pathname of a file
2813 in which methods and the like are defined, rather than its
2815 * minor incompatible change: TRUENAME now considers the truename of
2816 a file naming a directory to be the pathname with :DIRECTORY
2817 component indicating that directory.
2818 * minor incompatible change: a NAMED clause in the extended form of
2819 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2820 reason for the previous behaviour is unclear.
2821 * more systematization and improvement of CLOS and MOP conformance
2822 in PCL (thanks to Gerd Moellman and Pierre Mai):
2823 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2825 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2826 primary methods with no specializers;
2827 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2829 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2830 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2831 FORWARD-REFERENCED-CLASSes; error reporting on
2832 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2834 ** SXHASH on CLOS instances now uses a slot internal to the
2835 instance to return different numbers on distinct instances,
2836 while preserving the same return value through invocations of
2838 ** DEFMETHOD signals errors when methods with longer incongruent
2839 lambda lists are added to generic functions;
2840 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2841 CLASS, as specified in AMOP;
2842 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2843 on the order returned by the primary method for classes of
2844 class STANDARD-CLASS;
2845 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2846 * fixed some bugs shown by Paul Dietz' test suite:
2847 ** DOLIST puts its body in TAGBODY;
2848 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2850 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2851 value producing form;
2852 ** if more variables are given to PROGV than values, extra
2853 variables are bound and made to have no value;
2854 ** NSUBSTITUTE on list arguments gets the right answer with
2856 ** ELT signals an error of type TYPE-ERROR when the index argument
2857 is not a valid sequence index;
2858 ** LOOP signals (at macroexpansion time) an error of type
2859 PROGRAM-ERROR when duplicate variable names are found;
2860 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2861 ** FUNCALL of special-operators now cause an error of type
2863 ** PSETQ now works as required in the presence of side-effecting
2864 symbol-macro places;
2865 ** NCONC accepts any object as its last argument;
2866 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2868 ** loop-for-as-package does not require a package to be explicitely
2870 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2872 * fixed bug 166: compiler preserves "there is a way to go"
2873 invariant when deleting code.
2874 * fixed bug 172: macro lambda lists with required arguments after
2875 &REST arguments now cause an error to be signalled. (thanks to
2877 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2878 bug. (thanks to Matthew Danish)
2879 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2881 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2882 arguments are lexically rebound. (thanks to Gerd Moellmann and
2884 * fixed bug 194: error messages are now more informative when there
2885 is no primary method applicable in a call to a generic
2886 function. (thanks to Gerd Moellmann)
2887 * fixed bug in command line argument checking (thanks to Julian
2889 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2890 (thanks to Matthew Danish)
2891 * incremented fasl file version number, because of the
2892 SXHASH-related changes in the layout of CLOS data structures
2894 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2895 * minor incompatible change: The runtime (the Unix executable named
2896 "sbcl") is now much pickier about the .core files it will load.
2897 Essentially it now requires .core files to descend from the same
2898 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2899 as the runtime does. (The intent is to prevent the crashes which
2900 can occur, and which can even be reported as mysterious failures,
2901 when people patch the sources or change the build parameters
2902 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2903 sbcl and .core files.)
2904 * fixed bug: VALUES-LIST is no longer optimized away.
2905 * fixed bug 142: The FFI conversion of C string values to Lisp
2906 string values no longer conses excessively. (thanks to Nathan
2907 Froyd porting Raymond Toy's fix to CMU CL)
2908 * began to systematize and improve MOP conformance in PCL (thanks to
2909 Nathan Froyd, Gerd Moellman and Pierre Mai):
2910 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2912 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2913 ** FINALIZE-INHERITANCE is now called on class finalization;
2914 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2915 argument precedence order.
2916 * fixed bug 202: The compiler no longer fails on functions whose
2917 derived types contradict their declared type.
2918 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2919 so it can be non-toplevel.
2920 * The fasl file version number has changed (because of the new
2921 implementation of DEFMACRO).
2922 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2923 safe code, that any length requirement by their type-specifier
2924 argument is valid. The exceptions to this are described in bug
2926 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2927 functions now check the types of their inputs as required by ANSI.
2928 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2929 introduced symbol is DECLAREd to be SPECIAL.
2930 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2931 * fixed bug: :COUNT argument to sequence functions may be negative
2932 * fixed bug: body of DO-SYMBOLS may contain declarations
2933 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2934 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2935 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2936 symbol macro only once
2937 * fixed printing of call frame when argument list is unavailable
2938 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2939 * compiler no longer signals WARNING on unknown keyword
2942 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2943 * A beta-quality port to the mips architecture running Linux,
2944 based on the old CMUCL backend, has been made. It has been tested
2945 on a big-endian kernel, and works sufficiently well to be able to
2946 rebuild itself; it has not been tested in little-endian mode.
2947 * fixed an inconsistency between gencgc.c and purify.c which made
2948 dumping/loading .core files unreliable
2949 * fixed bug 120a: The compiler now deals correctly with IFs where
2950 the consequent is the same as the alternative, instead of
2951 misderiving the return type. (thanks to Alexey Dejneka)
2952 * fixed bug 113: Logical pathnames are now dumpable (the logical
2953 host is resolved at load-time, throwing an error if it is not
2955 * fixed bug 174: FORMAT's error message is slightly clearer when a
2956 non-printing character is used in a format directive.
2957 * fixed several bugs in compiler checking of type declarations, i.e.
2958 violations of the Python "declarations are assertions" principle
2959 (thanks to Alexey Dejneka)
2960 * fixed several bugs in PCL's error checking (thanks to Gerd
2962 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2964 * fixed bug in compilation of functions as first class values
2965 (thanks to Antonio Martinez-Shotton)
2966 * The compiler's handling TYPE-ERRORs which it can prove will
2967 inevitably happen at runtime has been cleaned up and corrected
2968 in several ways. (thanks to Alexey Dejneka)
2969 * improved argument type checking for various basic arithmetic
2970 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2971 lavished on them in the past that they can be compiled in many
2972 ways in different special cases
2973 * fixed bug 181: compiler checks validity of user supplied type
2975 * cleaned up code flushing in optimization: Function calls which
2976 should signal errors for safety purposes (e.g. which ANSI says
2977 should signal errors when their arguments are of incorrect type)
2978 are no longer optimized away.
2979 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2980 * incremented fasl file version number, because changes in the
2981 implementation of sequence functions like COERCE caused
2982 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2983 in old inline expansions) to become undefined. (Actually these
2984 changes were later undone, so we might very well be binary
2985 compatible with 0.7.7 after all, but leaving the version number
2986 incremented seemed like the simplest and most conservative
2989 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2990 * An alpha-quality port to the parisc architecture running Linux,
2991 based on the old CMUCL backend, has been made. This, even more so
2992 than the other backends, should be considered still a work in
2993 progress; known problems include that the Linux kernel in 64-bit
2994 mode does not propagate the correct sigcontext structure to
2995 userspace, and consequently SBCL on a parisc64 kernel will not
2997 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2998 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2999 suppressed.) Also now that the compiler is looking at declarations
3000 in the environment, it checks optimization declarations as well,
3001 and suppresses inlining when (> DEBUG SPEED).
3002 * More fixes have been made to treatment of floating point exception
3003 treatment and other Unix signals. In particular, floating point
3004 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3005 * The detection and handling of control stack exhaustion (infinite
3006 or very deeply nested recursion) has changed. Stack exhaustion
3007 detection is now done by write-protecting pages at the OS level
3008 and applies at all optimization settings; when found, a
3009 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3010 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3011 be caught using IGNORE-ERRORS.
3012 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3013 correctly when transforming recursive LABELS functions to LETs.
3014 (thanks to Alexey Dejneka)
3015 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3016 that are names of constants or global variables.
3017 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3018 alien routines with docstrings.
3019 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3020 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3022 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3023 (thanks to Eric Marsden)
3024 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3025 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3026 * Bug fix: The SPARC backend can now compile functions involving
3027 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3028 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3029 to a functions expecting a PATHNAME-DESIGNATOR.
3030 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3031 lambda lists. (thanks to Alexey Dejneka)
3032 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3033 functions so that EXPORTing the name of the class doesn't cause
3034 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3035 to Antonio Martinez for reporting this)
3036 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3037 constant-folded without failing an assertion. (thanks to Einar
3038 Floystad Dorum for reporting this)
3039 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3040 types are now dealt with more correctly by the compiler.
3041 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3042 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3043 behaviour analogous to RENAME-FILE. This puts its behaviour more
3044 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3046 * The fasl file version number has changed again. (because of the
3047 bug fix involving the names of PCL MAKE-INSTANCE functions)
3049 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3050 * bug fix: Floating point exceptions are treated much more
3051 consistently on the x86/Linux and PPC/Linux platforms.
3052 * Array initialization with :INITIAL-ELEMENT is now much faster for
3053 cases when the compiler cannot open code the array creation, but
3054 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3055 array accesses have also seen a speed increase.
3056 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3057 specified. (This was at the root of some bad interactions between
3058 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3060 * bug fix: Internal error arguments for undefined functions are now
3061 computed correctly on the PPC/Linux platform.
3062 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3063 Raymond Toy's patch for CMU CL)
3064 * Support for the Solaris 9 operating environment has been included
3065 (thanks to Daniel Merritt)
3066 * A very ugly but hopefully complete draft of the missing FFI chapter
3067 of the manual has been created by reformatting the corresponding
3068 CMU CL manual chapter into (currently very ugly and incoherent)
3069 DocBook and bringing it up to date for SBCL behavior. Thus, the
3070 manual is now essentially complete, at least by my extreme
3071 once-and-only-once standards, whereby it's acceptable to refer to
3072 the doc strings of SB-EXT functions as the primary documentation.
3073 * The fasl file version number has changed again, due to cleanup of
3074 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3075 is no longer a static symbol.)
3077 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3078 * SBCL now builds with OpenMCL (version 0.12) as the
3079 cross-compilation host; also, more progress has been made toward
3080 bootstrapping under CLISP.
3081 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3083 * bug 158 fixed: The compiler can now deal with integer loop
3084 increments different from 1; fixing this turned out also to fix
3086 * bug 169 fixed: no more bogus warnings about using lexical bindings
3087 despite the presence of perfectly good SPECIAL declarations (thanks
3088 to David Lichteblau)
3089 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3090 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3091 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3093 * bug 180 fixed: Method combination specifications no longer ignore
3094 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3095 * bug fix: Structure type predicate functions now check their argument
3096 count as they should.
3097 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3098 correctly. (thanks to Pierre Mai)
3099 * minor incompatible change: The --noprogrammer option is deprecated
3100 in favor of the new --disable-debugger option, which is very similar.
3101 (The major difference is that it takes effect at a slightly different
3102 time at startup, causing handling of errors in --sysinit and
3103 --userinit files will be affected differently.) The
3104 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3105 been added to allow this functionality to be controlled from ordinary
3106 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3107 maintainers, who might want to run non-interactive scripts to
3108 build SBCL cores which will later be used interactively.)
3109 * minor incompatible change: The LOAD function no longer, when given
3110 a wild pathname to load, loads all files matching that pathname.
3111 Instead, an error of type FILE-ERROR is signalled.
3113 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3114 * bug 147 fixed: The compiler preserves its block link/count
3115 invariants more correctly now so that it doesn't crash. (thanks
3117 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3119 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3120 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3121 no-value-for-this-character as NIL (as specified by ANSI).
3122 * HOST-NAMESTRING on physical pathnames now returns a string that is
3123 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3124 (thanks to Christophe Rhodes)
3125 * The Alpha port handles icache flushing more correctly. (thanks to
3127 * More progress has been made toward bootstrapping under CLISP. (thanks
3128 to Christophe Rhodes)
3129 * The fasl file format has changed again, because dynamic loading
3130 on OpenBSD (which has non-ELF object files) motivated some cleanups
3131 in the way that foreign symbols are transformed and passed around.
3132 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3133 is no longer treated as whitespace by the reader, but instead as
3134 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3135 |A
\7fB|, instead of A as it used to.
3137 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3138 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3139 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3140 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3141 this as of 0.7.1.45, but was left out of the previous news section
3142 (thanks to Dan Barlow)
3143 * SBCL now runs on the Solaris operating system on SPARC architectures
3144 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3145 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3146 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3147 * SPARC backend cleanups, allowing builds of cores optimized for V8
3148 and V9 SPARCS, and also emission of code targeted to a particular
3149 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3151 * SBCL is closer to bootstrapping under CLISP, thanks to various
3152 fixes by Christophe Rhodes.
3153 * The fasl file format has changed again, to allow the compiler's
3154 INFO database to support symbol macros.
3155 * The user manual (in doc/) is formatted into HTML more nicely.
3156 (thanks to coreythomas)
3157 * The system is smarter about SUBTYPEP relationships, especially
3158 those involving NOT types (including types such as ATOM which are
3159 represented internally using NOT types). Thus SUBTYPEP is less
3160 likely to return (VALUES NIL NIL) in general, and in particular
3161 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3163 * The fasl file format has changed again, because the internal
3164 representation of types now includes a new slot to support the new
3165 SUBTYPEP-of-NOT-types logic.
3166 * (not a change in the main branch of SBCL, but a related prototype
3167 which can hopefully be merged into the main branch of SBCL in the
3168 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3169 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3170 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3171 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3172 dispatch macro character combinations. (thanks to Alexey Dejneka)
3173 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3174 that superficially look like logical namestrings correctly.
3175 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3177 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3178 default supplied-p) for &optional and &key arguments. (thanks to
3181 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3182 * incompatible change: The compiler is now less aggressive about
3183 tail call optimization, doing it only when (> SPACE DEBUG) or
3184 (> SPEED DEBUG). (This is an incompatible change because there are
3185 programs which relied on the old CMU-CL-style behavior to optimize
3186 away their unbounded recursion which will now die of stack overflow.)
3187 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3188 for non-GENCGC systems has been increased to 20M (since that
3189 seems much closer to the likely performance optimum for modern
3190 systems than the old 4M value was)
3191 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3192 and *DEBUG-PRINT-LEVEL*
3193 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3194 of CMU CL's support for SPARC, and various endianness and other
3195 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3196 * new syntactic sugar for the Unix command line: --load foo.bar is now
3197 an alternate notation for --eval '(load "foo.bar")'.
3199 ** The system now detects stack overflow and handles it gracefully,
3200 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3201 optimization settings. (This is a good thing in general, and
3202 its introduction in this version should be particularly timely
3203 for anyone whose code fails because of suppression of tail
3205 ** The system now hunts for the C variable "environ" in a more
3206 devious way, to avoid segfaults when the C library version differs
3207 between compile time and run time. (thanks to Christophe Rhodes)
3208 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3209 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3210 ** The compiler no longer issues bogus style warnings for undefined
3211 classes in the same source file as the DEFCLASSes which defined
3212 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3214 ** fixes in CONDITION class precedence list for undefined function
3215 errors (thanks to Alexei Dejneka)
3216 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3217 correctly. (thanks to Dan Barlow)
3218 ** portability fixes aiming at bootstrapping under CLISP (thanks
3219 to Dave McDonald and Christophe Rhodes)
3220 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3221 ** fixes in type translation and and type inference (thanks to
3223 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3224 ** various fixes in the new ports (thanks to Dan Barlow)
3225 * several changes related to debugging:
3226 ** suppression of tail recursion, as noted above
3227 ** stack overflow detection, as noted above
3228 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3229 is now the default. (For some time encapsulation has been more
3230 reliable than the breakpoint-based :ENCAPSULATE NIL
3231 implementation, at least on X86 systems; and I just noticed that
3232 encapsulation also seems closer to the spirit of the ANSI
3235 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3237 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3238 up properly again. (There was a packaging bug in 0.7.0 which
3239 left their definitions in SB-SYS::LOAD-FOREIGN and
3240 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3241 vital for most things which interface to C-level interfaces,
3242 like extensions working with sockets or databases or
3243 Perl-compatible regexes or whatever, and the need to fix
3244 this bug is the main reason that 0.7.1 was released so
3246 ** DEFGENERIC is now choosier about the methods it redefines, so that
3247 reLOADing a previously-LOADed file containing DEFGENERICs does
3248 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3249 cycle now works as it should. (thanks to Alexey Dejneka)
3250 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3251 fixed by Christophe Rhodes. (He actually submitted this patch
3252 months ago, and I delayed until after 0.7.0.)
3253 ** Bug 111 (internal compiler confusion about runtime checks on
3254 FUNCTION types) was fixed by Alexey Dejneka.
3255 * Some internal cleanups (getting rid of variables which aren't
3256 needed now that the byte interpreter is gone) caused the fasl
3257 file format number to change again.
3259 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3260 * major incompatible change: The default fasl file extension, i.e. the
3261 default extension for files produced by COMPILE-FILE, has changed
3262 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3264 ** There are many changes in the implementation of the compiler.
3265 SBCL is now essentially a compiler-only implementation of ANSI
3266 Common Lisp. EVAL still "interprets" a few special cases, but
3267 almost all the interesting cases are handled by creating
3268 a LAMBDA expression, calling COMPILE on it, then calling
3269 FUNCALL on the result.
3270 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3271 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3272 Since the code is newer, there might still be some new bugs
3273 (though not as many as before Martin Atzmueller's fixes:-). But
3274 the new code is substantially simpler and clearer, and hopefully
3275 any remaining bugs will be simpler, less fundamental, and more
3276 fixable then the bugs in the old code.
3277 ** The revised compiler is still a little unsteady on its feet.
3279 *** The debugging information it produces (particularly the names
3280 of FUNCTION objects) is sometimes much less useful than what
3281 the old compiler produced.
3282 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3283 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3284 within a MACROLET) has been temporarily weakened.
3285 ** There are new compiler optimizations for various functions:
3286 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3287 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3288 *** the math functions TRUNCATE, FLOOR, and CEILING
3289 *** the function-of-all-trades COERCE
3290 Mostly these should be transparent, but there's one
3291 potentially-annoying problem (bug 117): when the compiler
3292 inline-expands a function and does type analysis on the result,
3293 it can create control paths which have type mismatches, and
3294 when it can't prove that those control paths aren't taken,
3295 it will issue WARNINGs about the type mismatches. This is
3296 a particular problem in practice for the new sequence functions.
3297 It's not clear how this should be fixed, and for now, a
3298 workaround is given in the entry for 117 in the BUGS file.
3299 ** (Because of the interaction between the two previous items --
3300 occasional inlining problems and new inline expansions -- some
3301 of the new sequence function optimizations won't really kick in
3302 completely until debugging information, and then inlining, are
3303 straightened out in some future version.)
3304 * minor incompatible changes:
3305 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3306 DIRECTORY no longer implicitly promotes NIL slots of its
3307 pathname argument to :WILD. In particular, when you ask for the
3308 contents of a directory (which you used to be able to do without
3309 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3310 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3311 ** changes in behavior that ANSI explicitly defines to be
3312 implementation dependent:
3313 *** The new compiler-only implementation still conforms with ANSI,
3314 but acts a little different than before. Besides the obvious
3315 changes in performance tradeoffs (that the cost per form passed
3316 to EVAL has gone up, and the cost per form executed by EVAL
3317 has gone down), the behavior of the system changes a little
3318 because there are no longer any interpreted function objects.
3319 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3320 e.g. doing COMPILE on the output of interactive DEFUN is
3322 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3324 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3325 has changed from (:CL) to NIL.
3326 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3327 e.g. #\Newline and #\Tab, don't have names specified in the
3328 ANSI Common Lisp standard, is now based on their ASCII symbolic
3329 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3330 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3331 are no longer used for output.
3332 ** changes in internal implementation constants:
3333 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3334 4 million. (If your application spends a lot of time GCing and
3335 you have a lot of RAM, you might want to experiment with
3336 increasing it even more.)
3337 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3338 However, almost all old code should still continue to work without
3339 immediate update, as SB-C-CALL is now a (deprecated) nickname
3341 ** Old operator names in the style DEF-FOO are now deprecated in
3342 favor of new corresponding names DEFINE-FOO, for consistency with
3343 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3344 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3345 internal symbols, but a few supported extensions like
3346 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3347 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3348 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3349 etc. as you get deeper into recursive calls to the debugger
3350 command loop, instead of the old "5]", "5]]", "5]]]"
3351 sequence. (I was motivated to do this when squabbles between
3352 ILISP and SBCL left me very deeply nested in the debugger. In the
3353 short term, this change will probably provoke more ILISP/SBCL
3354 squabbles, but hopefully it will be an improvement in the long run.)
3355 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3356 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3357 function, since declaiming return types involving aliens is
3358 (1) annoyingly messy to do by hand and (2) vital to efficient
3359 compilation of code which calls such functions.
3360 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3361 longer reexported by the SB-EXT package. They're solely useful
3362 for alien code, so it seems more logical that you should get
3363 them from the SB-ALIEN package, not in SB-EXT.
3364 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3365 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3366 features. Instead, the code that they used to control is always
3367 built into the system.
3368 * many other bug fixes
3369 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3370 advantage of the new EVAL-WHEN stuff and to clean them up in
3371 general, and they are now more ANSI-compliant in a number of
3372 ways. Martin Atzmueller is responsible for a lot of this.
3373 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3375 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3376 *** correct ERROR type for various file operations
3377 *** some fixes for Lisp streams
3378 *** DEFMETHOD syntax checking
3379 *** changing old weird representation of debug information as
3380 strings (which, among their other deficiencies, don't transform
3381 correctly when you rename packages, and don't change their
3382 print representation when you change things like *PACKAGE*
3383 and *PRINT-LENGTH*) to symbols and lists of symbols
3384 He also made several improvements and fixed several bugs in DESCRIBE.
3385 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3387 *** misbehavior of WRITE-STRING/WRITE-LINE
3388 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3389 and several other LOOP problems as well
3390 *** DIRECTORY when similar filenames are present
3391 *** DEFGENERIC with :METHOD options
3392 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3393 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3394 *** argument ordering in FIND with :TEST option
3395 *** mishandled package designator argument in APROPOS-LIST
3396 *** various problems in the backquote readmacro
3397 *** a bug in APROPOS
3398 *** probably some others that I'm not describing very well here,
3399 since the CVS log documents them by reference to sbcl-devel
3400 messages, and the SourceForge archives aren't working well.:-(
3401 ** Dan Barlow improved the Alpha port (and is making progress on the
3402 PPC port, for those of you who think different).
3403 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3404 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3405 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3406 continues to work on the SPARC port (for those of you in a position
3407 to look down upon our little PC-compatible boxes from a great height).
3408 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3409 and uses that copy, rather than the current dynamic value, when
3410 it's trying to decide whether to truncate output. Thus e.g.
3411 (let ((*print-lines* 50))
3412 (pprint-logical-block (stream nil)
3414 (let ((*print-lines* 8))
3415 (print (aref possiblybigthings i) stream)))))
3416 should now truncate the logical block only at 50 lines, instead of
3417 often truncating it at 8 lines, as it did before.
3418 * The doc/cmucl/ directory, containing old CMU CL documentation
3419 from the time of the fork, is no longer part of the base system.
3420 SourceForge has shut down its anonymous FTP service, and with it
3421 my original plan for distributing the old CMU CL documentation
3422 there. For now, if you need these files you can download an old
3423 SBCL source release and extract them from it.
3424 * The fasl file version number changed again, for dozens of reasons,
3425 some of which are apparent above.
3427 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3428 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3429 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3430 patch, so that DISASSEMBLE output is much nicer.
3431 * The code in the SB-PROFILE package now seems reasonably stable.
3432 I still haven't decided what the final interface should look like
3433 (I'd like PROFILE to interact cleanly with TRACE, since both
3434 facilities use function encapsulation) but if you have a need
3435 for profiling now, you can probably use it successfully with
3436 the current CMU-CL-style interface.
3437 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3438 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3439 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3440 link now returns the dangling link itself, and for similar
3441 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3442 link itself. (In these cases the old code signalled an error and
3443 looped endlessly, respectively.) Thus, DIRECTORY now works even
3444 in the presence of dangling and cyclic symbolic links.
3445 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3446 is now a supported extension again, since the consensus on
3447 sbcl-devel was that it can be useful for ordinary development
3448 work, not just for debugging SBCL itself.
3449 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3450 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3451 that currently-defined functions might be redefined later with
3452 different return types.
3453 * Hash tables can be printed readably, as inspired by CMU CL code
3454 of Eric Marsden and SBCL code of Martin Atzmueller.
3455 * better error handling in CLOS method combination, thanks to
3456 Martin Atzmueller porting Pierre Mai's CMU CL patches
3457 * more overflow fixes for >16Mbyte I/O buffers
3458 * A bug in READ has been fixed, so that now a single Ctrl-D
3459 character suffices to cause end-of-file on character streams.
3460 In particular, now you only need one Ctrl-D at the command
3461 line (not two) to exit SBCL.
3462 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3464 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3465 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3466 does the right thing.
3467 * The implementation of some type tests, especially for CONDITION
3468 types, is now tidier and maybe faster, due to CMU CL code
3469 originally by Douglas Crosher, ported by Martin Atzmueller.
3470 * Some math functions have been fixed, and there are new
3471 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3472 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3473 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3474 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3475 likely to be built into sbcl-0.7.0.)
3476 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3477 CL, ported to SBCL by Martin Atzmueller)
3478 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3479 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3480 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3481 * There's a new slam.sh hack to shorten the edit/compile/debug
3482 cycle for low-level changes to SBCL itself, and a new
3483 :SB-AFTER-XC-CORE target feature to control the generation of
3484 the after-xc.core file needed by slam.sh.
3485 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3486 COMPILE-FILE is no longer supported, so that now every function
3487 gets an entry point, so that block compilation looks a little
3488 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3489 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3490 returns the number of bytes consed since the system started,
3491 rather than the number consed since the first time the function
3492 was called. (The new definition parallels ANSI functions like
3493 CL:GET-INTERNAL-RUN-TIME.)
3494 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3495 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3496 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3497 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3498 * Fasl file version numbers are now independent of the target CPU,
3499 since historically most system changes which required version
3500 number changes have affected all CPUs equally. Similarly,
3501 the byte fasl file version is now equal to the ordinary
3504 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3505 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3506 is no longer recognized. I apologize for this, because it was
3507 listed in SB-EXT as a supported extension, but I found that
3508 its existing behavior was poorly specified, as well as incorrectly
3509 specified, and it looked like too much of a mess to straighten it
3510 out. I have enough on my hands trying to get ANSI stuff to work..
3511 * many patches ported from CMU CL by Martin Atzmueller, with
3512 half a dozen bug fixes in pretty-printing and the debugger, and
3513 half a dozen others elsewhere
3514 * fixed bug 13: Floating point infinities are now supported again.
3515 They might still be a little bit flaky, but thanks to bug reports
3516 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3517 as flaky as they were.
3518 * The --noprogrammer command line option is now supported. (Its
3519 behavior is slightly different in detail from what the old man
3520 page claimed it would do, but it's still appropriate under the
3521 same circumstances that the man page talks about.)
3522 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3523 are now supported, and enabled by default. Thus, the compiler can
3524 handle many floating point and complex operations much less
3525 inefficiently. (Thus e.g. you can implement a complex FFT
3527 * The compiler now detects type mismatches between DECLAIM FTYPE
3528 and DEFUN better, and implements CHECK-TYPE more correctly, and
3529 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3530 thanks to patches from Martin Atzmueller.
3531 * various fixes to make the cross-compiler more portable to
3532 ANSI-conforming-but-different cross-compilation hosts (notably
3533 Lispworks for Windows, following bug reports from Arthur Lemmens)
3534 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3535 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3536 fixed thanks to Pierre Mai's CMU CL patches.
3537 * fixes to keep the system from overflowing internal counters when
3538 it tries to use i/o buffers larger than 16M bytes
3539 * fixed bug 45a: Various internal functions required to support
3540 complex special functions have been merged from CMU CL sources.
3541 (When I was first setting up SBCL, I misunderstood a compile-time
3542 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3543 * improved support for type intersection and union, fixing bug 12
3544 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3545 more obscure bugs as well
3546 * some steps toward byte-compiling non-performance-critical
3547 parts of the system, courtesy of patches from Martin Atzmueller
3548 * Christophe Rhodes has made some debian packages of sbcl at
3549 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3550 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3551 stable, but are nonetheless usable. When he's ready, I'd be happy
3552 to add them to the SourceForge "File Releases" section. (And if
3553 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3554 * new fasl file format version number (because of changes in
3555 internal representation of (OR ..) types to accommodate the new
3556 support for (AND ..) types, among other things)
3558 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3559 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3561 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3563 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3564 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3565 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3566 * DESCRIBE now works on structure objects again.
3567 * Most function call argument type mismatches are now handled as
3568 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3569 know whether the function will be redefined before the call is
3570 executed. (The compiler could flag local calls with full WARNINGs,
3571 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3572 it doesn't keep track of enough information to know whether calls
3573 are local in this sense.)
3574 * Compiler output is now more verbose, with messages truncated
3575 later than before. (There should be some supported way for users
3576 to override the default verbosity, but I haven't decided how to
3577 provide it yet, so this behavior is still controlled by the internal
3578 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3579 src/compiler/ir1util.lisp.)
3580 * Fasl file format version numbers have increased again, because
3581 support for the Gray streams extension changes the layout of the
3582 system's STREAM objects.
3583 * The Gray subclassable streams extension now works, thanks to a
3584 patch from Martin Atzmueller.
3585 * The full LOAD-FOREIGN extension (not just the primitive
3586 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3587 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3588 but like most other programs, it defaults to copying the Unix
3589 environment from the original process instead of starting the
3590 new process in an empty environment.
3591 * Extensions which manipulate the Unix environment now support
3592 an :ENVIRONMENT keyword option which doesn't smash case or
3593 do other bad things. The CMU-CL-style :ENV option is retained
3594 for porting convenience.
3595 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3596 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3598 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3600 * A patch from Martin Atzmueller seems to have solved the SIGINT
3601 problem, and as far as we know, signal-handling now works cleanly.
3602 (If you find any new bugs, please report them!)
3603 * The system no longer defaults Lisp source file names to types
3604 ".l", ".cl", or ".lsp", but only to ".lisp".
3605 * The compiler no longer uses special default file extensions for
3606 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3607 seems to expect a single default extension for all compiled code,
3608 and there's no compelling reason to try to stretch the standard
3609 to allow two different extensions.) Instead, byte-compiled files
3610 default to the same extension as native-compiled files.
3611 * Fasl file format version numbers have increased again, because
3612 a rearrangement of internal implementation packages made some
3613 dumped symbols in old fasl files unreadable in new cores.
3614 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3615 many fewer weird special cases.
3616 * Bug #17 (differing COMPILE-FILE behavior between logical and
3617 physical pathnames) has been fixed, and some related misbehavior too,
3618 thanks to a patch from Martin Atzmueller.
3619 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3620 by Tim Moore, ported to SBCL by Martin Atzmueller.
3621 * Martin Atzmueller fixed several filesystem-related problems,
3622 including bug #36, in part by porting CMU CL patches, which were
3623 written in part by Paul Werkowski.
3624 * More compiler warnings in src/runtime/ are gone, thanks to
3625 more patches from Martin Atzmueller.
3626 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3629 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3631 * DESCRIBE now works on CONDITION objects.
3632 * The debugger now handles errors which arise when trying to print
3633 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3635 * The build system now uses an additional file, customize-target-features.lisp,
3636 to allow local modifications to the target *FEATURES* list. (The point of
3637 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3638 debugging features enabled, without having to worry about propagating it
3639 into everyone's system when I do a "cvs update".) When no
3640 customize-target-features.lisp file exists, the target *FEATURES* list
3641 should be constructed the same way as before.
3642 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3643 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3644 warn about behavior which is undefined under the ANSI spec. Note: This
3645 is specified by ANSI, but it's not very popular with programmers.
3646 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3647 macro in the SBCL sources for an example of a workaround which you
3648 might use to make portable ANSI-standard code which does what you want.
3649 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3650 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3651 when it's used as a non-toplevel form. (This is required in order
3652 to implement the DEFCONSTANT-EQX macro.)
3653 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3654 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3655 funny things, due to bugs in the implementation of EVAL-WHEN
3656 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3657 non-toplevel DEFCONSTANTs are uncommon.)
3658 * The core file version number and fasl file version number have been
3659 incremented, because the old noncompliant DEFCONSTANT behavior involved
3660 calling functions which no longer exist, and because I also took the
3661 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3662 * fixed bug 1 (error handling before read-eval-print loop starts), and
3663 redid debugger restarts and related debugger commands somewhat while
3665 ** The QUIT debugger command is gone, since it did something
3666 rather different than the SB-EXT:QUIT command, and since it never
3667 worked properly outside the main toplevel read/eval/print loop.
3668 Invoking the new TOPLEVEL restart provides the same functionality.
3669 ** The GO debugger command is also gone, since you can just invoke
3670 the CONTINUE restart directly instead.
3671 ** The TOP debugger command is also gone, since it's redundant with the
3672 FRAME 0 command, and since it interfered with abbreviations for the
3674 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3676 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3677 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3678 some porting work by Martin Atzmueller)
3679 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3680 with a boatload of patches which he ported to SBCL. Now that those
3682 ** The system tries to make sure that its low-priority messages
3683 are prefixed by semicolons, to help people who like to use
3684 syntax highlighting in their ILISP buffer. (This patch
3685 was originally due to Raymond Toy.)
3686 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3687 patches originally written by Raymond Toy.
3688 ** The compiler understands coercion between single-value and
3689 multiple-VALUES type expressions better, getting rid of some very
3690 weird behavior, thanks to patches originally by Robert MacLachlan
3691 and Douglas Crosher.
3692 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3693 lambda lists, thanks to a patch originally by Pierre Mai.
3694 ** The system no longer bogusly warns about "abbreviated type
3696 ** The compiler gets less confused by inlining and RETURN-FROM,
3697 thanks to some patches originally by Tim Moore.
3698 ** The system no longer hangs when dumping circular lists to fasl
3699 files, thanks to a patch originally from Douglas Crosher.
3700 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3701 undefined function error.
3702 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3703 in structure slot initforms) for now, documented workaround instead:-|
3704 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3705 * fixed bug 5: added stubs for various Gray stream functions called
3706 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3707 installed, at least appropriate type errors are generated
3708 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3709 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3710 consistently in DEFMETHOD forms.
3711 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3712 it doesn't seem to affect SBCL after all
3713 * The C runtime system now builds with better optimization and many
3714 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3716 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3718 * The system is now under CVS at SourceForge (instead of the
3719 CVS repository on my home machine).
3720 * The new signal handling code has been tweaked to treat register
3721 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3722 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3723 downstream, e.g. in debug-int.lisp, has implicit dependencies
3724 on the unsignedness of integer representation of machine words,
3725 and that caused the system to bomb out with infinite regress
3726 when trying to recover from type errors involving signed values,
3727 e.g. (BUTLAST '(1 2 3) -1).)
3728 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3729 (This was one of the bugs Peter Van Eynde reported back in July.)
3730 * The system now uses code inspired by Colin Walters' O(N)
3731 implementation of MAP (from the cmucl-imp@cons.org mailing
3732 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3733 inline the MAP operation, and there is more than one
3734 sequence argument to the MAP call (so that it can't just
3735 do ETYPECASE once and for all based on the type of the
3736 single sequence argument). (The old non-inline implementation
3737 of the general M-argument sequence-of-length-N case required
3738 O(M*N*N) time when any of the sequence arguments were LISTs.)
3739 * The QUIT :UNIX-CODE keyword argument has been renamed to
3740 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3741 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3743 * Raymond Wiker's patches to port dynamic loading from Linux to
3744 FreeBSD have been added.
3745 * The BUGS file is now more nearly up to date, thanks in large part
3746 to Martin Atzmueller's review of it.
3747 * The debugger now flushes standard output streams before it begins
3748 its output ("debugger invoked" and so forth).
3749 * The core version number and fasl file version number have both
3750 been incremented, because of incompatible changes in the layout
3752 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3753 used to not be called for a saved Lisp image.)
3754 * Martin Atzmueller's version of a patch to fix a compiler crash,
3755 as posted on sbcl-devel 13 September 2000, has been installed.
3756 * Instead of installing Martin Atzmueller's patch for the
3757 compiler transform for SUBSEQ, I deleted the compiler transform,
3758 and transforms for some similar consing operations.
3759 * A bug in signal handling which kept TRACE from working on OpenBSD
3761 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3762 away by constant folding
3763 * The system now defines its address space constants in one place
3764 (in the Lisp sources), and propagates them automatically elsewhere
3765 (through GENESIS and the sbcl.h file). Therefore, patching the
3766 address map is less unnecessarily tedious and error-prone. The
3767 Lisp names of address space constants have also been systematized.
3768 * CVS tags like dollar-Header-dollar have been removed from
3769 the sources, because they have never saved me trouble and
3770 they've been source of trouble working with patches and other
3771 diff-related operations.
3772 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3773 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3775 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3777 * The system has been ported to OpenBSD.
3778 * The system now compiles with a simple "sh make.sh" on the systems
3779 that it's supported on. I.e., now you no longer need to tweak
3780 text in the target-features.lisp-expr and symlinks in src/runtime/
3781 by hand, the make.sh takes care of it for you.
3782 * The system is no longer so grossly inefficient when compiling code
3783 involving vectors implemented as general (not simple) vectors (VECTOR T),
3784 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3785 which dares to use the various sequence functions on non-simple
3786 vectors, takes less of a performance hit.
3787 * There is now a primitive type predicate VECTOR-T-P
3788 to test for the (VECTOR T) type, so that e.g.
3789 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3790 can now be compiled with some semblance of efficiency. (The old code
3791 turned the type declaration into a full call to %TYPEP at runtime!)
3792 * AREF on (VECTOR T) is still not fast, since it's still compiled
3793 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3794 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3795 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3796 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3797 without making full calls to SUBTYPEP at runtime.
3798 (Some analogous efficiency issues for non-simple vectors specialized to
3799 element types other than T, or for non-simple multidimensional arrays,
3800 have not been addressed. They could almost certainly be handled the
3801 same way if anyone is motivated to do so.)
3802 * The changes in array handling break binary compatibility, so
3803 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3804 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3805 instead of (VALUES T T).
3806 * By following the instructions that Dan Barlow posted to sbcl-devel
3807 on 2 July 2000, I was able to enable primitive dynamic object
3808 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3809 functionality is not implemented (since it calls ld to resolve
3810 library references automatically, requiring RUN-PROGRAM for its
3811 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3812 try to resolve library references) is now supported.
3813 * The system now flushes the standard output streams when it terminates,
3814 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3815 them at several other probably-convenient times, e.g. in each pass of
3816 the toplevel read-eval-print loop, and after evaluating a form given
3817 as an "--eval" command-line option. (These changes were motivated by a
3818 discussion of stream flushing issues on cmucl-imp in August 2000.)
3819 * The source transform for TYPEP of array types no longer assumes
3820 that an array whose element type is a not-yet-defined type
3821 is implemented as an array of T, but instead punts, so that the
3822 type will be interpreted at runtime.
3823 * There is now some support for cross-compiling in make.sh: each of
3824 the phases of make.sh has its own script. (This should be transparent
3825 to people doing ordinary, non-cross-compile builds.)
3826 * Since my laptop doesn't have hundreds of megabytes of memory like
3827 my desktop machine, I became more motivated to do some items on
3828 my to-do list in order to reduce the size of the system a little:
3829 ** Arrange for various needed-only-at-cold-init things to be
3830 uninterned after cold init. To support this, those things have
3831 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3832 symbols with such names are now uninterned after cold init).
3833 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3834 fasl files for cold load.
3835 ** Remove the old compiler structure pooling code (which used to
3836 be conditional on the target feature :SB-ALLOC) completely.
3837 ** Redo the representation of some data in cold init to be more compact.
3838 (I also looked into supporting byte compiled code at bootstrap time,
3839 which would probably reduce the size of the system a lot, but that
3840 looked too complicated, so I punted for now.)
3841 * The maximum signal nesting depth in the src/runtime/ support code has
3842 been reduced from 4096 to 256. (I don't know any reason for the very
3843 large old value. If the new smaller value turns out to break something,
3844 I'll probably just bump it back up.)
3845 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3847 * Many, many bugs reported by Peter Van Eynde have been added to
3848 the BUGS list; some have even been fixed.
3849 * While enabling dynamic object file loading, I tried to make the
3850 code easier to understand, renaming various functions and variables
3851 with less ambiguous names, and changing some function calling
3852 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3853 * While trying to figure out how to do the OpenBSD port, I tried to
3854 clean up some of the code in src/runtime/. In particular, I dropped
3855 support for non-POSIX signal handling, added various comments,
3856 tweaked the code to reduce the number of compilation warnings, and
3857 renamed some files to increase consistency.
3858 * To support the new automatic configuration functionality in make.sh,
3859 the source file target-features.lisp-expr has been replaced with the
3860 source file base-target-features.lisp-expr and the machine-generated
3861 file local-target-features.lisp-expr.
3862 * fixed a stupid quoting error in make.sh so that using CMU CL
3863 "lisp -batch" as cross-compilation host works again
3865 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3867 * DESCRIBE no longer tries to call itself recursively to describe
3868 bound/fbound values, so that it no longer fails on symbols which are
3869 bound to themselves (like keywords, T, and NIL).
3870 * DESCRIBE now works on generic functions.
3871 * The printer now prints less-screwed-up representations of closures
3872 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3873 * A private symbol is used instead of the :EMPTY keyword previously
3874 used to mark empty slots in hash tables. Thus
3875 (DEFVAR *HT* (MAKE-HASH-TABLE))
3876 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3877 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3878 now does what ANSI says that it should. (You can still get
3879 similar noncompliant behavior if bang on the hash table
3880 implementation with all the symbols you get back from
3881 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3882 This breaks binary compatibility, since tests for equality to
3883 :EMPTY are wired into things like the macroexpansion of
3884 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3886 * There's now a minimal placeholder implementation for CL:STEP,
3887 as required by ANSI.
3888 * An obscure bug in the interaction of the normal compiler, the byte
3889 compiler, inlining, and structure predicates has been patched
3890 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3891 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3893 * Missing ordinary arguments in a macro call are now detected even
3894 when the macro lambda list contains &KEY or &REST.
3895 * The debugger no longer complains about encountering the top of the
3896 stack when you type "FRAME 0" to explicitly instruct it to go to
3897 the top of the stack. And it now prints the frame you request even
3898 if it's the current frame (instead of saying "You are here.").
3899 * As specified by ANSI, the system now always prints keywords
3900 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3902 * The default initial SIZE of HASH-TABLEs is now smaller.
3903 * Type information from CLOS class dispatch is now propagated
3904 into DEFMETHOD bodies, so that e.g.
3905 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3907 is now basically equivalent to
3908 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3909 (DECLARE (TYPE SINGLE-FLOAT X))
3911 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3912 operation, without having to do run-time type dispatch.
3913 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3914 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3915 * Since I don't seem to be making big file reorganizations very often
3916 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3917 is overflowing my ability to conveniently back them up), I've finally
3918 checked the system into CVS. (The CVS repository is on my home system,
3919 not at SourceForge -- putting it on SourceForge might come later.)
3920 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3921 high-level GC-NOTIFY-FOO functions send their output. (There's
3922 still very little control of where low-level verbose GC functions
3923 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3924 less than it used to -- the GC-NOTIFY-FOO functions are now under
3925 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3926 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3927 in only one place in the source code, and propagates it automatically
3928 everywhere that it's needed. Thus e.g. when I bump the version from
3929 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3930 * The C source files now include boilerplate legalese and documentation
3931 at the head of each file (just as the Lisp source files already did).
3932 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3933 to his page will be replaced with a link to his new CLiki service.
3935 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3937 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3938 * The build process now looks for GNU make under the default name "gmake",
3939 instead of "make" as it used to. If GNU make is not available as "gmake"
3940 on your system, you can change this default behavior by setting the
3941 GNUMAKE environment variable.
3942 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3943 can build without error under CMU CL.
3945 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3947 * There is now a partial SBCL user manual (with some new text and some
3948 text cribbed from the CMU CL manual).
3949 * The beginnings of a profiler have been added (starting with the
3950 CMU CL profiler and simplifying and cleaning up). Eventually the
3951 main interface should be through the TRACE macro, but for now,
3952 it's still accessed through vaguely CMU-CL-style functions and macros
3953 exported from the package SB-PROFILE.
3954 * Some problems left over from porting CMU CL to the new
3955 cross-compilation bootstrap process have been cleaned up:
3956 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3957 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3958 part when getting rid of the compiler *BACKEND* stuff.)
3959 ** The value of *NULL-TYPE* was screwed up, because it was
3960 being initialized before the type system knew the final
3961 definition of the 'NULL type. This screwed up several key
3962 optimizations in the compiler, causing inefficiency in all sorts
3963 of places. (I found it because I wanted to understand why
3964 GET-INTERNAL-RUN-TIME was consing.)
3965 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3966 PROCLAIM FTYPE information. Unfortunately this broke binary
3967 compatibility again, since now the forms output by DEFGENERIC
3968 to refer to functions which didn't exist in 0.6.3.
3969 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3970 can use the new (as of 0.6.3) transform for SOME into MAP into
3972 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3973 values to (INTEGER 0 1000000), so that the time code will no longer
3974 occasionally get blown up by Linux returning 1000000 microseconds
3975 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3976 its output conform to the ANSI spec. (Alas, this makes its output
3977 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3978 * A full call to MAP NIL with a single sequence argument no longer conses.
3979 * fixes to problems pointed out by Martin Atzmueller:
3980 * The manual page no longer talks about multiprocessing as though
3981 it were currently supported.
3982 * The ILISP support patches have been removed from the distribution,
3983 because as of version 5.10.1, ILISP now supports SBCL without us
3984 having to maintain patches.
3985 * added a modified version of Raymond Toy's recent CMU CL patch for
3986 EQUALP comparison of HASH-TABLE
3988 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3990 * The system still can't cross-compile itself with
3991 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3992 entails), but at least it can compile more of itself that way
3993 than it used to be able to, and various buglets which were uncovered
3994 by trying to cross-compile itself that way have now been fixed.
3995 * This release breaks binary compatibility again. This time
3996 at least I've incremented the FASL file format version to 2, so that the
3997 problem can be detected reliably instead of just causing weird errors.
3998 * various new style warnings:
3999 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4000 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4001 ** using the lexical binding of a variable named in the *FOO* style
4002 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4003 as specified by ANSI.
4004 * *RANDOM-STATE* is no longer automatically initialized from
4005 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4006 default behavior of the system is to repeat its behavior every time
4007 it's run. If you'd like to change this behavior, you can always
4008 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4009 old convention there was no comparably easy way to get the system to
4010 repeat its behavior every time it was run.
4011 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4012 FTYPE declarations has been removed, in favor of their ANSI
4013 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4014 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4015 the types of their sequence arguments can be determined at compile time.
4016 This is done through a new open code expansion for MAP which eliminates
4017 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4018 argument types can be determined at compile time.
4019 * The optimizer now transforms COERCE into an identity operation when it
4020 can prove that the coerced object is already of the correct type. (This
4021 can be a win for machine generated code, including the output of other
4022 optimization transforms, such as the MAP transform above.)
4023 * Credit information has been moved from source file headers into CREDITS.
4024 * Source file headers have been made more standard.
4025 * The CASE macro now compiles without complaining even when it has
4028 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4030 * (Note that the way that the PCL macroexpansions were rewritten
4031 to accommodate the change in DEFGENERIC below breaks binary
4032 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4033 not run under sbcl-0.6.2. Once we get out of alpha releases,
4034 i.e. hit release 1.0.0, we'll probably try to maintain binary
4035 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4036 and sbcl-1.4.4. Until then, however, it might be fairly common
4037 for maintenance releases to break binary compatibility.)
4038 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4040 * The reporting of unhandled signals has been changed to print some
4041 explanatory text as well as the report form. (Previously only
4042 the report form was printed.)
4043 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4044 it defines, so that the compiler no longer issues undefined function
4045 warnings for compiled-but-not-yet-loaded generic functions.
4046 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4047 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4048 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4049 section of the ANSI spec are supported.
4050 * The "" nickname for the "KEYWORD" package has been removed.
4051 The reader still handles symbol tokens which begin with a package marker
4052 as keywords, but it doesn't expose its mechanism for doing so in the
4053 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4054 * The system now issues STYLE-WARNINGs for contradictory TYPE
4055 proclamations. (Warnings for contradictory FTYPE proclamations would
4056 be nice too, but those can't be done usefully unless the type system
4057 is made smarter about FUNCTION types.)
4058 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4059 systematized, so that "*target-*.lisp is supposed to exist only on the
4060 target and imply that there's a related file which exists on the
4061 host, and *host-*.lisp is supposed to exist only on the host and imply
4062 that there's a related file which exists on the target. This involves a
4063 lot of renaming. Hopefully the acute confusion caused by the renaming
4064 will be justified by the reduction in chronic confusion..
4065 ** runtime-type.lisp -> early-target-type.lisp
4066 ** target-type.lisp -> late-target-type.lisp
4067 ** early-host-format.lisp -> early-format.lisp
4068 ** late-host-format.lisp -> late-format.lisp
4069 ** host-error.lisp -> misc-error.lisp
4070 ** early-error.lisp -> early-target-error.lisp
4071 ** late-error.lisp -> late-target-error.lisp
4072 ** host-defboot.lisp -> early-defboot.lisp
4073 ** code/misc.lisp -> code/target-misc.lisp
4074 ** code/host-misc.lisp -> code/misc.lisp
4075 ** code/numbers.lisp -> code/target-numbers.lisp
4076 ** code/early-numbers.lisp -> numbers.lisp
4077 ** early-host-type.lisp -> early-type.lisp
4078 ** late-host-type.lisp -> late-type.lisp
4079 ** host-typep.lisp -> typep.lisp
4080 ** load.lisp -> target-load.lisp
4081 ** host-load.lisp -> load.lisp
4082 ** host-disassem.lisp -> disassem.lisp
4083 ** host-insts.lisp -> insts.lisp
4084 ** byte-comp.lisp -> target-byte-comp.lisp
4085 ** host-byte-comp.lisp -> byte-comp.lisp
4086 ** host-signal.lisp -> signal.lisp
4087 ** host-defstruct.lisp -> defstruct.lisp
4088 ** late-target-type.lisp -> deftypes-for-target.lisp
4089 Furthermore, several other previously target-only files foo.lisp (e.g.
4090 hash-table.lisp and random.lisp) have been split into a target-and-host
4091 foo.lisp file and a target-only target-foo.lisp file, with their key type
4092 definitions in the target-and-host part, so that the cross-compiler will
4093 know more about target types.
4094 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4095 gone away. In their place are various *BACKEND-FOO* variables
4096 corresponding to the slots of the old structure.
4097 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4098 into the target SBCL has been fixed.
4099 * The chill.lisp system for loading cold code into a running SBCL
4101 * Support for the CMU CL "scavenger hook" extension has been removed.
4102 (It was undocumented and unused in the CMU CL sources that SBCL was
4103 derived from, and stale in sbcl-0.6.1.)
4104 * Various errors in the cross-compiler type system were detected
4105 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4106 (enabling various consistency checks). Many of them were fixed,
4107 but some hard problems remain, so the compiler is back to
4108 running without *TYPE-SYSTEM-INITIALIZED* for now.
4109 * As part of the cross-compiler type system cleanup, I implemented
4110 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4111 * I have started adding UNCROSS calls throughout the type system
4112 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4113 on cross-compiler input files will be able to go away, and various
4115 * CONSTANTP now returns true for quoted forms (as explicitly required
4118 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4120 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4121 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4122 (MAKE-LIST -1). In the longer term, it ought to become true
4123 that declarations are assertions even at SAFETY 1. For now, it's not
4124 quite true even at SAFETY 3, but it's at least more nearly true..
4125 (Note that this change seems to increases the size of the system by
4126 O(5%) and to decrease the speed of the compiler by 20% or more.)
4127 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4128 for the problem of printing dozens of lines of distracting information
4129 about low-level system machinery as part of the top stack frame
4130 on entry to the debugger when an undefined function was called.
4131 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4132 is not reset to COMMON-LISP-USER.
4133 * Compilation of stuff related to dyncount.lisp has been made conditional
4134 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4135 smaller. The various dyncount-related symbols have been moved into
4136 a new "SB-DYNCOUNT" package.
4137 * tty-inspect.lisp has been renamed to inspect.lisp.
4138 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4139 feature has gone away. (When we eventually port to other flavors of
4140 libc and/or Unix, we'll try to make the differences between flavors
4141 invisible at the user level.)
4142 * Various other *FEATURES* tags, and/or their associated conditionals,
4143 have been removed if obsolescent, or given better documentation, or
4144 sometimes given more-mnemonic names.
4146 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4148 * tidied up "make.sh" script
4149 * tidied up system directory structure
4150 * better "clean.sh" behavior
4151 * added doc/FOR-CMUCL-DEVELOPERS
4152 * many many small tweaks to output format, e.g. removing possibly-confusing
4153 trailing #\. character in DESCRIBE-INSTANCE
4154 * (EQUALP #\A 'A) no longer signals an error.
4155 * new hashing code, including EQUALP hashing
4156 * tidied up Lisp initialization and toplevel
4157 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4158 * command line argument processing
4159 * added POSIX-GETENV function to deal with Unix-ish environment variables
4160 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4161 terminating SBCL on EOF
4162 * non-verbose GC by default
4163 * There is no more "sbcl" shell script; the sbcl file is now the C
4164 runtime executable (just like CMU CL).
4165 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4167 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4168 symbol and package information as Lisp native symbol and package objects
4169 instead of strings naming symbols and strings naming packages. This way,
4170 whenever packages are renamed (as in warm init), debug information is
4171 transformed along with everything else.
4172 * tweaked the optimization policy declarations which control the building
4173 of SBCL itself. Now, among other things, the system no longer saves
4174 source location debugging information. (This helps two problems at once
4175 by reducing SBCL size and by keeping SBCL from trying to look for its
4176 sources -- which may not exist -- when reporting errors.)
4177 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4178 debugging and testing purposes
4179 * cleaned up printing, making the printer call PRINT-OBJECT for
4180 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4181 methods, giving nearly-ANSI behavior
4182 * converted almost all special variables to use *FOO* naming convention
4183 * deleted PARSE-TIME functionality, since it can be done portably
4184 * moved some files out of cold init into warm init
4185 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4187 * regularized formatting of source files
4188 * added an install.sh script
4189 * fixed ridiculous memory usage of cross-compiler by making
4190 compiler/alloc.lisp not try to do pooling unless it can hook
4191 itself into the GC of the cross-compilation host. Now the system
4192 builds nicely on my old laptop.
4193 * added :SB-ALLOC in target-features.lisp-expr
4194 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4195 was not implemented)
4196 * re-did condition handling and note reporting in the compiler. Notes
4197 are no longer handled by signalling conditions. Style warnings
4198 and warnings are handled more correctly and reported in such a way
4199 that it's easy to find one or the other in your output (so that you
4200 can e.g. figure out which of many problems caused COMPILE-FILE to
4202 * changed the severity of several compiler warnings from full WARNING
4203 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4204 compiler note reporting so that it doesn't use the condition system
4205 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4206 COMPILE-FILE command)
4207 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4208 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4209 no longer supported, since their implementation was deeply intertwingled
4210 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4211 * removed bogus "support" for compiler macros named (SETF FOO), and
4212 removed the compiler macro for SETF INFO (but only after making a fool
4213 of myself on the cmucl-imp mailing list by posting a bogus patch for
4214 DEFINE-COMPILER-MACRO..)
4215 * Compiled files containing forms which have side effects on the Lisp
4216 reader (such as DEFPACKAGE forms) are now handled more correctly.
4217 (Compiler queuing of top level lambdas has been suppressed by setting
4218 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4219 * deleted various currently-unused source files, e.g. gengc.lisp. They
4220 may be added back at some point e.g. when porting to other architectures,
4221 but until they are it's distracting to distribute them and to try to
4223 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4224 there were so many of them they're just distractions, and UNCROSS is
4225 known to be able to handle the current sources
4226 * moved PROFILE functionality into TRACE, so that it will be clear
4227 how the wrapping and unwrapping of functions when you profile them
4228 interacts with the wrapping and unwrapping of functions when you
4229 trace them. (Actually, the functionality isn't there yet, but at least
4230 the interface specification is there. Hopefully, the functionality will
4231 arrive with some future maintenance release.)
4232 * removed host-oops.lisp
4233 * changed signature of QUIT function to allow UNIX-CODE argument
4234 * fixed READ-SEQUENCE bug
4235 * tweaked verbose GC output so that it looks more like the progress
4236 output that ANSI specifies for functions like LOAD
4237 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4238 * added <http://sbcl.sourceforge.com> to the banner information printed by