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 * bug fix: / with an unused value was being deleted in safe code.
16 (thanks to Marco Monteiro and Kevin Reid)
17 * bug fix: number of characters that can be written onto a single
18 line in a file is unlimited.
19 * bug fix: some GC deadlocks caused by asynchronous interrupts have
20 been fixed by inhibiting interrupts for when GC is disbled.
21 * bug fix: GETHASH, PUTHASH, CLRHASH and REMHASH are now interrupt safe.
22 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
23 specifier no longer causes infinite recursion.
24 * improvement: the x86-64/darwin port now passes all tests and
25 should be considered non-experimental.
27 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
28 * new platform: experimental support for x86-64/darwin (MacOS).
29 * incompatible change: the thread-safe (on most platforms) getaddrinfo
30 and getnameinfo sockets functions are used instead of gethostbyaddr
31 and gethostbyname, on platforms where the newer functions are available.
32 As a result, the ALIASES field of HOST-ENT will always be NIL on these
34 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
35 don't print the startup banner, but behave as if --noinform was passed
36 as a command line argument. (thanks to Kevin Reid)
37 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
38 * optimization: code using alien values with undeclared types is much faster.
39 * optimization: the compiler is now able to open code SEARCH in more cases.
40 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
41 * bug fix: using standardized COMMON-LISP special variables as loop
42 variables no longer signals bogus package lock violations. (reported
44 * bug fix: declaring local loop variables to be of a range-limited type
45 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
46 (reported by Andras Simon)
47 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
48 bugs remain on x86-64.)
49 * bug fix: the #= and ## reader macros now interact reasonably with
50 funcallable instances.
51 * bug fix: type-checks for function arguments were compiled using the
52 compiler policy settings of the wrong lexical environment (in 1.0.2
54 * bug fix: SHADOW accepts characters as string designators, as required
55 by the spec (thanks to Eric Marsden)
56 * bug fix: fixed GC safety issues when foreign functions are called with
57 non-base strings as arguments
58 * bug fix: more consistent error output for fatal-compiler-warnings, like
60 * bug fix: functions with proper names were showing up as NIL in
61 backtraces (reported by Edi Weitz, regression in 1.0.2)
63 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
64 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
65 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
66 produce more efficient code on the x86 if the compiler can determine
67 that (+ INDEX <constant>) does not require a bounds check and FOO
68 has an element type at least 8 bits wide.
69 * bug fix: references to undefined variables are handled the same way
70 in toplevel forms as in the normal compiler
71 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
73 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
74 (thanks to Jon Buller)
75 * improvement: faster compilation times for complex functions
76 * improvement: added readlink support to SB-POSIX (thanks to Richard
79 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
80 * improvement: experimental support for mach exception handling on
81 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
83 * improvement: support for GBK external format.
84 (thanks to Chun Tian (binghe))
85 * improvement: the debugger now displays variables that have been closed
86 over, in code compiled with (DEBUG 2) or higher
87 * improvement: support for executable cores on NetBSD (thanks to
89 * new feature: added a RESTART-FRAME debugger command
90 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
91 be used to provide an implementation for FILE-POSITION on Gray streams
92 (thanks to Eric Marsden)
93 * new feature: add syslog, openlog and closelog support to SB-POSIX
94 (thanks to Richard Kreuter)
95 * optimization: the function call overhead in code compiled with
96 a high DEBUG optimization setting is significantly
97 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
98 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
99 (thanks to Tony Martinez)
100 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
101 evaluated once (reported by Kevin Reid)
102 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
103 works again (thanks to Kilian Sprotte)
104 * bug fix: an error is signaled for tagbodies with duplicate tags
105 (thanks to Stephen Wilson)
106 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
107 * bug fix: Win32 port can now handle foreign code unwinding Lisp
108 stack frames from alien callbacks.
109 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
110 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
111 65k pages (thanks to David Woodhouse)
112 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
114 changes in sbcl-1.0.1 relative to sbcl-1.0:
115 * new platform: FreeBSD/x86-64, including support for threading.
116 * new feature: the compiler stores cross-referencing information
117 abount function calls (who-calls), macroexpansion (who-macroexpands)
118 and special variables (who-binds, who-sets, who-references) for code
119 compiled with (< SPACE 3). This information is available through the
120 sb-introspect contrib.
121 * new feature: users may subclass SEQUENCE, and have instances of
122 these classes interoperate with standard Common Lisp functions if
123 a number of methods are defined. (This feature is experimental
124 and the interface subject to change based on feedback from SBCL
125 users and the general community)
126 * improvement: sb-sprof traces call stacks to an arbitrary depth on
127 x86/x86-64, rather than the previous fixed depth of 8
128 * improvement: another pthread back-end of mutex "pthread-futex".
129 * bug fix: non-ascii command-line arguments are processed correctly
130 (thanks to Yaroslav Kavenchuk)
131 * bug fix: non-required arguments were not passed correctly when a method
132 defined using DEFMETHOD was called from a mop-generated method using
133 CALL-NEXT-METHOD (reported by Pascal Costanza)
134 * bug fix: recursion is now permitted in accessors through
135 SLOT-UNBOUND. (reported by Pascal Costanza)
136 * bug fix: an error was signaled at startup if the HOME environment
137 variable was defined, but had an empty value (reported by Peter Van Eynde)
138 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
139 signal a PROGRAM-ERROR, not a BUG.
140 * bug fix: SB-POSIX:READDIR works when built with large file support.
141 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
143 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
144 for better automatic indentation support. (thanks to Matt Pillsbury)
145 * optimization: loading generic functions no longer takes O(n^2) time,
146 proportional to the amount of methods in the generic function
147 (reported by Todd Sabin and Jeremy Brown)
148 * optimization: the FIND and POSITION family of sequence functions
149 are significantly faster on arrays whose element types have been
151 * improvements to the Windows port:
152 ** Intermittent heap corruption problems have been fixed. (thanks
153 to Alastair Bridgewater)
154 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
156 ** Lisp is able to unwind foreign exception frames from alien
157 callbacks. (thanks to Alastair Bridgewater)
159 changes in sbcl-1.0 relative to sbcl-0.9.18:
160 * improvement: experimental support for threading on FreeBSD/x86.
161 (thanks to NIIMI Satoshi)
162 * improvement: runtime option --dynamic-space-size can be used
163 to set the size of the dynamic space reserved on startup.
164 * improvement: floating point modes in effect are now saved in
165 core, and restored on startup.
166 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
167 startup, not time since first call to GET-INTERNAL-REAL-TIME.
168 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
169 threads are not running after *SAVE-HOOKS* have run.
170 * improvement: writes to CLOS instance slots are type-checked in code
171 compiled with (SAFETY 3)
172 * improvement: floating-point exception handling on FreeBSD (thanks to
174 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
175 (thanks to Zach Beane)
176 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
178 * improvement: added support for the Shift-JIS external format.
179 (contributed by NIIMI Satoshi)
180 * improvement: callbacks are supported on Linux/PPC. (thanks to
182 * bug fix: compiler bug triggered by a (non-standard) VALUES
183 declaration in a LET* was fixed. (reported by Kaersten Poeck)
184 * bug fix: file compiler no longer confuses validated and already
185 dumped structurres. (reported by Kaersten Poeck)
186 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
187 fill-pointer signals a type-error as required. (thanks to
189 * bug fix: disassemly of funcallable instances works.
190 * bug fix: single stepping on PPC.
191 * bug fix: fix thread-safety problems in the type system (generally
192 manifesting as nonsensical errors like "STRING is a bad type specifier
193 for sequences" or "The value 1 is not of type FIXNUM")
194 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
195 (reported by Josip Gracin)
196 * bug fix: an error is signaled for attempts to displace arrays with
197 incompatible element types (thanks to Mario Mommer)
198 * bug fix: more correct handling of wide characters in debug info
199 (bug reported by Attila Lendvai and fixed by Juho Snellman)
200 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
201 and don't cause extra consing
202 * optimization: MAP and MAP-INTO are significantly faster on vectors
203 whose elements types have been declared.
204 * Improvements to SB-SPROF:
205 ** Support for allocation profiling
206 ** Reduced profiling overhead, especially for long profiling runs
207 * Improvements to the Windows port:
208 ** floating point exceptions are now reported correctly.
209 ** stack exhaustion detection works partially.
210 ** more accurate GET-INTERNAL-REAL-TIME.
211 ** better RUN-PROGRAM behaviour in case of errors in the child
213 ** PROBE-FILE now simplifies pathnames correctly.
214 ** DIRECTORY now works correctly with :WILD-INFERIORS.
215 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
216 1st 1970 (although time timezone data will be unreliable).
217 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
219 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
220 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
221 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
223 * bug fix: two potential GC deadlocks affecting threaded builds.
224 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
225 returns T (reported by Anton Kazennikov)
226 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
227 with non-variable places
228 * bug fix: remove a race condition in the setting of
229 funcallable-instance functions, this should make threaded CLOS
230 code more stable against memory faults.
231 * bug fix: corruption of specials when unbinding is interrupted by an
232 asynchronous unwind (reported by Hannu Koivisto)
233 * improvement: the debugger will now also display local variables that
234 are only used once, for code compiled with a DEBUG optimization quality
237 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
238 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
239 * incompatible change: External-format support for FFI calls. The
240 SB-ALIEN:C-STRING no longer implies an ASCII
241 external-format. Instead the string is subject to external-format
242 conversion. Additionally return values of type SB-ALIEN:C-STRING
243 are by default of type (SIMPLE-ARRAY CHARACTER), not
244 SIMPLE-BASE-STRING. For an alien type that behaves like the old
245 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
246 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
247 * incompatible change: SB-EXT package no longer contains the
248 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
249 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
251 * incompatible change: the single-stepper is no longer available
252 on Alpha, Mips and Sparc platforms.
253 * minor incompatible change: the direct superclasses of
254 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
255 not (STANDARD-OBJECT FUNCTION). This makes the
256 class-precedence-lists of GENERIC-FUNCTION and
257 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
259 * minor incompatible change: the default stream external format on
260 non-unicode SBCL is no longer detected from the locale, but is always
262 * new feature: Add a version of evaluator that uses an interpreter instead
263 of the compiler. EVAL still uses the compiler by default, to switch it
264 to use the interpreter, set the value of the variable
265 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
266 * minor incompatible change: the single-stepper REPL has been merged
267 with the normal debugger (see the "Stepping" heading of the debugger help
268 for more details). The debugger command STEP will no longer switch
269 to the single-stepper REPL.
270 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
271 for a type now works.
272 * bug fix: loading of default sysinit file works. (thanks to Leonid
274 * bug fix: better detection of circularities in the file-compiler.
275 (reported by Marco Monteiro)
276 * bug fix: the CL pathname functions now work with files that have
277 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
278 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
279 systems with Unix98 pty semantics.
280 * bug fix: ASDF-INSTALL will now work with bsd tar.
281 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
283 * bug fix: timers expiring in dead threads no longer cause a
284 type-error (reported by Paul "Nonny Mouse").
285 * bug fix: thanks to more lightweight single-stepper instrumentation,
286 code compiled with (DEBUG 3) will compile and execute significantly faster,
287 and will have more accurate type-inferencing than before
288 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
289 whose bindings are modified
290 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
291 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
292 ** replace dummy implementations of CL:MACHINE-INSTANCE and
293 CL:SOFTWARE-VERSION with proper version
295 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
296 * feature: implemented the READER-METHOD-CLASS and
297 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
298 as specified by AMOP.
299 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
301 * optimization: faster LOGCOUNT implementation on x86 and x86-64
302 (thanks to Lutz Euler)
303 * optimization: hashing of general arrays and vectors has been
304 improved. (reported by Any Fingerhut)
305 * enhancement: SB-INTROSPECT is now able to find definitions of
306 profiled functions. (thanks to Troels Henriksen)
307 * enhancement: compiler-macro expansion applies now to FUNCALL forms
308 as well, allowing compiler-macros for SETF-functions to expand.
309 * enhancement: step-instrumentation no longer wraps known
310 single-value functions with multiple-value context, allowing
311 better type inference.
312 * fixed bug #337: use of MAKE-METHOD in method combination now works
313 even in the presence of user-defined method classes. (reported by
314 Bruno Haible and Pascal Costanza)
315 * fixed bug #339(c): if there are applicable methods not part of any
316 long-form method-combination group, call INVALID-METHOD-ERROR.
317 (reported by Bruno Haible)
318 * fixed bug #361: the :FUNCTION initarg in the protocol for
319 initialization of methods can now be used to override
320 internally-produced optimized functions. (reported by Bruno
322 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
323 system-provided lambda expression no longer cause warnings about
324 unbound #:|pv-table| symbols.
325 * bug fix: improved the handling of type declarations and the
326 detection of violations for keyword arguments with non-constant
328 * bug fix: potentially erronous calls to PATHNAME and
329 MERGE-PATHNAMES were being flushed in some cases.
330 (reported by Richard Kreuter)
331 * bug fix: compiled calls to TYPEP were mishandling obsolete
332 instances. (reported by James Bielman and Attila Lendvai)
333 * bug fix: format strings with ~> without matching ~< no longer
334 trigger an AVER, but signal an understandable error instead.
335 (reported by Antonio Martinez)
336 * bug fix: specifying an output-file without a pathname-type for
337 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
338 of input-file instead of "fasl". (reported by Robert Dodier)
339 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
340 (reported by James Y Knight).
341 * bug fix: compiler-macros-function did not consider the environment
342 argument for shadowing by local functions.
343 * bug fix: compiler-macros expansion was inhibited by local INLINE
345 * bug fix: inline expansions of known functions were subject to
346 step-instrumentation in high debug policies, leading to problems
348 * bug fix: compiler failed to differentiate between different CONS
350 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
351 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
352 * bug fix: #368: incorrect use of expressed vs. upgraded array
354 * bug fix: #306a: more precise unions of array types.
355 * thread-safety improvements:
356 ** CONDITION-WAIT could return early on Linux, if the thread was
357 interrupted and subsequently continued with SIGCONT.
358 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
360 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
361 * added support for the ucs-2 external format. (contributed by Ivan
363 * minor incompatible change: pretty printing of objects of type
364 (cons symbol) is, in the default pprint-dispatch-table, now
365 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
367 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
368 called later in a class's lifetime, possibly as late as when the
369 first instance of the class is created. Previously,
370 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
371 class became finalizeable.
372 * fixed bug: FILE-POSITION sometimes returned inconsistent results
373 for multibyte external-format streams. (thanks to "vbzoli")
374 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
375 with :ALLOCATION :CLASS inherited from superclasses of the
377 * fixed bug: anonymous classes can now be created using the :NAME
378 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
379 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
380 * fixed bug: core-files saved with :EXECUTABLE T can again be
381 executed when SBCL_HOME isn't set. (reported by James Knight)
382 * fixed bug: toplevel LOCALLY forms with declarations could
383 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
384 * fixed bug: printing from several different threads using different
385 values of *print-case* could cause invalid output, due to
386 some internal special variables of the printer not being bound
387 thread-locally (reported by Max Mikhanosha)
388 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
389 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
390 (reported by Pascal Costanza)
391 * fixed bug: FINALIZE-INHERITANCE is called from
392 REINITIALIZE-INSTANCE on classes when the class has previously
393 been finalized, as required by AMOP.
394 * minor code generation optimizations:
395 ** better register allocation in CLOS dispatching functions
396 ** overflow detection when coercing signed bytes to fixnums on x86-64
397 ** is now implemented with one IMUL instruction instead of three shifts
398 ** more efficient bit-vector access on x86 and x86-64
399 ** more efficient access to raw structure slots on x86 and x86-64
400 * fixed some bugs revealed by Paul Dietz' test suite:
401 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
404 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
405 * feature: thread support on Solaris/x86, and experimental thread support
407 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
409 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
410 default method for SLOT-UNBOUND.
411 * minor incompatible change: prevent the user from specializing the
412 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
413 somewhat counter to the intent of the protocol, I (CSR) think, and
414 additionally it just doesn't work in SBCL as currently
415 implemented, thanks to optimizations (that are always valid for
416 the other three SLOT-VALUEish functions, but not for the setter).
417 * minor incompatible change: the :SB-LDB feature is now enabled by
418 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
419 the low-level debugger.
420 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
421 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
422 down to an SB-SYS:FD-STREAM, allowing direct communication in
423 more cases. Temporary files and pipes are still used as a fallback
425 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
426 on the non-reentrant C functions localtime(3) and gmtime(3).
427 * bug fix: global optimization policy was not visible in LOCALLY and
429 * bug fix: class objects can be used as specializers in methods.
430 (reported by Pascal Costanza)
431 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
432 with a type error. (reported by blitz_ on #lisp)
433 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
434 failed with a type error. (reported by Pascal Bourguignon)
435 * bug fix: merging pathnames against defaults with :DIRECTORY
436 starting with '(:RELATIVE :BACK) should preserve the :BACK.
437 (reported by James Y Knight)
438 * bug fix: saving large (>2GB) cores on x86-64 now works
439 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
440 constant index and a (simple-array (signed-byte 32)) array
441 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
442 error (patch by Robert J. Macomber)
443 * feature: TIME now displays the amount of run-time spent in GC
444 * bug fix: The debugger now does a better job of respecting
445 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
446 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
447 workaround for bug 403.)
448 * bug fix: floating point exception handling now works on FreeBSD
449 (thanks to NIIMI Satoshi)
450 * fixed some bugs revealed by Paul Dietz' test suite:
451 ** MISC.641: LET-conversion were not supposed to work in late
453 * improvements to the Win32/x86 port:
454 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
455 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
457 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
458 * new feature: source path information is generated for macro-expansion
459 errors for use in IDE's like Slime (thanks to Helmut Eller)
460 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
461 * bug fix: compiler does not loop forever on an invalid type in
463 * improvement: compilation of most CLOS applications is significantly
465 * optimization: added a limited bytecode compiler for simple toplevel
466 forms, speeding up compilation and FASL loading
467 * bug fix: the statistical profiler now properly distinguishes anonymous
470 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
471 * minor incompatible change: in sbcl-0.9.11 (but not earlier
472 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
473 into subclasses of the finalized class. Now user calls to
474 FINALIZE-INHERITANCE finalize just the one class, and calls by the
475 system return before any subclasses are finalized.
476 * minor incompatible change: The reader no longer ignores errors
477 regarding non-existent packages in #+ and #- feature tests.
478 * new feature: command line options --no-sysinit, --no-userinit to
479 inhibit loading the corresponding init files
480 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
481 for sending data through UDP sockets (thanks to François-René Rideau)
482 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
483 error is signalled instead (thanks to François-René Rideau)
484 * minor incompatible change: Error signalling behaviour of lexical
485 operations violating package locks has changed slightly. Refer to
486 documentation on package locks for details.
487 * enhancement: EVAL can process IF-expressions without resorting to the
489 * bug fix: Some operations involving SIN, COS, or TAN weren't being
490 constant-folded properly due to missing out-of-line functions.
491 (reported by Mika Pihlajamäki)
492 * bug fix: LISTEN sometimes returned T even in cases where no data was
493 immediately available from the stream
494 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
495 were reversed. (reported by Levente Mészáros)
496 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
497 when given a vector displaced to an adjustable array. (reported
499 * bug fix: derivation of float boundaries from numbers outside the
500 appropriate float range (reported by John Wiseman)
501 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
503 * bug fix: precomputing cache entries for generic functions with
504 some subclasses of specializers as yet invalid does not attempt to
505 fill a cache line with a negative offset. (reported by Levente
507 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
508 allow condition class objects as arguments to DOCUMENTATION and
509 (SETF DOCUMENTATION); only find and set documentation for
510 structure names for the STRUCTURE doc-type. (suggested by Gary
512 * improvements to the Win32/x86 port:
513 ** (user-homedir-pathname) and default initialization file
514 locations now know about the user's "Documents and Settings"
515 directory (thanks to Yaroslav Kavenchuk)
516 ** run-program is implemented (thanks to Mike Thomas)
517 ** sockets support (thanks to Timothy Ritchey)
518 ** better backtrace support (thanks to Alastair Bridgewater)
519 ** sb-grovel supported
520 ** asdf-install and sb-posix work somewhat
521 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
522 * improvements to the Solaris/x86 port:
523 ** works on Solaris 11/Solaris Express
524 ** floating-point exception handling support
525 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
526 * fixed some bugs revealed by Paul Dietz' test suite:
527 ** REMOVE-METHOD returns its generic function argument even when
528 the method is not one of the generic functions' methods.
529 ** objects declared as MEMBER types can be admissible arguments to
531 ** printing characters should simply be printed by the FORMAT ~:C
533 ** compiler failure when compiling functions with hairy constant
534 defaults for optional parameters.
535 ** compiler produces wrong code when MAYBE-INLINE-expanding a
536 function, which is already optimized.
538 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
539 * new platform: experimental support for SBCL x86/Darwin, including
540 MacOS X 10.4.5 on Intel.
541 * new feature: Unicode character names are now known to the system
542 (through CHAR-NAME and NAME-CHAR).
543 * new feature: the filesystem location of SBCL's core file is
544 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
545 * minor incompatible change: the contrib modules SB-POSIX and
546 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
547 this change is to make it easier to distribute
548 location-independent binaries.
549 * bug fix: as implied by AMOP, standardized classes no longer have
550 slots named by external symbols of public packages. (reported by
552 * optimization: calling functions via constant symbols -- as in
553 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
554 via the function object as in (FUNCALL #'FOO).
555 * optimization: on x86 and x86-64, the calling convention for the
556 case of dealing with an unknown number of values has been altered
557 to be friendlier to the prediction heuristics implemented,
558 particularly with reference to CALL and RET pairing. (thanks to
559 Alastair Bridgewater)
560 * optimization: on x86, the code for access to thread-local symbol
561 values and binding/unbinding of thread-local symbols is smaller.
562 (thanks to Alastair Bridgewater)
563 * enhancement: CONSTANTP is now able to determine constantness of
564 more complex forms, including calls to constant-foldable standardized
565 functions and some special forms beyond QUOTE.
566 * fixed bug: occasional GC crashes on Solaris/x86
567 * optimization: x86-64 supports stack allocation of results of simple
568 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
569 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
570 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
571 (thanks to James Knight)
572 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
573 by the standard (thanks to Stephen Compall)
575 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
576 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
577 be used for bundling the runtime and the core file into one
578 executable binary. This feature is not currently supported on all SBCL
579 platforms. (thanks to James Bielman and NIIMI Satoshi)
580 * new feature: a generational or ephemeral garbage collector is now
581 the default on the PowerPC platform (both Linux and Darwin). The
582 old Cheney (stop and copy) collector is a build-time option.
583 (thanks to Cyrus Harmon, after Raymond Toy)
584 * minor incompatible change: the method by which SBCL finds its
585 contributed modules has changed; it no longer relies on symbolic
586 links from an $SBCL_HOME/systems directory, but searches directly
587 in the subdirectories of $SBCL_HOME.
588 * enhancement: the dynamic heap size on the Linux/PPC platform is
589 markedly larger, even using the older Cheney garbage collector.
590 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
591 certain complicated string types. (reported by Gary King)
592 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
593 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
594 * fixed bug: DOCUMENTATION on structures no longer signals an
595 error if no structure type of the right name exists. (reported by
597 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
598 applicable non-standard primary or :AROUND method on
599 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
600 no longer cause unbound variable errors. (reported by Kevin Reid)
601 * fixed bug: metacircle resolution in cases where methods have slots
602 added before the slots from STANDARD-METHOD. (reported by Jean
604 * fixed bug: the Power PC assembly code for calling into lisp
605 sometimes computed its return address wrongly (depending on the
606 15th bit of the address, fixed at link time). This used to
607 manifest itself in a segmentation violation while building PCL.
608 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
609 many others over the years)
610 * fixed bug: ROOM no longer reports silly numbers for stack usage if
611 the stack pointer (treated as a signed integer) is negative.
612 (thanks to Peter van Eynde)
614 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
615 * new platform: experimental support for the Windows operating
616 system has been added. (thanks to Alastair Bridgewater)
617 * fixed several bugs in and robustified the PPC FFI (including
618 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
619 * bug fix: allow non-simple string symbol names (reported by Paul
621 * bug fix: interrupt handling on NetBSD (thanks to Richard M
623 * bug fix: saving a core corrupted callbacks on x86/x86-64
624 * bug fix: closed a loophole in metacircularity detection and
625 grounding in the PCL implementation of CLOS.
626 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
627 (reported by tomppa on #lisp)
628 * bug fix: clear the direction flag on Lisp -> C transitions, as
629 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
630 (reported by Andrej Grozin and Hendrik Maryns)
631 * optimization: major improvements to GC efficiency on GENCGC platforms
632 * optimization: faster implementation of EQUAL
633 * optimization: emit more efficient opcodes for some common
634 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
635 * optimization: several other minor code-generation improvements on x86-64
636 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
638 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
639 * minor incompatible change: (SETF CLASS-NAME) and (SETF
640 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
641 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
642 by ANSI as a generic function, and for consistency (SETF
643 GENERIC-FUNCTION-NAME) is treated likewise.
644 * fixed bug #233.b: make constraint propagation notice when a variable
645 value is changed after it is referenced but before it is used
646 * fixed bug #296: no more arbitrary behaviour from filenames with
647 odd characters as --load arguments. (reported by Adam Warner)
648 * fixed bug #390: :CHARACTER-SET pathname components now work as
649 expected. (reported by Tim Daly Jr)
650 * fixed bug #391: complicated :TYPE intersections in slot
651 definitions no longer cause an error in PCL internals.
652 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
653 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
654 returning the number of octets which would be written to the
655 file-stream. (thanks to Robert J. Macomber)
656 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
657 arguments to the internal stream functions. (thanks to David
659 * bug fix: stack exhaustion is now handled on the x86/SunOS
660 platform. (thanks to Daisuke Homma)
661 * bug fix: more accurate ROOM results on GENCGC platforms
662 * optimization: improved type inference for arithmetic-for
663 index variables in LOOP
664 * optimization: faster floating-point SQRT on x86-64
665 * fixed some bugs revealed by Paul Dietz' test suite:
666 ** DOCUMENTATION returns NIL instead of "" for method combinations
667 that don't have a docstring
669 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
670 * minor incompatible change: (SETF CLASS-NAME) and (SETF
671 GENERIC-FUNCTION-NAME) are no longer generic functions, and
672 therefore are not customizeable by user code (as seems to be at
673 least permitted and maybe required by AMOP). As a consolation,
674 however, the SBCL implementation of these functions now calls
675 REINITIALIZE-INSTANCE as specified by AMOP.
676 * bug fix: slot-definition documentation is propagated as per ANSI
677 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
678 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
679 argument, as do eslotd initialization methods. (from Pascal
680 Costanza's "Closer" project)
681 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
682 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
684 * bug fix: it is now possible to have more than one subclass of
685 STANDARD-GENERIC-FUNCTION without causing stack overflow.
686 (reported by Bruno Haible, Pascal Costanza and others)
687 * bug fix: the dependent update protocol now works for generic
688 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
690 * bug fix: condition-class instances corresponding to
691 DEFINE-CONDITION forms are now created eagerly. (reported by
692 Kalle Olavi Niemitalo on comp.lang.lisp)
693 * bug fix: floating point printing is more accurate in some
694 circumstances. (thanks to Simon Alexander)
695 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
696 merged with *DEFAULT-PATHNAME-DEFAULTS*.
697 * bug fix: callbacks on OS X now preserve stack-alignment.
698 (thanks to Cyrus Harmon)
699 * enhancement: the x86-64 disassembler is much better at
700 disassembling SSE instructions. (thanks to Lutz Euler)
701 * enhancement: CHECK-TYPE on a variable now allows type-inference
702 (thanks to Vincent Arkesteijn)
703 * optimization: improved performance of EUC-JP external format.
704 (thanks to NIIMI Satoshi)
705 * optimization: performance improvements to IO on file streams of
706 :ELEMENT-TYPE CHARACTER
707 * optimization: much faster memory allocation on x86-64
708 * optimization: faster garbage collections (latency and throughput)
710 * optimization: faster bignum subtraction and division on x86 and x86-64
711 (thanks to Lutz Euler)
712 * optimization: more accurate type-inference for loops that use a
713 floating point index variable or a negative step.
715 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
716 * bug fix: add a workaround to SBCL looping infinitely at startup on
717 Linux kernels with apparently buggy implementations of personality().
718 (thanks to Svein Ove Aas)
719 * bug fix: Unicode symbols are correctly printed in LDB backtraces
720 (thanks to David Lichteblau)
721 * bug fix: local bindings shadow global symbol macros for
722 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
723 * optimization: non-open coded uses of numeric comparison operators
724 (e.g. >) no longer cons when called with more than one parameter
725 on platforms supporting dynamic-extent allocation.
726 * enhancement: saving cores with foreign code loaded is now supported
727 on MIPS/Linux in addition to the previously supported platforms.
728 * bug fix: FIND-RESTART now tests for activity, not applicability when given
729 a restart object as identifier. (reported by Helmut Eller for CMUCL)
730 * bug fix: division by zero in sb-sprof when no samples were collected
731 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
732 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
733 (thanks to Alexander Kjeldaas)
734 * new feature: ignoring the return values of destructive functions
735 like NREVERSE signals a compile-time style-warning.
736 (thanks to Kevin Reid)
737 * new feature: the alignment of alien structure fields can be
738 explicitly specified. (thanks to Cyrus Harmon)
739 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
740 and dump core on SIGQUIT
742 ** incompatible change: threads do not inherit values of specials
743 from their parents (see manual)
744 ** bug fix: threads stacks belonging to dead threads are freed by the
745 next exiting thread, no need to gc to collect thread stacks anymore
746 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
747 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
748 ** bug fix: dynamic variable and thread start related gc lossage
749 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
751 * fixed some bugs revealed by Paul Dietz' test suite:
752 ** SUBTYPEP is slightly more accurate on heinously complicated
753 CONS types where some of the members have uncertain (in the
754 NIL, NIL sense) type relationships to each other.
756 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
757 no longer purified unless :PURIFY T is explicitly specified.
758 ** Non-purified cores are significantly smaller than before
760 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
761 * new feature: timers based on Zach Beane's excellent timer package
762 * added support for the following external formats: koi8-u,
763 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
764 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
765 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
766 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
767 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
768 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
769 * incompatible change: a threaded SBCL will no longer revert to
770 non-threaded mode on non-NPTL systems, but refuse to start entirely.
771 * bug fix: interrupts are disabled until startup is complete; no
772 more sigsegvs when receiving a signal to soon
773 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
775 * bug fix: add a workaround for the memory randomization features in
776 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
777 workaround will only be in effect on systems with the proc filesystem
779 * bug fix: printing objects of type HASH-TABLE signals a
780 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
782 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
783 have been running for over 50 days. (reported by Gilbert Baumann)
784 * bug fix: the logic for getting names of functions gets less
785 confused when confronded with alternate-metaclass
786 funcallable-instances. (reported by Cyrus Harmon)
787 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
788 consistent, even on internal alternate-metaclass objects.
789 * bug fix: flush closure information collected by physical
790 environment analysis prepass before the main pass. (bug reported
792 * bug fix: compiler pack phase does not modify a hashtable, which is
793 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
794 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
795 may be expanded. (reported by Alexander <malishev> on c.l.l)
796 * bug fix: no more style-warnings when compiling code using LAST
797 under high-SPEED low-SPACE settings. (reported by David Wallin
799 * fixed some incompatibilities between SBCL's MOP and the MOP
801 ** the METAOBJECT class is now implemented;
802 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
803 STANDARD-OBJECT, as required;
804 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
805 now compatible (as required by VALIDATE-SUPERCLASS); there
806 remains a consistency requirement over the presence or absence
807 of the FUNCTION class in the superclass list of finalized
808 classes; see the manual for more details;
809 ** the :AROUND method for COMPUTE-SLOTS on
810 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
811 requested slot ordering.
813 ** bug fix: parent thread now can be gc'ed even with a live
815 ** bug fix: binding a special with PROGV to no value is not
816 the same as that symbol not having been bound (thanks to
818 ** bug fix: binding specials is thread safe (thanks to
820 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
821 and SB-SPROF do not die with 'no handler for signal XX in
822 interrupt_handle_now(..)' anymore
823 ** bug fix: WITH-TIMEOUT works with multiple running threads
824 * fixed some bugs revealed by Paul Dietz' test suite:
825 ** ENSURE-GENERIC-FUNCTION should take a method class object for
826 the :method-class keyword argument.
828 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
829 * new port: the Solaris operating system on x86 processors is now
830 mostly supported, though some rough edges in the environment
831 remain. (thanks to Daisuke Homma)
832 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
833 space for its heap. (thanks to Thiemo Seufer)
834 * optimization: SBCL on MIPS platforms now supports dynamic-extent
835 closures. (thanks to Thiemo Seufer)
836 * minor incompatible change: eof selects abort in the debugger.
837 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
838 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
840 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
841 deprecated, and will go away in a future revision of SBCL.
842 * minor incompatible change: GC-ON and GC-OFF are no longer
843 implemented with a counter, it does not matter how many times gc
844 is switched on or off
845 * bug fix: discriminating functions for generic function classes
846 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
847 make invalid assumptions about method precedence order. (reported
849 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
850 platforms. (thanks to Thiemo Seufer)
851 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
852 * bug fix: correct pathname printing with printer escaping is on.
853 (thanks to Kevin Reid)
854 * bug fix: complex VOP definitions in "user-space" no longer trigger
855 package locks. (reported by Zach Beane)
856 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
857 for funcallable-instance functions, and is no different from
859 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
860 characters in the current readtable. (reported by Nicholas Neuss)
861 * bug fix: SUBTYPEP on various CONS types returns more a more
862 accurate acknowledgment of its certainty.
863 * optimizations: REMOVE-DUPLICATES now runs in linear time on
864 lists in some cases. This partially fixes bug 384.
865 * flush all standard streams before prompting in the REPL and the
867 * bug fix: signal handling and triggering gc do not conflict
868 directly anymore, in particular a high frequency sb-sprof does
869 not prevent gc from running
870 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
871 approximation for timezone and DST information between the
872 universal time epoch and the smallest negative 32-bit time_t.
873 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
874 year 1899 so long as the encoded time is non-negative.
875 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
876 the correct answer for non-constant rotate amounts when performing
877 an inline 32-bit rotation.
879 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
880 there is only one thread in the session
881 ** bug fix: memory leak for streams created in one thread and
882 written to in another
883 ** bug fix: lockup when compiled with gcc4
884 ** bug fix: race that allows the gc to be triggered when gc is
886 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
887 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
889 ** bug fix: finished threads are gc'ed properly
890 * fixed some bugs revealed by Paul Dietz' test suite:
891 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
892 has arguments with a different set of applicable methods from
893 the orignal arguments.
894 ** The type error thrown by MAP now has a correct expected-type
896 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
897 name a compiled function.
898 ** Three MISC tests where a large bignum was improperly coerced to
899 a float that couldn't represent that bignum during type
900 derivation were fixed.
901 ** SUBTYPEP can now handle types involving the negation of a
902 list-form FUNCTION type.
903 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
904 as (COMPLEX (AND RATIO (NOT FIXNUM))).
905 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
907 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
908 * New feature: Experimental support for bivalent streams: streams
909 opened with :element-type :default now allow character and binary
910 (unsigned-byte 8) I/O
911 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
912 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
914 * bug fix: no more highly sporadic "couldn't check whether ~S is
915 readable" when reading a stream and an interrupt hits in the middle
916 of a select system call
917 * compiler better recognizes complex arrays (reported by Eduardo
919 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
920 for variables on SBCL built with linkage-tables. (reported by Luis
922 * various error reporting improvements.
923 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
924 (Thanks to Thiemo Seufer)
925 * bug fix: nested reader invokations work correctly
926 * bug fix: it is possible to have simultaneous references to foreign
927 code and foreign data with the same name.
929 ** added x86-64 support
930 ** incompatible change: the threading api now works with thread
931 objects instead of thread ids
932 ** bug fix: threads are protected from signals and interruption when
933 starting up or going down
934 ** bug fix: a race where an exiting thread could lose its stack to gc
935 ** bug fix: don't halt on infinite error in threads if possible
936 ** fixed numerous gc deadlocks introduced in the pthread merge
937 ** bug fix: fixed thread safety issues in read and print
938 ** bug fix: debugger doesn't hang on session lock if interrupted at
939 an inappropriate moment
940 ** bug fix: run-program is now thread safe(r)
941 ** bug fix: inner with-recursive-lock no longer releases the mutex
942 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
943 * fixed some bugs revealed by Paul Dietz' test suite:
944 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
945 EXPECTED-TYPE slots filled.
946 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
948 ** MAKE-CONDITION accepts classes as type-designators.
949 ** COMPILE may never return NIL.
950 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
951 range before calling Unix time functions
953 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
954 * numerous signal handling fixes to increase stability
955 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
956 * minor incompatible change: we now correctly canonize default
957 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
958 as per the MOP, rather than the historical (INITARG INITFUNCTION
959 INITFORM). (reported by Bruno Haible)
960 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
961 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
962 for more information.
963 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
964 pathname is a directory pathname.
965 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
966 * dynamic space size on PPC has been increased to 768Mb. (thanks to
968 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
969 :METACLASS argument in addition to a class name. (reported by
970 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
971 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
972 (reported by Bruno Haible; patch from Gerd Moellmann)
973 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
975 * bug fix: more cleanups to the floating point exception handling on
976 x86-64 (thanks to James Knight)
977 * bug fix: the compiler does not try to constant fold calls of
978 COERCE and BIT-* functions when they return freshly constructed
979 objects. (reported by Nikodemus Siivola and Paul Dietz)
980 * optimization: improved the MIPS versions of generic (in the
981 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
982 * optimization: direct conversion of (unsigned-byte 32) to floats on
983 the PowerPC platform.
984 * optimization: structure instances with raw slots now use less
985 memory, and probably show better memory locality. (thanks to
987 * optimization: DYNAMIC-EXTENT declarations for lists and closures
988 are treated as requests for stack allocation on the x86-64,
989 Alpha-32, PPC and SPARC platforms.
990 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
991 the wrong file descriptor; implementation of SOCKET-OPEN-P.
992 (thanks to Tony Martinez)
994 ** gcing a dead thread can no longer lead to lockups
995 ** threads block signals until they are set up properly
996 ** errno is no longer shared by threads
997 ** interrupt-thread restores the eflags register on x86
998 ** fixed some lockups due to gc/thread interaction
999 * fixed some bugs revealed by Paul Dietz' test suite:
1000 ** invalid dotted lists no longer raise a read error when
1001 *READ-SUPPRESS* is T
1002 ** signal an error if a symbol that names a declaration is used
1003 as the name of a type, or vice versa
1004 ** allow using the (declare (typespec var*)) abbreviation for
1005 (declare (type typespec var*)) with all type specifiers
1006 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1007 ** always return NIL from PPRINT-POP when OBJECT is NIL
1008 ** don't signal errors when pretty-printing malformed LABELS,
1009 FLET or MACROLET forms
1010 ** declarations in a DOLIST body are also in scope for the
1012 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1014 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1017 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1018 * fixed cross-compiler leakages that prevented building a 32-bit
1019 target with a 64-bit host compiler.
1020 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1021 opened with :IF-EXISTS :OVERWRITE.
1022 * fixed bug 281: error for an invalid qualifier in a short-form method
1023 combination method is not signalled until the faulty method is called.
1024 * bug fix: iteration variable type inferrer failed to deal with open
1025 intervals. (reported by Alan Shields)
1026 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1027 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1028 or more runtime options were provided to the sbcl binary.
1029 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1030 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1031 * optimization: there's now a fast-path for fixnum arguments in the
1032 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1033 * optimization: the code generated on x86-64 is more compact thanks
1034 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1035 * bug fix: floating-point exception handling now partly works on
1036 x86-64. (Thanks to James Knight)
1037 * improvement to the MIPS backend from Thiemo Seufer: C-style
1038 64-bit long long arguments and return values to alien functions
1040 * fixed some bugs revealed by Paul Dietz' test suite:
1041 ** the type-error signalled from WARN has a filled-in DATUM slot.
1042 ** the type-error required when a stream is not associated with
1043 a file has the stream as its datum.
1044 ** type-errors on single-floats on x86-64 no longer have
1045 :INVALID-OBJECT as the datum
1046 ** the type-errors signalled for invalid function names now have
1047 a correct expected type
1048 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1049 for structures defined with a :TYPE. Documentation strings for
1050 typed structures are no longer immediately discarded
1051 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1052 broadcast streams and synonym streams. FILE-LENGTH now also works
1053 on broadcast streams.
1055 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1056 * incompatible change: the --noprogrammer option, deprecated since
1057 version 0.7.5, has been removed. Please use the equivalent
1058 --disable-debugger option instead.
1059 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1061 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1062 inoperational for a while now) has been completely removed.
1063 * null lexical environments are now printed as #<NULL-LEXENV>,
1064 significantly reducing the amount of clutter in typical backtraces.
1065 * documentation on weak pointers, finalization, and after GC hooks
1066 has been added to the manual.
1067 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1068 as other specialized array types, is much faster. SUBSEQ and
1069 COPY-SEQ on such arrays have also been sped up.
1070 * optimization: EQL is now more efficient when at least other argument
1071 is known to be of type (OR FIXNUM (NOT NUMBER)).
1072 * fixed bug: the runtime is now less vulnerable to changes in the
1073 size of the SBCL object on OS X, and virtual memory is reserved for
1074 all spaces that need to be at a fixed address.
1075 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1076 * fixed bug: finalizers and after GC hooks that cause consing are now
1078 * fixed bug: compiler error messages and summaries are now printed to
1079 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1080 * fixed inference of the upper bound of an iteration variable.
1081 (reported by Rajat Datta).
1082 * fixed bug 211e: calling local functions with duplicated constant
1083 keyword argument no longer causes a bogus style warning about an
1085 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1086 ftype declaration to be disregarded. (reported by Dave Roberts)
1087 * fixed bug 373: caused by erronous compilation of references to alien
1088 variables in the runtime on ppc/darwin.
1089 * fixed bug 376: CONJUGATE type deriver.
1090 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1091 (reported by Baughn on #lisp)
1092 * a cross-compiler bug on non-x86 platforms has been identified and
1093 fixed. (thanks to Bruno Haible)
1094 * improvements to the MIPS runtime code for increased stability.
1095 (thanks to Thiemo Seufer)
1096 * increased the maximimum compact environment size to allow
1097 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1098 * partial workaround for bug 135: don't name the function we're
1099 calling for hairy cases of EVAL, so as not to accumulate one environment
1100 entry per call to EVAL. (reported by Kevin Reid)
1101 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1102 * optimization: type testing for non-vector arrays should be faster.
1103 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1105 * bug fix: setting 31st element of a bit vector to zero did not work
1107 * bug fix: redefining a class definition which failed due to a
1108 previous accessor / function clash now works (but see BUGS entry
1109 #380 for more problems in this area). (thanks to Zach Beane)
1110 * the long form of DEFINE-METHOD-COMBINATION disables method group
1111 checking when given a single method group with pattern *. (thanks
1113 * on x86 compiler supports stack allocation of results of simple
1114 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1115 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1116 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1117 * fixed some bugs related to Unicode integration:
1118 ** the restarts for recovering from input and output encoding
1119 errors only appear when there is in fact such an error to
1121 * fixed some bugs revealed by Paul Dietz' test suite:
1122 ** MISC.549 and similar: late transformation of unsafe type
1123 assertions into derived types caused unexpected code
1125 ** SCALE-FLOAT type deriver is less wrong.
1126 ** type derivers for EXP, LOG and similar functions compute result
1127 types for complex arguments better.
1128 ** (MISC.563) CONJUGATE type deriver works for very restricted
1130 ** out-of-line type testers for character strings are available.
1131 ** EQUAL compiler transform understands specialness of objects
1133 ** accessing double-floats stored on the stack now works on x86-64.
1134 ** debugger internals could sometimes create invalid lispobjs,
1135 resulting in GC crashes.
1136 ** MISC.548: type check weakening can convert required type into
1138 ** initialization forms for bindings are not in scope of free special
1141 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1142 * incompatible change: thread support for non-NPTL systems has
1143 been removed - locking is buggy and unreliable. A threaded
1144 SBCL build will now warn at startup and refuse to create
1145 new threads, unless futex support is detected in the kernel
1146 * incompatible change: the top level REPL now has only an ABORT
1147 restart associated with it, not TOPLEVEL and ABORT as it used to.
1148 TOP and TOPLEVEL are now available as debugger commands for
1149 returning to the top level.
1150 * incompatible change: forms evaluated in the REPL now use the
1151 global optimization policy.
1152 * incompatible change: user- and system-initialization files are
1153 no longer processed with LOAD, but by READ and EVAL; hence the
1154 global optimization policy, startup package, readtable, etc,
1156 * The .fasl file format number has been incremented because of
1157 various incompatible changes.
1158 * internal entry point details and argument counts no longer appear
1159 in backtraces unless explicitly requested by setting
1160 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1161 * built-in and standard functions no longer have names like "top
1162 level local call to FOO".
1163 * fixed bug 32: functions defined in non-null lexical environments
1164 now have more legible printed representation
1165 * fixed bug 33: functions defined in non-null lexical environemnts
1166 are now more amenable to inspection by INSPECT.
1167 * workaround for bug 354: XEPs no longer appear in backtraces unless
1168 explicitly requested.
1169 * fixed bug: receiving the signal which results from attempting to
1170 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1171 SIGBUS on other *BSDs) on architectures where the C stack is also
1172 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1173 to debugger confusion or wild instability. (reported by Cheuksan
1175 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1176 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1177 (reported by Lutz Euler)
1178 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1179 compiled code. (reported by Paul Dietz)
1180 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1181 specialized parameter to a method to something that is not TYPEP
1182 the specializer is now possible.
1183 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1184 face of package deletion.
1185 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1186 pathname. (reported by David Tolpin on c.l.l)
1187 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1188 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1189 STRING-STREAM and FILE-STREAM, respectively.
1190 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1191 than previously; in addition, its test suite is now run on build.
1192 (thanks to Luke Gorrie)
1193 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1194 * optimization: added a immediate representation for single-floats
1196 * fixed some bugs revealed by Paul Dietz' test suite:
1197 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1198 ** attempting to create a package with a colliding nickname causes
1199 correctable errors to be signalled.
1200 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1201 ** several x86-64 backend bugs related to sign-extension of immediate
1204 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1205 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1206 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1208 * fixed loading of multiply forward-referenced layouts.
1209 (thanks to Cheuksan Wang)
1210 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1211 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1212 (see COMPILE-FILE documentation for details of :PRINT option)
1213 * fixed bugs 19 and 317: fixed-format floating point printing is
1214 more accurate. This also fixes a bug reported by Adam Warner
1215 related to the ~@F format directive.
1216 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1218 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1219 dispatch table if the from-char is a dispatch macro character.
1220 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1221 by the wordsize no longer ignore the last word. (reported by Lutz
1223 * fixed bug in type checking of dynamic-extent variables. (reported
1225 * optimization: sequence traversal functions use their freedom to
1226 coerce function designators to functions.
1227 * optimization: code with many calls to CLOS methods specialized on
1228 CLOS classes has had redundant type checks removed.
1229 * optimization: added declarations to speed up operations that access
1230 the internal character database (for example STRING-UPCASE)
1231 * optimization: comparison operations between floats and sufficiently small
1232 fixnums no longer create extra rationals
1233 * fixed some bugs related to Unicode integration:
1234 ** portions of multibyte characters at the end of buffers for
1235 character-based file input are correctly transferred to the
1236 start of the buffer at the next read.
1237 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1238 passing it through to OPEN.
1239 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1240 argument given to internal calls to OPEN.
1241 ** the built-in comment readers (introduced by character sequences
1242 ";" and "#|") are more forgiving to encoding errors; they will
1243 STYLE-WARN and then attempt to resync the stream at a character
1244 boundary. (thanks to Teemu Kalvas)
1245 * fixed some bugs in the x86-64 port:
1246 ** Negative short int return values from c-calls are sign-extended
1248 ** The stack is aligned to 16-bytes for c-calls, as required by
1249 the ABI. (thanks to Cheuksan Wang)
1250 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1251 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1252 for enums. (thanks to Vincent Arkesteijn)
1253 ** Multiple small optimizations and bugfixes for floating point
1255 * fixed some bugs revealed by Paul Dietz' test suite:
1256 ** Space, Tab, Linefeed, Return and Page have the invalid
1257 secondary constituent character trait.
1258 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1260 ** WITH-INPUT-FROM-STRING should only update the index place on
1262 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1263 works more reliably.
1264 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1265 with one-dimensional arrays with fill pointers.
1266 ** TYPE-OF failed on a complex with an integer realpart and a
1268 ** compiler failure during type inference for the code of form
1269 (IF (EQL X (THE ...)) ...) (MISC.535).
1271 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1272 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1273 platform. The port supports 61-bit fixnums, large memory spaces
1274 and reloading shared object files.
1275 * enhancement: saving cores with foreign code loaded is now
1276 supported on ppc/Darwin in addition to the previously supported
1278 * enhancement: the statistical profiler now walks deeper into the
1279 call stack for more meaningful call-graphs and accrued time
1280 reports (x86/x86-64 only). It also now reports time spent in
1282 * enhancement: it is now possible to trace most individual methods
1283 of a generic function in addition to tracing the generic function
1285 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1287 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1288 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1289 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1290 produces an error. (thanks to Vincent Arkesteijn)
1291 * bug fix: NAMESTRING on pathnames with :WILD components in their
1292 directories works correctly. (thanks to Artem V. Andreev)
1293 * fixed bug 125: compiler preserves identity of closures. (reported
1295 * bug fixed: functions with &REST arguments sometimes failed with
1296 "Undefined function" when compiled with (DEBUG 3). (reported by
1298 * bug fix: overflow during compiling of setting element of a bit
1299 vector with constant index and value. (reported by Timmy Douglas)
1300 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1301 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1302 * on x86 compiler partially supports stack allocation of dynamic-extent
1304 * GO and RETURN-FROM do not check the extent of their exit points
1305 when compiled with SAFETY 0.
1306 * fixed some bugs related to Unicode integration:
1307 ** encoding and decoding errors are now much more robustly
1308 handled; it should now be possible to recover even from invalid
1309 input or output to the terminal. (thanks to Teemu Kalvas)
1310 ** provided a first cut at implementing STRING-TO-OCTETS and
1311 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1312 ** altered the SB-MD5 contributed module slightly, changing the
1313 interface just enough for it to be supportable for builds where
1314 lisp characters are not eight bits.
1315 * fixed some bugs revealed by Paul Dietz' test suite:
1316 ** the FORMATTER-generated functions for ~V[ conditionals require
1317 the correct number of arguments.
1318 ** READ-FROM-STRING returns the mandated second value when applied
1319 to displaced strings.
1320 ** the #\Rubout and #\Backspace characters are treated as invalid
1321 constituent characters by the tokenizer.
1323 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1324 * new feature: reloading changed shared object files with
1325 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1326 * new feature: references to foreign variables and functions
1327 can now be compiled and loaded before the corresponding shared
1328 object file is loaded, as long as the foreign definitions are
1329 available at runtime.
1330 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1331 supported. (thanks to Dan Debertin)
1332 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1333 just keywords. (thanks to Vincent Arkesteijn)
1334 * fixed bug #331: structure-class instances corresponding to
1335 DEFSTRUCT forms are now created eagerly.
1336 * fixed bug #345: backtraces from calls to undefined functions work
1337 on x86 as well. Related bug #61 is now also partially fixed on x86
1338 (backtraces from throws to unknown catch tags.)
1339 * bug fix: lambda-list parsing is now stricter vrt. order and number
1340 of lambda-list keywords.
1341 * bug fix: as specified by AMOP, an error is signalled if a
1342 class-option appears multiple times in a DEFCLASS form. (reported
1344 * bug fix: RANDOM can be compiled when the compiler derives the type
1345 of its numeric argument as a disjoint set of small integers.
1346 (reported by Paul Dietz)
1347 * bug fix: starting a core saved with shared objects loaded when
1348 those objects are not available no longer causes threaded SBCL to
1349 hang. (reported by Sean Ross)
1350 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1352 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1353 argument, which must always be NIL. (reported by Kalle Niemitalo)
1354 * bug fix: printing 1.0d+23 no longer results in an error.
1355 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1356 * bug fix: structure slot setters preserve evaluation order. (thanks
1358 * fixed some bugs related to Unicode integration:
1359 ** RUN-PROGRAM can allow its child to take input from a Lisp
1360 stream. (reported by Stefan Scholl)
1361 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1362 locales. (reported by Ken Causey)
1363 * fixed some bugs revealed by Paul Dietz' test suite:
1364 ** INCF, DECF and REMF evaluate their place form as specified in
1366 ** FORMATTER expands ~{ ~} iteration directives with V or #
1367 parameters correctly.
1368 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1369 consequent uses no arguments correctly.
1370 ** the system has a partial understanding of the (COMPLEX RATIO)
1372 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1375 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1376 * new feature: a build-time option (controlled by the :SB-UNICODE
1377 keyword feature, enabled by default) for building the system with
1378 support for the entire 21-bit character space defined by the
1380 * new feature: the system now has rudimentary external-format
1381 support; the primary user-visible change at this time is that
1382 characters with the high bit set (such as lower-case-e-acute) will
1383 print correctly to a terminal in a UTF-8 environment.
1384 * minor incompatible change: BASE-CHAR no longer names a class;
1385 however, CHARACTER continues to do so, as required by ANSI.
1386 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1387 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1388 has been moved to the SB-EXT package (temporarily re-exported from
1390 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1391 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1393 * fixed bug #351: better error-handlind and reporting for malformed
1395 * fixed bug #350: bignum-printing is now more memory-efficient,
1396 allowing printing of very large bignums, eg. (expt 2 10000000).
1397 (reported by Bruno Haible)
1398 * fixed bug #302: better primitive-type selection for intersection
1400 * fixed bug #308: non-graphic characters now all have names, as
1401 required. (reported by Bruno Haible)
1402 * bug fix: redefining a class with different superclasses now correctly
1403 removes it from the direct-subclasses of its previous superclasses.
1404 (reported by David Morse)
1405 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1406 new value now works. (reported by Bruno Haible)
1407 * bug fix: correct canonicalization of multiple non-standard slot
1408 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1409 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1410 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1411 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1412 now exists, an signals an error.
1413 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1414 during macroexpansion for non-iteration variables as well. (reported
1415 by Bruno Haible for CMUCL)
1416 * bug fix: Cyclic structures and unprintable objects in compiler
1417 messages no longer cause errors. (reported by Bruno Haible)
1418 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1419 and READ-FROM-STRING all now return a primary value of NIL if
1420 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1421 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1422 (reported by Bruno Haible for CMUCL)
1423 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1424 do not designate a condition. (reported by Bruno Haible for
1426 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1427 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1428 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1429 permitting resolution in favour of any of the conflicting symbols.
1430 (reported by Bruno Haible for CMUCL)
1431 * bug fix: EQUAL compiler optimizations is less aggressive on
1432 strings which can potentially compare true despite having distinct
1433 specialized array element types.
1434 * bug fix: unit enumerations can be defined without dividing by
1435 zero. (thanks to Vincent Arkesteijn)
1436 * FORMAT compile-time argument count checking has been enhanced.
1437 (report from Bruno Haible for CMUCL)
1438 * a partial workaround for the bug 262: the compiler does not try to
1439 inline-expand a local function doing RETURN-FROM from a deleted
1440 BLOCK. (thanks to Peter Denno for the bug report and to David
1441 Wragg for the simple test case)
1442 * fixed some bugs revealed by Paul Dietz' test suite:
1443 ** RENAME-PACKAGE allows all package designators as new package
1445 ** constraint propagation and lambda variable substitution are
1446 more cautious in dealing with partially deleted code.
1447 ** compiler sometimes lost reoptimization passes.
1448 ** CERROR, given a condition as condition designator, passes its
1449 remaining arguments to the continue format control without
1451 ** Case-altering FORMAT directives work correctly on non-ASCII
1453 ** The REAL type specifier handles bounds outside the floating
1454 point ranges without signalling FLOATING-POINT-OVERFLOW.
1455 ** Functions with IR1-transformations can create intercomponent
1456 references to global functions.
1457 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1459 ** FORMAT ~R treats a nil value for its first parameter correctly.
1461 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1462 * enhancement: saving cores with foreign code loaded is now
1463 supported on x86/NetBSD and sparc/Linux in addition to the previously
1464 supported platforms.
1465 * bug fix: on some platforms repeated installations caused multiple
1466 copies of HTML documentation to be installed -- should not happen
1467 any more. (reported by Stefan Scholl)
1468 * bug fix: parsing self-recursive alien record types multiple times
1469 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1470 original patch by Helmut Eller for CMUCL)
1471 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1472 (thanks to Richard Kreuter)
1473 * bug fix: defining classes whose accessors are methods on existing
1474 generic functions in other (locked) packages no longer signals
1475 bogus package lock violations. (reported by François-René Rideau)
1476 * bug fix: special variables as DEFMETHOD parameters no longer have
1477 associated bogus type declarations. (reported by David Wragg and
1479 * bug fix: read-write consistency on streams of element-type
1480 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1481 * bug fix: redefiniton of the only method of a generic function with
1482 no DEFGENERIC no longer emits a full WARNING. In addition,
1483 redefinition of generic functions with no DEFGENERIC to an
1484 incompatible lambda list now signals an error. (thanks to Zach
1486 * bug fix: DEFGENERIC now works even when there's a function of the
1487 same name in an enclosing lexical environment. (thanks to Zach
1489 * fixed compiler failure, caused by instrumenting code during
1490 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1491 * optimization: added loop analysis and improved register allocation
1492 * fixed some bugs revealed by Paul Dietz' test suite:
1493 ** POSITION on displaced vectors with non-zero displacement
1494 returns the right answer.
1495 ** (SIMPLE-STRING) is a valid type specifier for sequence
1497 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1499 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1500 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1502 ** The justification version of the FORMAT ~< directive treats
1503 non-zero minpad parameter correctly.
1505 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1506 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1507 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1508 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1509 the supported interface.
1510 * new feature: Single-stepping of code compiled with DEBUG 2 or
1511 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1512 * new feature: saving cores with foreign code loaded is now
1513 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1514 Timothy Moore's work for CMUCL)
1515 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1516 parameters to * instead of NIL if no initform is supplied.
1517 (reported by Johan Bockgaard)
1518 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1519 correctly when FUNCALL appears as the car of the form. Note:
1520 despite this FUNCALL forms are not currently subject to
1521 compiler-macro expansion. (port of Raymond Toy's fix for the
1522 same from CMUCL, reported by Johan Bockgaard)
1523 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1524 (thanks for Teemu Kalvas)
1525 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1526 case instead of terminating the iteration (thanks for Julian
1527 Squires, Sean Champ and Raymond Toy)
1528 * bug fix: incorrect expansion of defgeneric that caused a style
1529 warning. (thanks for Zach Beane)
1530 * bug fix: slot accessor effective method computation works properly
1531 for classes with multiple non-standard applicable methods on
1532 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1533 * on x86 compiler supports stack allocation of results of LIST and
1534 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1535 CMUCL implementation by Gerd Moellmann)
1536 * fixed some bugs revealed by Paul Dietz' test suite:
1537 ** FORMAT strings with both the ~<~:;~> form of the justification
1538 directive and pretty-printing directives cause an error.
1540 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1541 * incompatible change: the internal functions
1542 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1543 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1544 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1545 instead of the old functions.
1546 * new feature: on platforms where "dladdr" is available foreign
1547 function names now appear in backtraces. (based on Helmut Eller's
1549 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1550 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1552 * documentation: ASDF manual is now distributed with SBCL.
1553 * bug fix: SBCL can now load its contributed modules with REQUIRE
1554 even if the system-provided entries have been removed from
1555 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1557 * bug fix: backtraces involving undefined functions or assembly
1558 routines are more informative. (thanks to Brian Downing)
1559 * bug fix: mutually referent alien structures now work correctly.
1560 (reported by Rick Taube)
1561 * bug fix: structures defined by WITH-ALIEN can be referred to
1562 within other definitions in the same WITH-ALIEN.
1563 * bug fix: division operators (MOD, TRUNCATE and the like) with
1564 constant zero divisors and integer dividends no longer generate
1566 * bug fix: provide default methods for INPUT-STREAM-P and
1567 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1568 * bug fix: improve the detection and resolution of MOP metacycles
1569 (where effective method computation depends on the generic function
1570 having its effective method computed). (reported by Bruno Haible)
1571 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1572 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1573 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1574 * bug fix: pathnames with relative directory components can be
1575 represented relative to default pathnames.
1576 * optimization: in taking the GCD of bignums, reduce the two bignums
1577 to approximately the same size (using Euclid's algorithm) before
1578 applying the more sophisticated binary GCD. (thanks to Juho
1580 * optimization: COUNT on bitvectors now operates word-at-a-time.
1581 * optimization: ASH with a positive, but not necessarily constant,
1582 (leftwards) shift, when in a modular context, is compiled to a
1584 * fixed some bugs revealed by Paul Dietz' test suite:
1585 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1586 if the corresponding argument is NIL.
1587 ** FORMAT directives accepting a minpad parameter treat negative
1589 ** literal commas as character parameters to FORMAT directives are
1591 ** literal spaces directly after ~<Newline> directives within a
1592 format-logical-block (~:< ~@:>) do not induce :FILL-style
1593 conditional newlines.
1594 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1595 as specified: it no longer includes conditional newlines).
1596 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1598 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1599 :PER-LINE-PREFIX argument does not evaluate to a string.
1601 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1602 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1603 the manual for details; for now, package locks can be disabled by
1604 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1605 if no major problems are found then it is likely that they will be
1606 compiled in unconditionally.
1607 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1608 unsupported operators on all platforms. To load a shared library
1609 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1610 object file, link it into a shared library outside of SBCL and
1611 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1613 * bug workaround: ROOM T and the GC were not working together
1614 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1615 reporting facility aren't preserved. That reporting has been
1616 disabled, so now until and unless someone figures out how to make it
1617 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1618 * minor incompatible change: as threatened around sbcl-0.8.0, the
1619 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1620 The symbols are also exported from SB-PCL for backwards
1621 compatibility, but more so than before SB-PCL should be treated as
1622 an implementation-internal package.
1623 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1625 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1626 It is known as of this release not to work on the Alpha, however.
1627 * fixed bug #167: errors signalled due to illegal syntax in method
1628 bodies are now more legible.
1629 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1630 designators and can hence be used with TYPEP.
1631 * fixed bug #333: CHECK-TYPE now ensures that the type error
1632 signalled, if any, has the right object to be accessed by
1633 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1634 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1635 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1637 * fixed bug #334: programmatic addition of slots using specialized
1638 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1639 and :ALLOCATION :CLASS slots.
1640 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1641 float exponents. (rereported by Peter Seibel)
1642 * fixed a bug: #\Space (and other whitespace characters) are no
1643 longer considered to be macro characters in standard syntax by
1644 GET-MACRO-CHARACTER.
1645 * fixed bug: initialization of condition class metaobjects no longer
1646 causes an instance of the condition to be created. (reported by
1648 * fixed bug: it is now possible to have slots such that
1649 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1650 description is neither :INSTANCE nor :CLASS.
1651 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1652 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1653 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1654 non-standard SLOT-DEFINITION-ALLOCATION values.
1655 * fixed bugs: various race conditions handling exiting threads.
1656 CL-PPCRE's thread torture test now passes.
1657 * fixed arguably-a-bug: GC time proportional to number of threads in
1658 system even when most of them are idle
1659 * optimization: improved performance of BIT and SBIT on bit-vectors.
1660 * .fasl file incompatibility: The fasl file version number has
1661 been incremented because of changes associated with package locks.
1663 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1664 * minor incompatible change: the system no longer provides
1665 optimization hints (conditions which are TYPEP
1666 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1667 mode; these hints are still emitted when higher SPEED optimization
1669 * new contrib module: a sampling profiler (profiling by statistical
1670 sampling, rather than by instrumenting functions) is available as
1671 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1672 * the behaviour of the standard function ED is now customizeable by
1673 third parties through a hook variable: see ED's documentation
1674 string for information on the protocol.
1675 * the compiler no longer emits efficiency notes for (FUNCALL X)
1676 when the type of X is uncertain under default optimization
1678 * fixed bug 276: mutating a binding of a specialized parameter to a
1679 method to something that is not TYPEP the specializer is now
1681 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1682 exhibit the required behaviour on the x86 platform. (thanks to
1683 Peter van Eynde, Eric Marsden and Bruno Haible)
1684 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1685 even for difficult arguments. (reported by Peter Graves)
1686 * fixed bug 141a: the backquote printer now descends quoted
1688 * fixed another bug in backquote printing: no more destructive
1689 modification of the form's list structure. (reported by Brian
1691 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1692 move between its address being taken and the call to
1693 interrupt_thread, fixing a crashing race condition.
1694 * the SB-POSIX contrib implementation has been adjusted so that it
1695 no longer exhibits ridiculously poor performance when constructing
1696 instances corresponding to C structs.
1698 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1699 * minor incompatible change: the sb-grovel contrib now treats C
1700 structures as alien (in the sense of SB-ALIEN) objects rather than
1701 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1702 has implications for memory management of client code
1703 (sb-grovel-returned objects must now be manually managed) and for
1704 type safety (alien objects now have full types).
1705 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1706 used to control emission of compiler diagnostics, rather than the
1707 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1708 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1709 quality should be considered deprecated.
1710 * install.sh now installs the user manual as well
1711 * (not quite a new documentable feature, but worth considering in
1712 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1713 beginnings of a semantically meaningful condition hierarchy is
1714 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1716 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1717 designator as the defaults argument.
1718 * fixed bug: Displaced arrays whose displaced-to array has become
1719 too small now cause ARRAY-DIMENSION to signal an error, providing
1720 for safer bounds-checking. (reported by Bruno Haible)
1721 * fixed bug: DEFCLASS slot definitions with identical :READER and
1722 :WRITER names now signal a reasonable error. (reported by Thomas
1724 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1726 * fixed bug: Invalid slot specification errors now print correctly.
1727 (thanks to Zach Beane)
1728 * fixed bug 320: Shared to local slot value transfers in class
1729 redefinitions now happen corrently from superclasses as
1730 well. (reported by Bruno Haible)
1731 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1733 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1734 improper lists correctly. (reported by Bruno Haible)
1735 * fixed bug 313: source-transform for <fun-name> was erroneously
1736 applied to a call of a value of a variable with name <fun-name>.
1737 (reported by Antonio Menezes Leitao)
1738 * fixed bug 307: The obsolete instance protocol ensures that
1739 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1740 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1741 remove dead unknown-values globs from the middle of the stack.
1742 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1743 easier to resume long computations after using *BREAK-ON-SIGNALS*
1744 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1745 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1746 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1747 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1748 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1749 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1751 * fixed a (fairly theoretical) bug in string printing: if
1752 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1753 does not have array-element-type equal to the most general string
1755 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1756 function name. (thanks to Nikodemus Siivola)
1757 * fixed bug: compile-time format string checker failed on
1758 non-closed ~{. (reported by Thomas F Burdick)
1759 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1760 don't warn on legal loop code involving more than one aggregate
1761 boolean. (thanks to Nikodemus Siivola)
1762 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1763 in keyword position to MAKE-PACKAGE was not being reported as
1765 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1766 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1768 * fixed bugs 280 and 312: the checking for multiple definitions in a
1769 file is less likely to become confused by uses of inline
1771 * fixed bug: the #S reader macro performs the keyword coercion
1772 specified for slot names. (reported by Kalle Niemitalo)
1773 * fixed bug: lambda lists may contain symbols whose names start with
1774 & but are not lambda-list-keywords; their occurrence triggers a
1776 * fixed bug 321: define-method-combination argument lambda lists do
1777 not cause invalid code to be generated when &optional or &aux
1778 variables have default values. (reported by Bruno Haible)
1779 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1780 have CLOS classes; this appears also to have fixed reported
1781 instability in stack exhaustion detection.
1782 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1783 MUFFLE-WARNING when no associated restart is present is now
1785 * optimization: rearranged the expansion of various defining macros
1786 so that each expands into only one top-level form in a
1787 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1789 * optimization: used a previously little-used slot in symbols to
1790 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1792 * fixed some bugs revealed by Paul Dietz' test suite:
1793 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1795 ** Fixed an optimization bug related to inheritance of initforms
1796 from local to shared slots.
1797 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1798 ** CAST optimizer forgot to flush argument derived type.
1799 ** print/read consistency on floats is now orders of magnitude
1800 more likely. (thanks also to Bruno Haible for a similar report
1802 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1803 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1804 for objects of type REAL. Make it so.
1805 ** SXHASH is specified (infelicitously) to respect similarity,
1806 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1807 it so. (thanks to Markus Ziegler)
1808 ** on the Alpha, the compiler succeeds in compiling functions
1809 returning a known number of arguments greater than 63.
1810 ** fixed handling of invalid NIL arguments in keyword position
1812 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1814 ** provided a readably-printable representation for RANDOM-STATE
1816 ** ensured that pathnames, where they have a namestring, always
1817 print using #P"..." syntax.
1819 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1820 * Support for the forthcoming 2.0 version of the NetBSD kernel
1821 running on x86 hardware has been added. (thanks to Perry
1822 E. Metzger most immediately, and others for their past work)
1823 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1824 current release version; SBCL's OpenBSD support had been broken
1825 since about the time of OpenBSD's switch to ELF binary format.)
1826 * [placeholder for DX summary]
1827 ** user code with &REST lists declared dynamic-extent, under high
1828 speed or space and low safety and debug optimization policy.
1829 * The manual has been converted to Texinfo format and the debugger
1830 chapter from the cmucl manual has been added.
1831 * A facility has been added to extract documentation strings from
1832 sbcl and store them as Texinfo-formatted snippets for inclusion in
1833 the manual (via Texinfo's @include directive)
1834 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1835 the test case to Dave Roberts)
1836 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1837 pointers. (reported by Sean Ross)
1838 * bug fix: PROFILE output is printed nicely even for large numerical
1839 values. (thanks to Zach Beane)
1840 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1841 greater than 32 handle EOF correctly.
1842 * bug fix: on X86 an immediate argument of the IMUL instruction is
1843 correctly printed in disassembly. (thanks to Lutz Euler)
1844 * bug fix: class slots in redefined classes preserve their old
1845 values. (thanks to Bruno Haible and Nikodemus Siivola)
1846 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1847 succeeds. (reported by Marco Baringer)
1848 * fixed some bugs revealed by Paul Dietz' test suite:
1849 ** READ-SEQUENCE now works on ECHO-STREAMs.
1850 ** RATIONALIZE works more according to its specification. (thanks
1853 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1854 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1855 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1856 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1857 should matter to you only if you rebind the printer control
1858 variables and then find you want different bindings in the
1859 debugger than in the ordinary execution of your program.)
1860 * The runtime build system has been tweaked to support building
1861 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1862 assembler and linker. (thanks to Nikodemus Siivola)
1863 * Unbound, undefined, undeclared variables now trigger full
1864 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1865 more often programmer error than deliberate exploitation of undefined
1867 * optimization: the hash algorithm for strings has changed to one
1868 that is less vulnerable to spurious collisions. (thanks to Juho
1870 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1871 needless bounds checking. (thanks to Juho Snellman)
1872 * optimization: implemented multiplication as a modular
1873 (UNSIGNED-BYTE 32) operation on the PPC backend.
1874 * fixed some bugs revealed by Paul Dietz' test suite:
1875 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1877 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1878 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1879 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1881 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1883 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1885 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1886 their output stream on EOF from read.
1887 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1888 have been read to end-of-file.
1889 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1891 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1892 description of determination of which consecutive characters
1894 ** Printing the "Space" character with escaping on now yields "#\\ ",
1895 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1896 ** Reading floating-point numbers with *READ-BASE* set to a number
1897 less than 10 works correctly.
1898 ** Reading floating-point numbers with *READ-BASE* set to a number
1899 more than 10 works correctly.
1900 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1901 the readtable currently in effect.
1903 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1904 * minor incompatible change: parsing of namestrings on a physical
1905 (Unix) host has changed; numbers after the final #\. in a
1906 namestring are no longer interpreted as a version field. This is
1907 intented to be largely invisible to the user, except that the
1908 meaning of the namestring "*.*.*" has changed: it now refers to a
1909 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1910 should usually be replaced by
1911 (make-pathname :name :wild :type :wild :version :wild)
1912 with the added benefit that this is more likely to be portable.
1913 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1914 option to OPEN now signals an error if the file being opened
1915 exists; this may have an impact on existing code.
1916 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1917 signals received, on the PowerPC platforms (both Linux and
1918 Darwin). (thanks to Pierre Mai for pointing out the location of
1920 * several fixes on OS X: The system now builds and runs cleanly on
1921 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1922 (thanks to Brian Mastenbrook)
1923 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1924 second-resolution: integer multiples of 1/3600 between -24 and 24.
1925 (thanks to Vincent Arkesteijn)
1926 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1927 type when called with 1 argument; PEEK-CHAR checked type of
1928 PEEK-TYPE only after having read first character from a
1929 stream. (reported by Peter Graves)
1930 * bug fix: the garbage collector now has much better locality
1931 behaviour, and in particular no longer treats all memory as being
1932 exhausted when many small objects point to each other in a deeply
1934 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1935 recognized as being TYPEP their class.
1936 * bug fix: the PUSHNEW documentation string has been corrected.
1937 (thanks to Vincent Arkesteijn)
1938 * bug fix: defaulting of the value for the last of an atypically
1939 large number of multiple values being bound was not being
1940 performed correctly on the Alpha or PPC platforms
1941 * optimization: implemented multiplication as a modular
1942 (UNSIGNED-BYTE 32) operation on the x86 backend.
1943 * optimization: SEARCH on simple-base-strings can now be open-coded.
1944 (see also contrib/compiler-extras.lisp for inspiration for
1945 teaching the compiler about the Boyer-Moore algorithm).
1946 * value, returned by MAX (and MIN) called with several EQUALP, but
1947 not EQL, arguments now does not depend on compiler settings.
1948 * fixed some bugs revealed by Paul Dietz' test suite:
1949 ** in stack analysis liveness information is propagated from
1950 non-local entry points.
1951 ** pathwise CAST removing failed when the CAST node did not start
1953 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1954 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1956 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1957 host is already defined.
1958 ** RENAME-FILE works on streams instead of signalling an internal
1960 ** PEEK-CHAR uses the current readtable when determining whether
1961 or not a character is whitespace.
1962 ** MERGE-PATHNAMES handles the case when the pathname does not
1963 specify a name while the default-pathname specifies a version.
1964 ** Pathnames now stand a better chance of respecting print/read
1966 ** Attempting to use standardized file system operators with a
1967 pathname with invalid :DIRECTORY components signals a
1969 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1971 ** DIRECTORY on logical pathnames is more correct.
1972 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1973 signal a TYPE-ERROR if their argument is not a stream.
1974 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1975 designator argument does not designate a stream.
1976 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1977 examining the synonym.
1978 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1980 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1981 element-type larger than ([UN]SIGNED-BYTE 32).
1983 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1984 * When built with the :SB-FUTEX feature, threaded builds now take
1985 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1986 for faster/more reliable mutex and condition variable support.
1987 * Incompatible change (but one you probably shouldn't have been using
1988 anyway): the interface and code for arbitrating between multiple
1989 threads in the same user session has been redesigned.
1990 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1991 error when called without an explicit environment argument.
1992 (thanks to Nikodemus Siivola)
1993 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1994 confused about their position. (thanks to Adam Warner and Gerd
1996 * bug fix: Pretty printing backquoted forms with unquotations in the
1997 argument list position of various code constructs such as LAMBDA
1998 now works correctly. (reported by Paul Dietz)
1999 * bug fix: Pretty printing unquotations no longer loses all
2000 stream position information.
2001 * optimization: performance of string output streams is now less
2002 poor for multiple small sequence writes.
2003 * optimization: performance of CSUBTYPEP in the presence of complex
2004 expressions involving CONS and NOT many times has been improved.
2005 (reported by Paul Dietz)
2006 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2008 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2009 streams for the REPL, for the debugger, and so on. (thanks to
2011 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2012 * fixed some bugs revealed by Paul Dietz' test suite:
2013 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2014 is ignored if it is too small, rather than propagating through
2015 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2016 ** extremely complex negations of CONS types were not being
2017 sufficiently canonicalized, leading to inconsistencies in
2019 ** VALUES tranformer lost derived type.
2021 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2022 * fixed a bootstrapping bug: the build process no longer assumes
2023 that the various BOOLE-related constants have the same value in
2024 host and target lisps. (noted by Paul Dietz' test suite on an
2025 SBCL binary built from CLISP)
2026 * The system can now be dynamically linked on the MIPS platform,
2027 which enables dynamic loading of foreign code from Lisp. (thanks
2028 to Ralf Baechle for discussions on the MIPS ABI)
2029 * The system now records debugging information for its own source
2030 files in a filesystem-position-independent manner, relative to
2031 the "SYS" logical host.
2032 * fixed a compiler bug: MV-LET convertion did not check references
2033 to the "max args" entry point. (reported by Brian Downing)
2034 * tweaked disassembly notes to be less confident about proclaiming
2035 some instruction as an LRA. (thanks to Brian Downing)
2036 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2037 now each have their own history, command character, and other
2038 characteristics. (thanks to David Lichteblau)
2039 * fixed some bugs revealed by Paul Dietz' test suite:
2040 ** compiler failure in compiling LOGAND expressions including a
2042 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2043 shift greater than 32.
2044 ** FLUSH-DEST did not mark blocks for type check regeneration.
2045 ** HANDLER-CASE failed to accept declarations in handler clauses
2046 in some circumstances.
2048 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2049 * New code in contrib/sb-introspect (still probably not entirely
2050 stable yet) provides some support for smart Lisp development
2051 environments like SLIME.
2052 * The conditions signalled for errors occurring when loading .fasl
2053 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2054 in a way which should help ASDF recover gracefully.
2055 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2056 stuff has been cleaned up. If you code contrib/ stuff, this might
2057 affect you, and you can look at contrib/README, contrib/STANDARDS,
2058 and/or the 0.8.4.27 diff to check.
2059 * In full calls the compiler now does not generate checks for declared
2060 argument types for all arguments.
2061 * various threading fixes
2062 ** and some experimental patches which didn't make it into
2063 the main tree for this release, but which are shipped in
2064 contrib/experimental-thread.patch as a possible fix for some
2065 failures (deadlock, spinning...) in GC-intensive multithreaded
2067 * fixed PPC build problem (source code incompatibility of different
2068 library versions): added offsetof() hackery which attempts to divine
2069 where glibc maintainers put uc_mcontext today
2070 * fixed bug 282: compiler does not trust type assertions while passing
2071 arguments to a full call.
2072 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2073 &OPTIONAL VALUES type parameter.
2074 * fix bug 214: algorithm for noting rejected templates is now more
2075 similar to that of template seletion. (also reported by rydis on
2077 * fixed bug 141b: printing backquoted information readably and prettily
2078 inserts a space where necessary.
2079 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2080 (CHARACTER 10) are now reported as errors, rather than propagated
2081 as unknown types. (reported by piso on #lisp)
2082 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2083 with values NIL and :ERROR. (thanks to Milan Zamazal)
2084 * fixed bug 191c: CLOS now does proper keyword argument checking as
2085 described in CLHS 7.6.5 and 7.6.5.1.
2086 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2087 longer raise an error; further, using a list as a for-as-arithmetic
2088 counter now raises a meaningful error.
2089 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2090 now understood by sequence creation functions such as MAKE-SEQUENCE
2092 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2093 a STRING-INPUT-STREAM.
2094 * compiler enhancement: SIGNUM is now better able to derive the type
2096 * type declarations inside WITH-SLOTS are checked. (reported by
2098 * fixed some bugs revealed by Paul Dietz' test suite:
2099 ** incorrect optimization of TRUNCATE for a positive first
2100 argument and negative second.
2101 ** compiler failure in let-convertion during flushing dead code.
2102 ** compiler failure while deriving type of TRUNCATE on an
2103 interval, containing 0.
2104 ** ASH of a negative bignum by a negative bignum count now returns
2106 ** intersection of CONS types now canonicalizes properly, fixing
2107 inconsistencies in SUBTYPEP.
2109 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2110 * incompatible change: The --disable-debugger command line
2111 option now clobbers the debugger at a more fundamental
2112 level, by redefining #'INVOKE-DEBUGGER instead of by
2113 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2114 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2115 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2116 enter the debugger REPL and then suffer recursive errors
2117 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2118 while under the new system, BREAK in --disable-debugger mode
2119 terminates the system just as an unhandled error would.
2120 * fixed compiler performance when processing loops with a step >1;
2121 * bug fix: DOCUMENTATION now retrieves generic function
2122 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2123 support has been systematized, and now supports the methods
2124 specified by ANSI, along with a default method and a method for
2125 slot documentation. (reported by Nathan Froyd)
2126 * bug fix: effective methods associated with a generic function are
2127 no longer cached over a change of that generic function's method
2128 combination. (reported by Andreas Fuchs)
2129 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2130 names a non-existent directory. (thanks to Andreas Fuchs)
2131 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2132 the PPC platform, lead to stack corruption; this has been fixed.
2133 (reported by Rainer Joswig)
2134 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2135 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2136 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2137 arithmetic in general, is now much more reliable on the PPC
2139 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2140 platform now returns the right answer.
2141 * optimization: restored some effective method precomputation in
2142 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2143 precomputation is now tunable.
2144 * optimization: compiler-internal data structure use has been
2145 reviewed, and changes have been made that should improve the
2146 performance of the compiler by about 20%.
2147 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2148 simple-base-strings and simple-bit-vectors is improved.
2149 * optimization: the optimization of 32-bit logical and arithmetic
2150 functions introduced in version 0.8.3 on the x86 has been
2151 implemented on the mips, ppc and sparc platforms; an
2152 implementation of the same facility, but for 64-bit arithmetic,
2153 has been added for the alpha.
2154 * microoptimization: the compiler is better able to make use of the
2155 x86 LEA instruction for multiplication by constants.
2156 * bug fix: in some situations compiler did not report usage of
2157 generic arithmetic in (SPEED 3) policy.
2158 * bug 145b fix: compiler used wrong type specifier while converting
2159 MEMBER-types to numeric.
2160 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2162 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2163 index argument now works properly on the Alpha platform.
2164 * bug fix: floating point exception treatment on the Alpha platform
2166 * bug fix: FILE-POSITION works much better on string input and
2167 output streams. (thanks to Nikodemus Siivola)
2168 * bug fix: many threading/garbage collection symptoms sorted.
2169 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2170 might be pseudo-atomic.
2171 * internal change: Stopping for GC is now done with signals not
2172 ptrace. GC is now done in whichever thread wanted it, instead of
2174 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2176 * bug fix: result form in DO is not contained in the implicit
2178 * incompatible change: ICR structure is changed; the value part of
2179 CONTINUATION is now called LVAR; corresponding functions are
2180 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2181 * added type deriver for ISQRT (thanks to Robert E. Brown).
2182 * bug fix: better support for loading from the command line when an
2183 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2185 * fixed some bugs revealed by Paul Dietz' test suite:
2186 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2187 ** ROUND and FROUND now give the right answer when given very
2188 small float arguments.
2189 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2191 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2192 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2193 ** LCM should return a non-negative integer.
2194 ** PARSE-INTEGER returned the index of a terminator instead of the
2195 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2196 ** PARSE-INTEGER returned an incorrect index being applied to a
2198 ** LCM with two arguments of 0 returns 0 rather than signalling
2200 ** unsigned addition of a 32-bit constant with the high bit set no
2201 longer causes an internal compiler error on the x86.
2202 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2203 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2204 with negative last argument.
2205 ** byte specifiers with zero size and position no longer cause
2206 an error during type derivation.
2207 ** bignum multiplication on the Alpha platform now returns the
2209 * porting: The system now builds on SuSE AMD64, although it still
2210 generates a 32-bit binary.
2211 * .fasl file incompatibility: The fasl file version number has
2212 been incremented (because of the changes to internal compiler
2213 data structures referred to above).
2215 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2216 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2217 more accurately, on the Darwin kernel running on PowerPC hardware.
2218 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2219 * Compiler code deletion notes now signal a condition of type
2220 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2221 an associated MUFFLE-WARNING restart.
2222 * The compiler now performs limited argument count validation of
2223 constant format strings in FORMAT, and where appropriate in ERROR,
2224 CERROR and WARN. (thanks to Gerd Moellmann)
2225 * New ASDF-INSTALL contrib can be used for automatic download and
2226 installation of third-party Lisp code from CCLAN or other sites
2228 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2229 which forces another thread to execute a function supplied by the
2231 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2232 now accept and act upon their :ELEMENT-TYPE keyword argument.
2233 (reported by Martin Atzmueller, Edi Weitz)
2234 * bug fix: FILE-POSITION now accepts position designators up to
2235 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2236 is the greater. (thanks to Patrik Nordebo)
2237 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2238 Kalle Olavi Niemitalo)
2239 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2240 printable. (reported by Eric Marsden)
2241 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2242 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2243 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2244 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2245 (thanks to Stig Sandoe)
2246 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2247 UNREAD-CHAR now correctly includes the unread character in the
2248 target sequence. (thanks to Gerd Moellmann)
2249 * bug fix in threaded builds: the system can now be suspended and
2250 resumed by shell job control with minimal disruption.
2251 * bug fixes in times and timezones >2038AD
2252 * better handling of "where is GNU make?" problem in build scripts
2253 (thanks to Nikodemus Siivola)
2254 * new optimization: inside a named function any reference to a
2255 function with the same name is considered to be a self-reference;
2256 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2257 optimization quality.
2258 * new optimization on x86: logical functions and + now have
2259 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2260 used when the result is truncated to 32 bits.
2261 * VALUES declaration is partially enabled.
2262 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2263 * fixed some bugs revealed by Paul Dietz' test suite:
2264 ** The system now obeys the constraint imposed by
2265 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2266 types form a lattice under type intersection.
2267 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2268 ** ASSOC now ignores NIL elements in an alist.
2269 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2270 and (1+ MOST-POSITIVE-FIXNUM) answers.
2271 ** The addition of a method with invalid qualifiers to a generic
2272 function does not cause an error to be signalled immediately;
2273 a warning is signalled, and the error is generated only on
2274 calling the generic function.
2275 * changed .fasl file version number, in part to add type codes for
2276 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2277 obscure ANSI requirements
2279 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2280 * fixed bug 148: failure to inline-expand a local function left
2281 garbage, confusing the compiler.
2282 * fixed bugs 3cd: structure slot readers perform type check if the
2283 slot can have an invalid value (i.e. it is either not initialized
2284 or can be written with a less specific slot writer).
2285 * bug fix: the compiler now traps array references to elements off
2286 the end of an array; previously, the bounds checking in some
2287 circumstances could go off-by-one.
2288 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2290 * type declarations for array element types now obey the description
2291 on the CLHS page "Declaration TYPE", as per discussions on
2292 sbcl-help around 2003-05-08. This means that a declaration
2293 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2294 declaration, all references to BAR will be asserted or assumed
2295 (with THE, so dependent on compiler policy) to involve objects of
2296 type FOO. Note that no such declaration is implied in
2297 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2298 * declared types of functions from the "Conditions"
2299 chapter. (reported by Paul Dietz)
2300 * bug fix: CERROR accepts a function as its first argument.
2301 * bug fix: NTH an NTHCDR accept a bignum as index
2302 arguments. (reported by Adam Warner)
2303 * optimization: character compare routines now optimize comparing
2304 against a constant character. (reported by Gilbert Baumann)
2305 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2306 argument now works properly on the MIPS platform.
2307 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2308 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2309 declared to have a functional type, does not cause an error
2311 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2312 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2313 * fixed bug: if last continuation of a deleted block has a
2314 destination, this destination should be deleted too. (reported by
2316 * fixed a bug in the bootstrap process: the host compiler's values
2317 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2318 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2319 test case from Patrik Nordebo)
2320 * improved the ability of the disassembler on the PPC platform to
2321 provide helpful disassembly notes.
2322 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2323 the class in more cases than previously.
2324 * bug fix: FILE-POSITION now understands :START and :END for
2325 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2326 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2327 8) data. (thanks to David Lichteblau)
2328 * bug fix: it is possible to add a method to a generic function
2329 without lambda list.
2330 * bug fix: reader failed to signal END-OF-FILE inside an
2331 object representation. (reported by Nikodemus Siivola)
2332 * fixed some bugs revealed by Paul Dietz' test suite:
2333 ** LAST and [N]BUTLAST should accept a bignum.
2334 ** condition slot accessors are methods.
2335 ** (VECTOR NIL) is a subtype of STRING.
2337 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2338 * minor incompatible change: some nonsensical specialized lambda
2339 lists (used in DEFMETHOD) which were previously ignored now signal
2341 * minor incompatible change: the system is now aware of the types of
2342 variables in the COMMON-LISP package, and will signal errors for
2343 most violations of these type constraints (where previously they
2344 were silently accepted).
2345 * minor incompatible change: COMPILE-FILE now uses the freedom
2346 afforded (ANSI 3.2.2.3) to use derived function types for
2347 functions defined in the same file. This also permits the system
2348 to warn on static type mismatches and function
2349 redefinition. (Currently it does not work with high DEBUG level.)
2350 * minor incompatible change: VALUES declaration is disabled.
2351 * When issuing notes, the compiler now signals a condition of type
2352 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2353 restart for use in user handlers. It is expected that the
2354 COMPILER-NOTE condition will eventually become a condition
2355 supertype to a hierarchy of note types, which will then be
2356 handleable in a similar fashion. However, at the moment, no such
2357 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2358 but it's an implementation detail, not a classification for the
2360 * Changes in type checking closed the following bugs:
2361 ** type checking of unused values (192b, 194d, 203);
2362 ** template selection based on unsafe type assertions (192c, 236);
2363 ** type checking in branches (194bc).
2364 * A short form of VALUES type specifier has ANSI meaning (it has
2365 increased the number of situations when SBCL cannot perform type
2367 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2368 keywords or constants is permissible.
2369 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2370 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2371 classes. (thanks to Antonio Martinez)
2372 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2373 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2374 argument to a pretty-print stream. Instead, it leaves any such
2375 operations to DESCRIBE-OBJECT methods.
2376 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2378 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2379 lambda lists. (thanks to David Lichteblau)
2380 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2381 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2382 (thanks to Gerd Moellmann)
2383 * fixed bug 246: increased compilation speed of long
2384 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2386 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2388 * DEFCONSTANT now throws a condition of type
2389 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2390 constant to a non-EQL value; CONTINUE and ABORT restarts
2391 respectively change and preserve the value.
2392 * fixed bug 63: The code walker, part of the implementation of CLOS,
2393 is now better at handling symbol macros.
2394 * bug fix: There is no longer an internal implementation type named
2395 CL:LENGTH. (reported by Raymond Toy)
2396 * bug fix: In macro-like defining macros/special operators the
2397 implicit block does not enclose the lambda list.
2398 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2399 enclosed in parentheses) are not suitable as type specifiers, and
2400 their use properly signals an error now.
2401 * bug fix: An argument count mismatch for a type specifier in code
2402 being compiled no longer causes an unhandled error at compile
2403 time, but signals a compile-time warning.
2404 * fixed simple vector readable printing
2405 * bug fix: DESCRIBE takes more care over whether the class
2406 precedence list slot of a class is bound before accessing it.
2407 (reported by Markus Krummenacker)
2408 * bug fix: FORMATTER can successfully compile pretty-printer format
2409 strings which use variants of the ~* directive inside.
2410 * bug fix: SEARCH now applies its TEST predicate to the elements of
2411 the arguments in the correct order. (thanks to Wolfhard Buss)
2412 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2413 in type inference. (thanks to Robert E. Brown)
2414 * bug fix: Reading in symbols with an explicit package name of ""
2415 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2416 * fixed some bugs revealed by Paul Dietz' test suite:
2417 ** NIL is now allowed as a structure slot name.
2418 ** Arbitrary numbers, not just REALs, are allowed in certain
2419 circumstances in LOOP for-as-arithmetic clauses.
2420 ** Multiple class redefinitions before slot access no longer
2421 causes a type error.
2422 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2423 association between the name and a class.
2424 ** Generic functions with non-standard method-combination and over
2425 five methods all of which return constants no longer return NIL
2426 after the first few invocations. (thanks to Gerd Moellmann)
2427 ** CALL-NEXT-METHOD with no arguments now passes the original
2428 values of the arguments, even in the presence of assignment.
2429 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2431 ** DISASSEMBLE works with closures and funcallable instances.
2432 ** ADD-METHOD now returns the generic function, not the new method.
2433 ** FIND-METHOD signals an error if the lengths of the specializers
2434 is incompatible with the generic function, even if the ERRORP
2436 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2437 which its argument is a member.
2438 ** DEFCLASS only redefines the class named by its class-name
2439 argument if that name is the proper name of the class;
2440 otherwise, it creates a new class.
2441 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2442 of the UNBOUND-SLOT condition to the name of the slot.
2443 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2444 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2445 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2446 SLOT-MAKUNBOUND in the specified fashion.
2448 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2449 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2450 cross-compilation host. As a consequence, we can now bootstrap our
2451 way up to SBCL starting with a bare gcc toolchain and human-readable
2452 source code (first the source to CLISP, then the source to SBCL).
2453 * A contributed module containing a partial implementation of the
2454 simple-streams interface has been included. (thanks to Rudi
2456 * A contributed module implementing the RFC1321 Message Digest
2457 Algorithm, known as MD5, has been included.
2458 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2459 no longer has any effect, as the code controlled by this feature
2460 has been deleted. (As far as we know, no-one has ever built using
2461 this feature, and its semantics were confused in any case).
2462 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2463 work on conditions (as required by the ANSI specification),
2464 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2465 expected behaviour on conditions. Users should note, however,
2466 that such behaviour is not required by the ANSI specification,
2467 and so use of this behaviour may render their code unportable.
2468 * fixed some bugs revealed by Paul Dietz' test suite:
2469 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2471 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2472 ** SXHASH on condition objects no longer returns NIL.
2473 ** :ALLOCATION :CLASS slots are better treated; their values are
2474 updated on class redefinition, and initforms inherited from
2475 superclasses are applied.
2476 ** REMOVE-METHOD returns its generic function argument even when
2477 no method was removed.
2478 ** SHARED-INITIALIZE now initializes the values of the requested
2479 slots, including those with :ALLOCATION :CLASS.
2480 ** ALLOCATE-INSTANCE now works on structure classes defined via
2481 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2483 ** SLOT-EXISTS-P now works on conditions, as well as structures
2485 ** MAKE-LOAD-FORM now has the required methods on
2486 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2487 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2488 keyword, and now implements the SLOT-NAMES argument.
2489 ** methods with &OPTIONAL arguments no longer allow too many
2490 arguments to be passed in the call without error.
2491 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2492 option is consistent with the required arguments of the generic
2493 function lambda list.
2494 * bug fix: REQUIRE accepts a string designator. (Thanks to
2496 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2497 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2498 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2499 * bug fix: The compiler now checks for duplicated variables in macro
2501 * bug fix: SETQ on globals returns the correct value.
2502 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2503 not a COMPILER-ERROR (followed by some other strange error on
2504 choosing the CONTINUE restart).
2505 * bug fix: make.sh and friends are now more consistent in the way that
2506 they look for GNU "make".
2508 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2509 * experimental native threads support (on x86 Linux >=2.4 only).
2510 This is not compiled in by default: you need to add :SB-THREAD to
2511 the target features. See the "Beyond ANSI" chapter of the manual
2513 * fix for longstanding nonANSIism: The old distinction between
2514 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2515 The return value from CL:FIND-CLASS is now a CLOS class, and
2516 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2517 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2518 * An interface to the MetaObject Protocol, as described in Kiczales,
2519 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2520 MIT Press, 1991, is available from the SB-MOP package.
2521 * incompatible change: the SB-PCL package should now be considered
2522 a private implementation detail, and no longer a semi-private MOP
2524 * minor incompatible change: due to rearrangement for threads, the
2525 control stack and binding stack are now allocated at arbitrary
2526 addresses instead of being hardcoded per-port. Users affected by
2527 this probably have to be doing advanced things with shared
2528 libraries, and will know who they are.
2529 * minor incompatible change: Previously, all --eval forms used were
2530 processed with READ before any of them were processed with EVAL.
2531 Now each --eval form is processed with both READ and EVAL before
2532 the next --eval form is processed. (Thus package operations like
2533 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2534 work as the user might reasonably expect.)
2535 * minor incompatible change: *STANDARD-INPUT* is now only an
2536 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2538 * minor incompatible change: Y-OR-N-P is now character-oriented,
2539 not line oriented. Also, YES-OR-NO-P now works without errors.
2540 (thanks to Antonio Martinez)
2541 * sb-aclrepl module improvements: an integrated inspector, added
2542 repl features, and a bug fix to :trace command.
2543 * Known functions, which cannot be open coded by the backend, are
2544 considered to be able to check types of their arguments. (fixing
2545 a bug report by Nathan J. Froyd)
2546 * fixed a bug in computing method discriminating functions: It is
2547 now possible to define methods specialized on classes which have
2548 forward-referenced superclasses. (thanks to Gerd Moellmann)
2549 * fixed evaluation order in optional entries (reported by Gilbert
2551 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2552 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2553 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2554 required-by-AMOP NAME argument, as well as CLASS and
2555 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2556 * fixed bug 20: DEFMETHOD can define methods using names that are
2557 not the proper names of classes to designate class specializers.
2558 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2559 Unix file descriptors, instead of blowing up. (thanks to Antonio
2561 * Garbage collection refactoring: user-visible change is that a
2562 call to the GC function during WITHOUT-GCING will not do garbage
2563 collection until the end of the WITHOUT-GCING. If you were doing
2564 this you were probably losing anyway.
2565 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2566 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2567 TYPEP the latter but not the former.
2568 * The compiler issues a full WARNING for calls to undefined functions
2569 with names from the CL package.
2570 * MAP-INTO for a vector destination is open coded. (reported by
2571 Brian Downing on c.l.l)
2572 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2573 documentation string.
2574 * fixed some bugs revealed by Paul Dietz' test suite:
2575 ** COPY-ALIST now signals an error if its argument is a dotted
2577 ** Condition slots are now accessed more correctly in the presence
2578 of multiple initargs for a given slot.
2579 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2580 exclude restarts of the same name associated with a different
2582 ** DEFCLASS of forward-referenced classes with another
2583 forward-referenced class in the superclasses list no longer
2585 ** Condition slots are now initialized once each, not multiple
2586 times. (thanks to Gerd Moellmann)
2587 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2589 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2590 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2591 arguments contain duplicated elements.
2592 ** RESTART-CASE understands local macros.
2593 ** RESTART-CASE associates exactly its own restarts with a condition.
2594 ** ENDP in safe mode checks its argument to be of type LIST.
2595 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2596 in question is unbound.
2597 ** Optimized MAKE-INSTANCE functions no longer cause internal
2598 assertion failures in the presence of duplicate initargs.
2599 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2600 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2601 propagated even in the ctor optimized implementation of
2603 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2605 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2606 * a better implementation of SXHASH on (simple) bit vectors,
2607 measured both in execution speed and in distribution of results
2608 over the positive fixnums, has been installed. Likewise, a better
2609 implementation of EQUAL for simple bit vectors is now available.
2610 * fixed CEILING optimization for a divisor of form 2^k.
2611 * fixed bug 240 (emitting extra style warnings "using the lexical
2612 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2613 by Antonio Martinez)
2614 * fixed SXHASH, giving different results for NIL depending on type
2615 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2616 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2617 variable. (found by Rolf Wester)
2618 * SBCL does not ignore type declarations for special
2619 variables. (reported by rif on c.l.l 2003-03-05)
2620 * some bug fixes in contrib/sb-aclrepl/
2621 * fixed some bugs revealed by Paul Dietz' test suite:
2622 ** a bug in the CONS type specifier, whereby the CAR and CDR
2623 types got intertwined, has been fixed;
2624 ** the type system is now able to reason about the interaction
2625 between INTEGER and RATIO types more completely;
2626 ** APPEND, [N]REVERSE and NRECONC check that those their
2627 arguments, which must be proper lists, are really so;
2628 ** An array specialized to be unable to hold elements has been
2629 implemented, as required -- yes, really -- by ANSI;
2630 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2631 on malformed property lists;
2633 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2634 * incompatible packaging change: in line with Unix convention,
2635 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2636 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2637 * REQUIRE and PROVIDE are now optionally capable of doing something
2638 useful. See the documentation string for REQUIRE.
2639 * infrastructure for a managed SBCL contrib system: contributed
2640 modules in this release include:
2641 ** the ASDF system definition facility;
2642 ** an interface to the BSD Sockets API;
2643 ** an ACL-like convenience interface to the repl;
2644 (thanks to Kevin Rosenberg)
2645 ** an implementation of ROTATE-BYTE, with efficient implementation
2647 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2648 gives a closer approximation to the right answer than previously.
2649 (thanks to Raymond Toy)
2650 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2651 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2652 optional environment arguments, as required by ANSI.
2653 * fixed bugs in other functions taking environment objects, allowing
2654 calls with an explicit NIL environment argument to be compiled
2656 * fixed bug 228: primary return values from
2657 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2658 COMPILE or FUNCTION.
2659 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2660 :TYPE will no longer signal errors on innocuous objects.
2661 * fixed bug 231b: SETQ is better at respecting type declarations in
2662 the lexical environment.
2663 * fixed a bug in DEFCLASS: classes named by symbols with no or
2664 unprintable packages can now be defined.
2665 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2666 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2667 * fixed compiler failure related to checking types of functions.
2668 (reported by Robert E. Brown)
2669 * the compiler is now much more consistent in its error-checking
2670 treatment of bounding index arguments to sequence functions: in
2671 (SAFETY 3) code, errors will be signalled in almost all cases if
2672 invalid sequence bounding indices are passed to functions defined
2673 by ANSI to operate on sequences.
2674 * fixed a bug in the build procedure: documentation of SBCL-specific
2675 packages is now preserved and available in the final Lisp image.
2676 * lifted FDEFINITION lookup out of loops in the implementation of
2677 many list operations. (thanks to Robert E. Brown)
2678 * fixed a bug in the reader: the #n# reader macro now works for
2679 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2680 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2681 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2682 E. Brown and Tony Martinez respectively)
2683 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2684 SLOT-BOUNDP: in particular, it is now able to optimize them much
2685 better, and is now not vulnerable to having packages renamed.
2686 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2687 always called when a slot is not present in an instance. (thanks
2689 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2690 longer signal an error on unreading a character following EOF on
2691 the previous constituent stream. (thanks to Tony Martinez)
2692 * fixed some bugs revealed by Paul Dietz' test suite:
2693 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2694 not just nonnegative fixnums;
2695 ** the logical bit-array operators such as BIT-AND now accept an
2696 explicit NIL for their "opt-arg" argument (to indicate a
2697 freshly-consed result bit-array);
2698 ** ELT now signals an error on an invalid sequence index in safe
2700 ** the type system is now cleverer about negations of numeric
2701 types, and consequently understands the BIGNUM and RATIO types
2703 ** the type system is now cleverer about the interaction between
2704 INTEGER and RATIO types: while bugs still remain, many more
2705 cases are accurately computed;
2706 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2707 if it is in the last clause;
2708 ** CONSTANTLY now correctly returns a side-effect-free function in
2710 ** DECLARE is no longer treated as a special-operator; in
2711 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2712 * incremented fasl file version number due to the change in the
2713 DEFSTRUCT-SLOT-DESCRIPTION structure.
2715 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2716 * minor incompatible change: code processed by the "interpreter" or
2717 EVAL now has a compilation optimization policy of (DEBUG 2)
2718 (changed from (DEBUG 1)) to improve debuggability of interactive
2719 development, and to allow the use of the debug RETURN command in
2721 * an experimental implementation of the RETURN command for the
2722 debugger has been included. (thanks to Frederik Kuivinen)
2723 * fixed bug 62: constraints were not propagated into a loop.
2724 * fixed bug in embedded calls of SORT (reported and investigated by
2726 * fixed some bugs revealed by Paul Dietz' test suite:
2727 ** printing and reading of arrays with some dimensions having
2728 length 0 (thanks to Gerd Moellmann);
2729 ** BOA constructor with &AUX argument without a default value does
2730 not cause a type error;
2731 ** CONSTANTP now returns true for all self-evaluating objects.
2733 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2734 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2735 accessors that are related by inheritance, as specified in the
2736 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2737 Valtteri Vuorikoski)
2738 * The compiler is now able to inline functions that were defined in
2739 a complex lexical environment (e.g. inside a MACROLET).
2740 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2742 * Support for the upcoming FreeBSD-5.0 release has been included.
2743 (thanks to Dag-Erling Smorgrav)
2744 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2745 effect when it is not in a toplevel context.
2746 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2747 stand a better chance of being correct. (thanks to Gerd
2749 * fixed bug in COERCE, which now signals an error on coercing a
2750 rational to a bounded real type which excludes the expected
2752 * The compiler is now able to derive types more accurately from the
2753 COERCE and COMPILE functions.
2754 * fixed bug 223: functional binding is considered to be constant
2755 only for symbols in the CL package.
2756 * fixed bug 231: SETQ did not check the type of a variable being set
2757 (reported by Robert E. Brown)
2758 * A new optimization for MAKE-INSTANCE has been included, fixing
2759 various bugs (including relating to :ALLOCATION :CLASS slots and
2760 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2762 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2763 ** As required by ANSI, LOOP now disallows anonymous collection
2764 clauses such as COLLECT I in conjunction with aggregate boolean
2765 clauses such as THEREIS (= I 1);
2766 ** LOOP now signals an error when any variable is reused in the
2767 same loop (including the potentially useful construct analogous
2768 to WITH A = 1 WITH A = (1+ A);
2769 ** IT is only a special loop symbol within the first clause of a
2770 conditional loop clause;
2771 ** LOOP with a typed iteration variable over a hashtable now
2772 signals a type error iff it should.
2773 * fixed some other bugs revealed by Paul Dietz' test suite:
2774 ** FILE-STREAM now names the class previously known as FD-STREAM;
2775 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2776 argument) no longer signals an error;
2777 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2778 of the slot symbol, rather than using the current package
2779 ((:CONC-NAME "") continues to intern the slot's name in the
2781 * incremented fasl file version number, because of the incompatible
2782 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2783 of the new implementation of DEFINE-COMPILER-MACRO.
2785 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2786 * Support for building SBCL for MIPS platforms running in
2787 little-endian mode has now been checked in, and basic
2788 functionality on said platforms verified.
2789 * minor incompatible change: PCL now records the pathname of a file
2790 in which methods and the like are defined, rather than its
2792 * minor incompatible change: TRUENAME now considers the truename of
2793 a file naming a directory to be the pathname with :DIRECTORY
2794 component indicating that directory.
2795 * minor incompatible change: a NAMED clause in the extended form of
2796 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2797 reason for the previous behaviour is unclear.
2798 * more systematization and improvement of CLOS and MOP conformance
2799 in PCL (thanks to Gerd Moellman and Pierre Mai):
2800 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2802 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2803 primary methods with no specializers;
2804 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2806 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2807 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2808 FORWARD-REFERENCED-CLASSes; error reporting on
2809 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2811 ** SXHASH on CLOS instances now uses a slot internal to the
2812 instance to return different numbers on distinct instances,
2813 while preserving the same return value through invocations of
2815 ** DEFMETHOD signals errors when methods with longer incongruent
2816 lambda lists are added to generic functions;
2817 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2818 CLASS, as specified in AMOP;
2819 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2820 on the order returned by the primary method for classes of
2821 class STANDARD-CLASS;
2822 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2823 * fixed some bugs shown by Paul Dietz' test suite:
2824 ** DOLIST puts its body in TAGBODY;
2825 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2827 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2828 value producing form;
2829 ** if more variables are given to PROGV than values, extra
2830 variables are bound and made to have no value;
2831 ** NSUBSTITUTE on list arguments gets the right answer with
2833 ** ELT signals an error of type TYPE-ERROR when the index argument
2834 is not a valid sequence index;
2835 ** LOOP signals (at macroexpansion time) an error of type
2836 PROGRAM-ERROR when duplicate variable names are found;
2837 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2838 ** FUNCALL of special-operators now cause an error of type
2840 ** PSETQ now works as required in the presence of side-effecting
2841 symbol-macro places;
2842 ** NCONC accepts any object as its last argument;
2843 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2845 ** loop-for-as-package does not require a package to be explicitely
2847 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2849 * fixed bug 166: compiler preserves "there is a way to go"
2850 invariant when deleting code.
2851 * fixed bug 172: macro lambda lists with required arguments after
2852 &REST arguments now cause an error to be signalled. (thanks to
2854 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2855 bug. (thanks to Matthew Danish)
2856 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2858 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2859 arguments are lexically rebound. (thanks to Gerd Moellmann and
2861 * fixed bug 194: error messages are now more informative when there
2862 is no primary method applicable in a call to a generic
2863 function. (thanks to Gerd Moellmann)
2864 * fixed bug in command line argument checking (thanks to Julian
2866 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2867 (thanks to Matthew Danish)
2868 * incremented fasl file version number, because of the
2869 SXHASH-related changes in the layout of CLOS data structures
2871 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2872 * minor incompatible change: The runtime (the Unix executable named
2873 "sbcl") is now much pickier about the .core files it will load.
2874 Essentially it now requires .core files to descend from the same
2875 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2876 as the runtime does. (The intent is to prevent the crashes which
2877 can occur, and which can even be reported as mysterious failures,
2878 when people patch the sources or change the build parameters
2879 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2880 sbcl and .core files.)
2881 * fixed bug: VALUES-LIST is no longer optimized away.
2882 * fixed bug 142: The FFI conversion of C string values to Lisp
2883 string values no longer conses excessively. (thanks to Nathan
2884 Froyd porting Raymond Toy's fix to CMU CL)
2885 * began to systematize and improve MOP conformance in PCL (thanks to
2886 Nathan Froyd, Gerd Moellman and Pierre Mai):
2887 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2889 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2890 ** FINALIZE-INHERITANCE is now called on class finalization;
2891 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2892 argument precedence order.
2893 * fixed bug 202: The compiler no longer fails on functions whose
2894 derived types contradict their declared type.
2895 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2896 so it can be non-toplevel.
2897 * The fasl file version number has changed (because of the new
2898 implementation of DEFMACRO).
2899 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2900 safe code, that any length requirement by their type-specifier
2901 argument is valid. The exceptions to this are described in bug
2903 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2904 functions now check the types of their inputs as required by ANSI.
2905 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2906 introduced symbol is DECLAREd to be SPECIAL.
2907 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2908 * fixed bug: :COUNT argument to sequence functions may be negative
2909 * fixed bug: body of DO-SYMBOLS may contain declarations
2910 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2911 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2912 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2913 symbol macro only once
2914 * fixed printing of call frame when argument list is unavailable
2915 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2916 * compiler no longer signals WARNING on unknown keyword
2919 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2920 * A beta-quality port to the mips architecture running Linux,
2921 based on the old CMUCL backend, has been made. It has been tested
2922 on a big-endian kernel, and works sufficiently well to be able to
2923 rebuild itself; it has not been tested in little-endian mode.
2924 * fixed an inconsistency between gencgc.c and purify.c which made
2925 dumping/loading .core files unreliable
2926 * fixed bug 120a: The compiler now deals correctly with IFs where
2927 the consequent is the same as the alternative, instead of
2928 misderiving the return type. (thanks to Alexey Dejneka)
2929 * fixed bug 113: Logical pathnames are now dumpable (the logical
2930 host is resolved at load-time, throwing an error if it is not
2932 * fixed bug 174: FORMAT's error message is slightly clearer when a
2933 non-printing character is used in a format directive.
2934 * fixed several bugs in compiler checking of type declarations, i.e.
2935 violations of the Python "declarations are assertions" principle
2936 (thanks to Alexey Dejneka)
2937 * fixed several bugs in PCL's error checking (thanks to Gerd
2939 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2941 * fixed bug in compilation of functions as first class values
2942 (thanks to Antonio Martinez-Shotton)
2943 * The compiler's handling TYPE-ERRORs which it can prove will
2944 inevitably happen at runtime has been cleaned up and corrected
2945 in several ways. (thanks to Alexey Dejneka)
2946 * improved argument type checking for various basic arithmetic
2947 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2948 lavished on them in the past that they can be compiled in many
2949 ways in different special cases
2950 * fixed bug 181: compiler checks validity of user supplied type
2952 * cleaned up code flushing in optimization: Function calls which
2953 should signal errors for safety purposes (e.g. which ANSI says
2954 should signal errors when their arguments are of incorrect type)
2955 are no longer optimized away.
2956 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2957 * incremented fasl file version number, because changes in the
2958 implementation of sequence functions like COERCE caused
2959 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2960 in old inline expansions) to become undefined. (Actually these
2961 changes were later undone, so we might very well be binary
2962 compatible with 0.7.7 after all, but leaving the version number
2963 incremented seemed like the simplest and most conservative
2966 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2967 * An alpha-quality port to the parisc architecture running Linux,
2968 based on the old CMUCL backend, has been made. This, even more so
2969 than the other backends, should be considered still a work in
2970 progress; known problems include that the Linux kernel in 64-bit
2971 mode does not propagate the correct sigcontext structure to
2972 userspace, and consequently SBCL on a parisc64 kernel will not
2974 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2975 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2976 suppressed.) Also now that the compiler is looking at declarations
2977 in the environment, it checks optimization declarations as well,
2978 and suppresses inlining when (> DEBUG SPEED).
2979 * More fixes have been made to treatment of floating point exception
2980 treatment and other Unix signals. In particular, floating point
2981 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2982 * The detection and handling of control stack exhaustion (infinite
2983 or very deeply nested recursion) has changed. Stack exhaustion
2984 detection is now done by write-protecting pages at the OS level
2985 and applies at all optimization settings; when found, a
2986 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2987 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2988 be caught using IGNORE-ERRORS.
2989 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2990 correctly when transforming recursive LABELS functions to LETs.
2991 (thanks to Alexey Dejneka)
2992 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2993 that are names of constants or global variables.
2994 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2995 alien routines with docstrings.
2996 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2997 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2999 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3000 (thanks to Eric Marsden)
3001 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3002 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3003 * Bug fix: The SPARC backend can now compile functions involving
3004 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3005 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3006 to a functions expecting a PATHNAME-DESIGNATOR.
3007 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3008 lambda lists. (thanks to Alexey Dejneka)
3009 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3010 functions so that EXPORTing the name of the class doesn't cause
3011 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3012 to Antonio Martinez for reporting this)
3013 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3014 constant-folded without failing an assertion. (thanks to Einar
3015 Floystad Dorum for reporting this)
3016 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3017 types are now dealt with more correctly by the compiler.
3018 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3019 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3020 behaviour analogous to RENAME-FILE. This puts its behaviour more
3021 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3023 * The fasl file version number has changed again. (because of the
3024 bug fix involving the names of PCL MAKE-INSTANCE functions)
3026 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3027 * bug fix: Floating point exceptions are treated much more
3028 consistently on the x86/Linux and PPC/Linux platforms.
3029 * Array initialization with :INITIAL-ELEMENT is now much faster for
3030 cases when the compiler cannot open code the array creation, but
3031 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3032 array accesses have also seen a speed increase.
3033 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3034 specified. (This was at the root of some bad interactions between
3035 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3037 * bug fix: Internal error arguments for undefined functions are now
3038 computed correctly on the PPC/Linux platform.
3039 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3040 Raymond Toy's patch for CMU CL)
3041 * Support for the Solaris 9 operating environment has been included
3042 (thanks to Daniel Merritt)
3043 * A very ugly but hopefully complete draft of the missing FFI chapter
3044 of the manual has been created by reformatting the corresponding
3045 CMU CL manual chapter into (currently very ugly and incoherent)
3046 DocBook and bringing it up to date for SBCL behavior. Thus, the
3047 manual is now essentially complete, at least by my extreme
3048 once-and-only-once standards, whereby it's acceptable to refer to
3049 the doc strings of SB-EXT functions as the primary documentation.
3050 * The fasl file version number has changed again, due to cleanup of
3051 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3052 is no longer a static symbol.)
3054 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3055 * SBCL now builds with OpenMCL (version 0.12) as the
3056 cross-compilation host; also, more progress has been made toward
3057 bootstrapping under CLISP.
3058 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3060 * bug 158 fixed: The compiler can now deal with integer loop
3061 increments different from 1; fixing this turned out also to fix
3063 * bug 169 fixed: no more bogus warnings about using lexical bindings
3064 despite the presence of perfectly good SPECIAL declarations (thanks
3065 to David Lichteblau)
3066 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3067 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3068 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3070 * bug 180 fixed: Method combination specifications no longer ignore
3071 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3072 * bug fix: Structure type predicate functions now check their argument
3073 count as they should.
3074 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3075 correctly. (thanks to Pierre Mai)
3076 * minor incompatible change: The --noprogrammer option is deprecated
3077 in favor of the new --disable-debugger option, which is very similar.
3078 (The major difference is that it takes effect at a slightly different
3079 time at startup, causing handling of errors in --sysinit and
3080 --userinit files will be affected differently.) The
3081 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3082 been added to allow this functionality to be controlled from ordinary
3083 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3084 maintainers, who might want to run non-interactive scripts to
3085 build SBCL cores which will later be used interactively.)
3086 * minor incompatible change: The LOAD function no longer, when given
3087 a wild pathname to load, loads all files matching that pathname.
3088 Instead, an error of type FILE-ERROR is signalled.
3090 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3091 * bug 147 fixed: The compiler preserves its block link/count
3092 invariants more correctly now so that it doesn't crash. (thanks
3094 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3096 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3097 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3098 no-value-for-this-character as NIL (as specified by ANSI).
3099 * HOST-NAMESTRING on physical pathnames now returns a string that is
3100 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3101 (thanks to Christophe Rhodes)
3102 * The Alpha port handles icache flushing more correctly. (thanks to
3104 * More progress has been made toward bootstrapping under CLISP. (thanks
3105 to Christophe Rhodes)
3106 * The fasl file format has changed again, because dynamic loading
3107 on OpenBSD (which has non-ELF object files) motivated some cleanups
3108 in the way that foreign symbols are transformed and passed around.
3109 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3110 is no longer treated as whitespace by the reader, but instead as
3111 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3112 |A
\7fB|, instead of A as it used to.
3114 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3115 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3116 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3117 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3118 this as of 0.7.1.45, but was left out of the previous news section
3119 (thanks to Dan Barlow)
3120 * SBCL now runs on the Solaris operating system on SPARC architectures
3121 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3122 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3123 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3124 * SPARC backend cleanups, allowing builds of cores optimized for V8
3125 and V9 SPARCS, and also emission of code targeted to a particular
3126 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3128 * SBCL is closer to bootstrapping under CLISP, thanks to various
3129 fixes by Christophe Rhodes.
3130 * The fasl file format has changed again, to allow the compiler's
3131 INFO database to support symbol macros.
3132 * The user manual (in doc/) is formatted into HTML more nicely.
3133 (thanks to coreythomas)
3134 * The system is smarter about SUBTYPEP relationships, especially
3135 those involving NOT types (including types such as ATOM which are
3136 represented internally using NOT types). Thus SUBTYPEP is less
3137 likely to return (VALUES NIL NIL) in general, and in particular
3138 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3140 * The fasl file format has changed again, because the internal
3141 representation of types now includes a new slot to support the new
3142 SUBTYPEP-of-NOT-types logic.
3143 * (not a change in the main branch of SBCL, but a related prototype
3144 which can hopefully be merged into the main branch of SBCL in the
3145 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3146 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3147 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3148 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3149 dispatch macro character combinations. (thanks to Alexey Dejneka)
3150 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3151 that superficially look like logical namestrings correctly.
3152 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3154 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3155 default supplied-p) for &optional and &key arguments. (thanks to
3158 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3159 * incompatible change: The compiler is now less aggressive about
3160 tail call optimization, doing it only when (> SPACE DEBUG) or
3161 (> SPEED DEBUG). (This is an incompatible change because there are
3162 programs which relied on the old CMU-CL-style behavior to optimize
3163 away their unbounded recursion which will now die of stack overflow.)
3164 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3165 for non-GENCGC systems has been increased to 20M (since that
3166 seems much closer to the likely performance optimum for modern
3167 systems than the old 4M value was)
3168 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3169 and *DEBUG-PRINT-LEVEL*
3170 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3171 of CMU CL's support for SPARC, and various endianness and other
3172 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3173 * new syntactic sugar for the Unix command line: --load foo.bar is now
3174 an alternate notation for --eval '(load "foo.bar")'.
3176 ** The system now detects stack overflow and handles it gracefully,
3177 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3178 optimization settings. (This is a good thing in general, and
3179 its introduction in this version should be particularly timely
3180 for anyone whose code fails because of suppression of tail
3182 ** The system now hunts for the C variable "environ" in a more
3183 devious way, to avoid segfaults when the C library version differs
3184 between compile time and run time. (thanks to Christophe Rhodes)
3185 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3186 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3187 ** The compiler no longer issues bogus style warnings for undefined
3188 classes in the same source file as the DEFCLASSes which defined
3189 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3191 ** fixes in CONDITION class precedence list for undefined function
3192 errors (thanks to Alexei Dejneka)
3193 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3194 correctly. (thanks to Dan Barlow)
3195 ** portability fixes aiming at bootstrapping under CLISP (thanks
3196 to Dave McDonald and Christophe Rhodes)
3197 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3198 ** fixes in type translation and and type inference (thanks to
3200 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3201 ** various fixes in the new ports (thanks to Dan Barlow)
3202 * several changes related to debugging:
3203 ** suppression of tail recursion, as noted above
3204 ** stack overflow detection, as noted above
3205 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3206 is now the default. (For some time encapsulation has been more
3207 reliable than the breakpoint-based :ENCAPSULATE NIL
3208 implementation, at least on X86 systems; and I just noticed that
3209 encapsulation also seems closer to the spirit of the ANSI
3212 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3214 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3215 up properly again. (There was a packaging bug in 0.7.0 which
3216 left their definitions in SB-SYS::LOAD-FOREIGN and
3217 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3218 vital for most things which interface to C-level interfaces,
3219 like extensions working with sockets or databases or
3220 Perl-compatible regexes or whatever, and the need to fix
3221 this bug is the main reason that 0.7.1 was released so
3223 ** DEFGENERIC is now choosier about the methods it redefines, so that
3224 reLOADing a previously-LOADed file containing DEFGENERICs does
3225 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3226 cycle now works as it should. (thanks to Alexey Dejneka)
3227 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3228 fixed by Christophe Rhodes. (He actually submitted this patch
3229 months ago, and I delayed until after 0.7.0.)
3230 ** Bug 111 (internal compiler confusion about runtime checks on
3231 FUNCTION types) was fixed by Alexey Dejneka.
3232 * Some internal cleanups (getting rid of variables which aren't
3233 needed now that the byte interpreter is gone) caused the fasl
3234 file format number to change again.
3236 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3237 * major incompatible change: The default fasl file extension, i.e. the
3238 default extension for files produced by COMPILE-FILE, has changed
3239 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3241 ** There are many changes in the implementation of the compiler.
3242 SBCL is now essentially a compiler-only implementation of ANSI
3243 Common Lisp. EVAL still "interprets" a few special cases, but
3244 almost all the interesting cases are handled by creating
3245 a LAMBDA expression, calling COMPILE on it, then calling
3246 FUNCALL on the result.
3247 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3248 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3249 Since the code is newer, there might still be some new bugs
3250 (though not as many as before Martin Atzmueller's fixes:-). But
3251 the new code is substantially simpler and clearer, and hopefully
3252 any remaining bugs will be simpler, less fundamental, and more
3253 fixable then the bugs in the old code.
3254 ** The revised compiler is still a little unsteady on its feet.
3256 *** The debugging information it produces (particularly the names
3257 of FUNCTION objects) is sometimes much less useful than what
3258 the old compiler produced.
3259 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3260 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3261 within a MACROLET) has been temporarily weakened.
3262 ** There are new compiler optimizations for various functions:
3263 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3264 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3265 *** the math functions TRUNCATE, FLOOR, and CEILING
3266 *** the function-of-all-trades COERCE
3267 Mostly these should be transparent, but there's one
3268 potentially-annoying problem (bug 117): when the compiler
3269 inline-expands a function and does type analysis on the result,
3270 it can create control paths which have type mismatches, and
3271 when it can't prove that those control paths aren't taken,
3272 it will issue WARNINGs about the type mismatches. This is
3273 a particular problem in practice for the new sequence functions.
3274 It's not clear how this should be fixed, and for now, a
3275 workaround is given in the entry for 117 in the BUGS file.
3276 ** (Because of the interaction between the two previous items --
3277 occasional inlining problems and new inline expansions -- some
3278 of the new sequence function optimizations won't really kick in
3279 completely until debugging information, and then inlining, are
3280 straightened out in some future version.)
3281 * minor incompatible changes:
3282 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3283 DIRECTORY no longer implicitly promotes NIL slots of its
3284 pathname argument to :WILD. In particular, when you ask for the
3285 contents of a directory (which you used to be able to do without
3286 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3287 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3288 ** changes in behavior that ANSI explicitly defines to be
3289 implementation dependent:
3290 *** The new compiler-only implementation still conforms with ANSI,
3291 but acts a little different than before. Besides the obvious
3292 changes in performance tradeoffs (that the cost per form passed
3293 to EVAL has gone up, and the cost per form executed by EVAL
3294 has gone down), the behavior of the system changes a little
3295 because there are no longer any interpreted function objects.
3296 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3297 e.g. doing COMPILE on the output of interactive DEFUN is
3299 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3301 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3302 has changed from (:CL) to NIL.
3303 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3304 e.g. #\Newline and #\Tab, don't have names specified in the
3305 ANSI Common Lisp standard, is now based on their ASCII symbolic
3306 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3307 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3308 are no longer used for output.
3309 ** changes in internal implementation constants:
3310 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3311 4 million. (If your application spends a lot of time GCing and
3312 you have a lot of RAM, you might want to experiment with
3313 increasing it even more.)
3314 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3315 However, almost all old code should still continue to work without
3316 immediate update, as SB-C-CALL is now a (deprecated) nickname
3318 ** Old operator names in the style DEF-FOO are now deprecated in
3319 favor of new corresponding names DEFINE-FOO, for consistency with
3320 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3321 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3322 internal symbols, but a few supported extensions like
3323 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3324 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3325 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3326 etc. as you get deeper into recursive calls to the debugger
3327 command loop, instead of the old "5]", "5]]", "5]]]"
3328 sequence. (I was motivated to do this when squabbles between
3329 ILISP and SBCL left me very deeply nested in the debugger. In the
3330 short term, this change will probably provoke more ILISP/SBCL
3331 squabbles, but hopefully it will be an improvement in the long run.)
3332 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3333 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3334 function, since declaiming return types involving aliens is
3335 (1) annoyingly messy to do by hand and (2) vital to efficient
3336 compilation of code which calls such functions.
3337 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3338 longer reexported by the SB-EXT package. They're solely useful
3339 for alien code, so it seems more logical that you should get
3340 them from the SB-ALIEN package, not in SB-EXT.
3341 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3342 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3343 features. Instead, the code that they used to control is always
3344 built into the system.
3345 * many other bug fixes
3346 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3347 advantage of the new EVAL-WHEN stuff and to clean them up in
3348 general, and they are now more ANSI-compliant in a number of
3349 ways. Martin Atzmueller is responsible for a lot of this.
3350 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3352 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3353 *** correct ERROR type for various file operations
3354 *** some fixes for Lisp streams
3355 *** DEFMETHOD syntax checking
3356 *** changing old weird representation of debug information as
3357 strings (which, among their other deficiencies, don't transform
3358 correctly when you rename packages, and don't change their
3359 print representation when you change things like *PACKAGE*
3360 and *PRINT-LENGTH*) to symbols and lists of symbols
3361 He also made several improvements and fixed several bugs in DESCRIBE.
3362 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3364 *** misbehavior of WRITE-STRING/WRITE-LINE
3365 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3366 and several other LOOP problems as well
3367 *** DIRECTORY when similar filenames are present
3368 *** DEFGENERIC with :METHOD options
3369 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3370 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3371 *** argument ordering in FIND with :TEST option
3372 *** mishandled package designator argument in APROPOS-LIST
3373 *** various problems in the backquote readmacro
3374 *** a bug in APROPOS
3375 *** probably some others that I'm not describing very well here,
3376 since the CVS log documents them by reference to sbcl-devel
3377 messages, and the SourceForge archives aren't working well.:-(
3378 ** Dan Barlow improved the Alpha port (and is making progress on the
3379 PPC port, for those of you who think different).
3380 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3381 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3382 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3383 continues to work on the SPARC port (for those of you in a position
3384 to look down upon our little PC-compatible boxes from a great height).
3385 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3386 and uses that copy, rather than the current dynamic value, when
3387 it's trying to decide whether to truncate output. Thus e.g.
3388 (let ((*print-lines* 50))
3389 (pprint-logical-block (stream nil)
3391 (let ((*print-lines* 8))
3392 (print (aref possiblybigthings i) stream)))))
3393 should now truncate the logical block only at 50 lines, instead of
3394 often truncating it at 8 lines, as it did before.
3395 * The doc/cmucl/ directory, containing old CMU CL documentation
3396 from the time of the fork, is no longer part of the base system.
3397 SourceForge has shut down its anonymous FTP service, and with it
3398 my original plan for distributing the old CMU CL documentation
3399 there. For now, if you need these files you can download an old
3400 SBCL source release and extract them from it.
3401 * The fasl file version number changed again, for dozens of reasons,
3402 some of which are apparent above.
3404 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3405 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3406 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3407 patch, so that DISASSEMBLE output is much nicer.
3408 * The code in the SB-PROFILE package now seems reasonably stable.
3409 I still haven't decided what the final interface should look like
3410 (I'd like PROFILE to interact cleanly with TRACE, since both
3411 facilities use function encapsulation) but if you have a need
3412 for profiling now, you can probably use it successfully with
3413 the current CMU-CL-style interface.
3414 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3415 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3416 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3417 link now returns the dangling link itself, and for similar
3418 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3419 link itself. (In these cases the old code signalled an error and
3420 looped endlessly, respectively.) Thus, DIRECTORY now works even
3421 in the presence of dangling and cyclic symbolic links.
3422 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3423 is now a supported extension again, since the consensus on
3424 sbcl-devel was that it can be useful for ordinary development
3425 work, not just for debugging SBCL itself.
3426 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3427 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3428 that currently-defined functions might be redefined later with
3429 different return types.
3430 * Hash tables can be printed readably, as inspired by CMU CL code
3431 of Eric Marsden and SBCL code of Martin Atzmueller.
3432 * better error handling in CLOS method combination, thanks to
3433 Martin Atzmueller porting Pierre Mai's CMU CL patches
3434 * more overflow fixes for >16Mbyte I/O buffers
3435 * A bug in READ has been fixed, so that now a single Ctrl-D
3436 character suffices to cause end-of-file on character streams.
3437 In particular, now you only need one Ctrl-D at the command
3438 line (not two) to exit SBCL.
3439 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3441 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3442 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3443 does the right thing.
3444 * The implementation of some type tests, especially for CONDITION
3445 types, is now tidier and maybe faster, due to CMU CL code
3446 originally by Douglas Crosher, ported by Martin Atzmueller.
3447 * Some math functions have been fixed, and there are new
3448 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3449 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3450 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3451 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3452 likely to be built into sbcl-0.7.0.)
3453 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3454 CL, ported to SBCL by Martin Atzmueller)
3455 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3456 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3457 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3458 * There's a new slam.sh hack to shorten the edit/compile/debug
3459 cycle for low-level changes to SBCL itself, and a new
3460 :SB-AFTER-XC-CORE target feature to control the generation of
3461 the after-xc.core file needed by slam.sh.
3462 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3463 COMPILE-FILE is no longer supported, so that now every function
3464 gets an entry point, so that block compilation looks a little
3465 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3466 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3467 returns the number of bytes consed since the system started,
3468 rather than the number consed since the first time the function
3469 was called. (The new definition parallels ANSI functions like
3470 CL:GET-INTERNAL-RUN-TIME.)
3471 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3472 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3473 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3474 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3475 * Fasl file version numbers are now independent of the target CPU,
3476 since historically most system changes which required version
3477 number changes have affected all CPUs equally. Similarly,
3478 the byte fasl file version is now equal to the ordinary
3481 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3482 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3483 is no longer recognized. I apologize for this, because it was
3484 listed in SB-EXT as a supported extension, but I found that
3485 its existing behavior was poorly specified, as well as incorrectly
3486 specified, and it looked like too much of a mess to straighten it
3487 out. I have enough on my hands trying to get ANSI stuff to work..
3488 * many patches ported from CMU CL by Martin Atzmueller, with
3489 half a dozen bug fixes in pretty-printing and the debugger, and
3490 half a dozen others elsewhere
3491 * fixed bug 13: Floating point infinities are now supported again.
3492 They might still be a little bit flaky, but thanks to bug reports
3493 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3494 as flaky as they were.
3495 * The --noprogrammer command line option is now supported. (Its
3496 behavior is slightly different in detail from what the old man
3497 page claimed it would do, but it's still appropriate under the
3498 same circumstances that the man page talks about.)
3499 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3500 are now supported, and enabled by default. Thus, the compiler can
3501 handle many floating point and complex operations much less
3502 inefficiently. (Thus e.g. you can implement a complex FFT
3504 * The compiler now detects type mismatches between DECLAIM FTYPE
3505 and DEFUN better, and implements CHECK-TYPE more correctly, and
3506 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3507 thanks to patches from Martin Atzmueller.
3508 * various fixes to make the cross-compiler more portable to
3509 ANSI-conforming-but-different cross-compilation hosts (notably
3510 Lispworks for Windows, following bug reports from Arthur Lemmens)
3511 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3512 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3513 fixed thanks to Pierre Mai's CMU CL patches.
3514 * fixes to keep the system from overflowing internal counters when
3515 it tries to use i/o buffers larger than 16M bytes
3516 * fixed bug 45a: Various internal functions required to support
3517 complex special functions have been merged from CMU CL sources.
3518 (When I was first setting up SBCL, I misunderstood a compile-time
3519 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3520 * improved support for type intersection and union, fixing bug 12
3521 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3522 more obscure bugs as well
3523 * some steps toward byte-compiling non-performance-critical
3524 parts of the system, courtesy of patches from Martin Atzmueller
3525 * Christophe Rhodes has made some debian packages of sbcl at
3526 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3527 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3528 stable, but are nonetheless usable. When he's ready, I'd be happy
3529 to add them to the SourceForge "File Releases" section. (And if
3530 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3531 * new fasl file format version number (because of changes in
3532 internal representation of (OR ..) types to accommodate the new
3533 support for (AND ..) types, among other things)
3535 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3536 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3538 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3540 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3541 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3542 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3543 * DESCRIBE now works on structure objects again.
3544 * Most function call argument type mismatches are now handled as
3545 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3546 know whether the function will be redefined before the call is
3547 executed. (The compiler could flag local calls with full WARNINGs,
3548 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3549 it doesn't keep track of enough information to know whether calls
3550 are local in this sense.)
3551 * Compiler output is now more verbose, with messages truncated
3552 later than before. (There should be some supported way for users
3553 to override the default verbosity, but I haven't decided how to
3554 provide it yet, so this behavior is still controlled by the internal
3555 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3556 src/compiler/ir1util.lisp.)
3557 * Fasl file format version numbers have increased again, because
3558 support for the Gray streams extension changes the layout of the
3559 system's STREAM objects.
3560 * The Gray subclassable streams extension now works, thanks to a
3561 patch from Martin Atzmueller.
3562 * The full LOAD-FOREIGN extension (not just the primitive
3563 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3564 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3565 but like most other programs, it defaults to copying the Unix
3566 environment from the original process instead of starting the
3567 new process in an empty environment.
3568 * Extensions which manipulate the Unix environment now support
3569 an :ENVIRONMENT keyword option which doesn't smash case or
3570 do other bad things. The CMU-CL-style :ENV option is retained
3571 for porting convenience.
3572 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3573 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3575 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3577 * A patch from Martin Atzmueller seems to have solved the SIGINT
3578 problem, and as far as we know, signal-handling now works cleanly.
3579 (If you find any new bugs, please report them!)
3580 * The system no longer defaults Lisp source file names to types
3581 ".l", ".cl", or ".lsp", but only to ".lisp".
3582 * The compiler no longer uses special default file extensions for
3583 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3584 seems to expect a single default extension for all compiled code,
3585 and there's no compelling reason to try to stretch the standard
3586 to allow two different extensions.) Instead, byte-compiled files
3587 default to the same extension as native-compiled files.
3588 * Fasl file format version numbers have increased again, because
3589 a rearrangement of internal implementation packages made some
3590 dumped symbols in old fasl files unreadable in new cores.
3591 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3592 many fewer weird special cases.
3593 * Bug #17 (differing COMPILE-FILE behavior between logical and
3594 physical pathnames) has been fixed, and some related misbehavior too,
3595 thanks to a patch from Martin Atzmueller.
3596 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3597 by Tim Moore, ported to SBCL by Martin Atzmueller.
3598 * Martin Atzmueller fixed several filesystem-related problems,
3599 including bug #36, in part by porting CMU CL patches, which were
3600 written in part by Paul Werkowski.
3601 * More compiler warnings in src/runtime/ are gone, thanks to
3602 more patches from Martin Atzmueller.
3603 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3606 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3608 * DESCRIBE now works on CONDITION objects.
3609 * The debugger now handles errors which arise when trying to print
3610 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3612 * The build system now uses an additional file, customize-target-features.lisp,
3613 to allow local modifications to the target *FEATURES* list. (The point of
3614 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3615 debugging features enabled, without having to worry about propagating it
3616 into everyone's system when I do a "cvs update".) When no
3617 customize-target-features.lisp file exists, the target *FEATURES* list
3618 should be constructed the same way as before.
3619 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3620 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3621 warn about behavior which is undefined under the ANSI spec. Note: This
3622 is specified by ANSI, but it's not very popular with programmers.
3623 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3624 macro in the SBCL sources for an example of a workaround which you
3625 might use to make portable ANSI-standard code which does what you want.
3626 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3627 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3628 when it's used as a non-toplevel form. (This is required in order
3629 to implement the DEFCONSTANT-EQX macro.)
3630 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3631 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3632 funny things, due to bugs in the implementation of EVAL-WHEN
3633 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3634 non-toplevel DEFCONSTANTs are uncommon.)
3635 * The core file version number and fasl file version number have been
3636 incremented, because the old noncompliant DEFCONSTANT behavior involved
3637 calling functions which no longer exist, and because I also took the
3638 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3639 * fixed bug 1 (error handling before read-eval-print loop starts), and
3640 redid debugger restarts and related debugger commands somewhat while
3642 ** The QUIT debugger command is gone, since it did something
3643 rather different than the SB-EXT:QUIT command, and since it never
3644 worked properly outside the main toplevel read/eval/print loop.
3645 Invoking the new TOPLEVEL restart provides the same functionality.
3646 ** The GO debugger command is also gone, since you can just invoke
3647 the CONTINUE restart directly instead.
3648 ** The TOP debugger command is also gone, since it's redundant with the
3649 FRAME 0 command, and since it interfered with abbreviations for the
3651 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3653 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3654 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3655 some porting work by Martin Atzmueller)
3656 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3657 with a boatload of patches which he ported to SBCL. Now that those
3659 ** The system tries to make sure that its low-priority messages
3660 are prefixed by semicolons, to help people who like to use
3661 syntax highlighting in their ILISP buffer. (This patch
3662 was originally due to Raymond Toy.)
3663 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3664 patches originally written by Raymond Toy.
3665 ** The compiler understands coercion between single-value and
3666 multiple-VALUES type expressions better, getting rid of some very
3667 weird behavior, thanks to patches originally by Robert MacLachlan
3668 and Douglas Crosher.
3669 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3670 lambda lists, thanks to a patch originally by Pierre Mai.
3671 ** The system no longer bogusly warns about "abbreviated type
3673 ** The compiler gets less confused by inlining and RETURN-FROM,
3674 thanks to some patches originally by Tim Moore.
3675 ** The system no longer hangs when dumping circular lists to fasl
3676 files, thanks to a patch originally from Douglas Crosher.
3677 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3678 undefined function error.
3679 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3680 in structure slot initforms) for now, documented workaround instead:-|
3681 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3682 * fixed bug 5: added stubs for various Gray stream functions called
3683 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3684 installed, at least appropriate type errors are generated
3685 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3686 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3687 consistently in DEFMETHOD forms.
3688 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3689 it doesn't seem to affect SBCL after all
3690 * The C runtime system now builds with better optimization and many
3691 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3693 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3695 * The system is now under CVS at SourceForge (instead of the
3696 CVS repository on my home machine).
3697 * The new signal handling code has been tweaked to treat register
3698 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3699 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3700 downstream, e.g. in debug-int.lisp, has implicit dependencies
3701 on the unsignedness of integer representation of machine words,
3702 and that caused the system to bomb out with infinite regress
3703 when trying to recover from type errors involving signed values,
3704 e.g. (BUTLAST '(1 2 3) -1).)
3705 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3706 (This was one of the bugs Peter Van Eynde reported back in July.)
3707 * The system now uses code inspired by Colin Walters' O(N)
3708 implementation of MAP (from the cmucl-imp@cons.org mailing
3709 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3710 inline the MAP operation, and there is more than one
3711 sequence argument to the MAP call (so that it can't just
3712 do ETYPECASE once and for all based on the type of the
3713 single sequence argument). (The old non-inline implementation
3714 of the general M-argument sequence-of-length-N case required
3715 O(M*N*N) time when any of the sequence arguments were LISTs.)
3716 * The QUIT :UNIX-CODE keyword argument has been renamed to
3717 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3718 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3720 * Raymond Wiker's patches to port dynamic loading from Linux to
3721 FreeBSD have been added.
3722 * The BUGS file is now more nearly up to date, thanks in large part
3723 to Martin Atzmueller's review of it.
3724 * The debugger now flushes standard output streams before it begins
3725 its output ("debugger invoked" and so forth).
3726 * The core version number and fasl file version number have both
3727 been incremented, because of incompatible changes in the layout
3729 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3730 used to not be called for a saved Lisp image.)
3731 * Martin Atzmueller's version of a patch to fix a compiler crash,
3732 as posted on sbcl-devel 13 September 2000, has been installed.
3733 * Instead of installing Martin Atzmueller's patch for the
3734 compiler transform for SUBSEQ, I deleted the compiler transform,
3735 and transforms for some similar consing operations.
3736 * A bug in signal handling which kept TRACE from working on OpenBSD
3738 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3739 away by constant folding
3740 * The system now defines its address space constants in one place
3741 (in the Lisp sources), and propagates them automatically elsewhere
3742 (through GENESIS and the sbcl.h file). Therefore, patching the
3743 address map is less unnecessarily tedious and error-prone. The
3744 Lisp names of address space constants have also been systematized.
3745 * CVS tags like dollar-Header-dollar have been removed from
3746 the sources, because they have never saved me trouble and
3747 they've been source of trouble working with patches and other
3748 diff-related operations.
3749 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3750 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3752 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3754 * The system has been ported to OpenBSD.
3755 * The system now compiles with a simple "sh make.sh" on the systems
3756 that it's supported on. I.e., now you no longer need to tweak
3757 text in the target-features.lisp-expr and symlinks in src/runtime/
3758 by hand, the make.sh takes care of it for you.
3759 * The system is no longer so grossly inefficient when compiling code
3760 involving vectors implemented as general (not simple) vectors (VECTOR T),
3761 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3762 which dares to use the various sequence functions on non-simple
3763 vectors, takes less of a performance hit.
3764 * There is now a primitive type predicate VECTOR-T-P
3765 to test for the (VECTOR T) type, so that e.g.
3766 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3767 can now be compiled with some semblance of efficiency. (The old code
3768 turned the type declaration into a full call to %TYPEP at runtime!)
3769 * AREF on (VECTOR T) is still not fast, since it's still compiled
3770 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3771 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3772 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3773 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3774 without making full calls to SUBTYPEP at runtime.
3775 (Some analogous efficiency issues for non-simple vectors specialized to
3776 element types other than T, or for non-simple multidimensional arrays,
3777 have not been addressed. They could almost certainly be handled the
3778 same way if anyone is motivated to do so.)
3779 * The changes in array handling break binary compatibility, so
3780 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3781 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3782 instead of (VALUES T T).
3783 * By following the instructions that Dan Barlow posted to sbcl-devel
3784 on 2 July 2000, I was able to enable primitive dynamic object
3785 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3786 functionality is not implemented (since it calls ld to resolve
3787 library references automatically, requiring RUN-PROGRAM for its
3788 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3789 try to resolve library references) is now supported.
3790 * The system now flushes the standard output streams when it terminates,
3791 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3792 them at several other probably-convenient times, e.g. in each pass of
3793 the toplevel read-eval-print loop, and after evaluating a form given
3794 as an "--eval" command-line option. (These changes were motivated by a
3795 discussion of stream flushing issues on cmucl-imp in August 2000.)
3796 * The source transform for TYPEP of array types no longer assumes
3797 that an array whose element type is a not-yet-defined type
3798 is implemented as an array of T, but instead punts, so that the
3799 type will be interpreted at runtime.
3800 * There is now some support for cross-compiling in make.sh: each of
3801 the phases of make.sh has its own script. (This should be transparent
3802 to people doing ordinary, non-cross-compile builds.)
3803 * Since my laptop doesn't have hundreds of megabytes of memory like
3804 my desktop machine, I became more motivated to do some items on
3805 my to-do list in order to reduce the size of the system a little:
3806 ** Arrange for various needed-only-at-cold-init things to be
3807 uninterned after cold init. To support this, those things have
3808 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3809 symbols with such names are now uninterned after cold init).
3810 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3811 fasl files for cold load.
3812 ** Remove the old compiler structure pooling code (which used to
3813 be conditional on the target feature :SB-ALLOC) completely.
3814 ** Redo the representation of some data in cold init to be more compact.
3815 (I also looked into supporting byte compiled code at bootstrap time,
3816 which would probably reduce the size of the system a lot, but that
3817 looked too complicated, so I punted for now.)
3818 * The maximum signal nesting depth in the src/runtime/ support code has
3819 been reduced from 4096 to 256. (I don't know any reason for the very
3820 large old value. If the new smaller value turns out to break something,
3821 I'll probably just bump it back up.)
3822 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3824 * Many, many bugs reported by Peter Van Eynde have been added to
3825 the BUGS list; some have even been fixed.
3826 * While enabling dynamic object file loading, I tried to make the
3827 code easier to understand, renaming various functions and variables
3828 with less ambiguous names, and changing some function calling
3829 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3830 * While trying to figure out how to do the OpenBSD port, I tried to
3831 clean up some of the code in src/runtime/. In particular, I dropped
3832 support for non-POSIX signal handling, added various comments,
3833 tweaked the code to reduce the number of compilation warnings, and
3834 renamed some files to increase consistency.
3835 * To support the new automatic configuration functionality in make.sh,
3836 the source file target-features.lisp-expr has been replaced with the
3837 source file base-target-features.lisp-expr and the machine-generated
3838 file local-target-features.lisp-expr.
3839 * fixed a stupid quoting error in make.sh so that using CMU CL
3840 "lisp -batch" as cross-compilation host works again
3842 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3844 * DESCRIBE no longer tries to call itself recursively to describe
3845 bound/fbound values, so that it no longer fails on symbols which are
3846 bound to themselves (like keywords, T, and NIL).
3847 * DESCRIBE now works on generic functions.
3848 * The printer now prints less-screwed-up representations of closures
3849 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3850 * A private symbol is used instead of the :EMPTY keyword previously
3851 used to mark empty slots in hash tables. Thus
3852 (DEFVAR *HT* (MAKE-HASH-TABLE))
3853 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3854 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3855 now does what ANSI says that it should. (You can still get
3856 similar noncompliant behavior if bang on the hash table
3857 implementation with all the symbols you get back from
3858 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3859 This breaks binary compatibility, since tests for equality to
3860 :EMPTY are wired into things like the macroexpansion of
3861 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3863 * There's now a minimal placeholder implementation for CL:STEP,
3864 as required by ANSI.
3865 * An obscure bug in the interaction of the normal compiler, the byte
3866 compiler, inlining, and structure predicates has been patched
3867 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3868 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3870 * Missing ordinary arguments in a macro call are now detected even
3871 when the macro lambda list contains &KEY or &REST.
3872 * The debugger no longer complains about encountering the top of the
3873 stack when you type "FRAME 0" to explicitly instruct it to go to
3874 the top of the stack. And it now prints the frame you request even
3875 if it's the current frame (instead of saying "You are here.").
3876 * As specified by ANSI, the system now always prints keywords
3877 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3879 * The default initial SIZE of HASH-TABLEs is now smaller.
3880 * Type information from CLOS class dispatch is now propagated
3881 into DEFMETHOD bodies, so that e.g.
3882 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3884 is now basically equivalent to
3885 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3886 (DECLARE (TYPE SINGLE-FLOAT X))
3888 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3889 operation, without having to do run-time type dispatch.
3890 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3891 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3892 * Since I don't seem to be making big file reorganizations very often
3893 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3894 is overflowing my ability to conveniently back them up), I've finally
3895 checked the system into CVS. (The CVS repository is on my home system,
3896 not at SourceForge -- putting it on SourceForge might come later.)
3897 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3898 high-level GC-NOTIFY-FOO functions send their output. (There's
3899 still very little control of where low-level verbose GC functions
3900 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3901 less than it used to -- the GC-NOTIFY-FOO functions are now under
3902 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3903 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3904 in only one place in the source code, and propagates it automatically
3905 everywhere that it's needed. Thus e.g. when I bump the version from
3906 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3907 * The C source files now include boilerplate legalese and documentation
3908 at the head of each file (just as the Lisp source files already did).
3909 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3910 to his page will be replaced with a link to his new CLiki service.
3912 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3914 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3915 * The build process now looks for GNU make under the default name "gmake",
3916 instead of "make" as it used to. If GNU make is not available as "gmake"
3917 on your system, you can change this default behavior by setting the
3918 GNUMAKE environment variable.
3919 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3920 can build without error under CMU CL.
3922 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3924 * There is now a partial SBCL user manual (with some new text and some
3925 text cribbed from the CMU CL manual).
3926 * The beginnings of a profiler have been added (starting with the
3927 CMU CL profiler and simplifying and cleaning up). Eventually the
3928 main interface should be through the TRACE macro, but for now,
3929 it's still accessed through vaguely CMU-CL-style functions and macros
3930 exported from the package SB-PROFILE.
3931 * Some problems left over from porting CMU CL to the new
3932 cross-compilation bootstrap process have been cleaned up:
3933 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3934 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3935 part when getting rid of the compiler *BACKEND* stuff.)
3936 ** The value of *NULL-TYPE* was screwed up, because it was
3937 being initialized before the type system knew the final
3938 definition of the 'NULL type. This screwed up several key
3939 optimizations in the compiler, causing inefficiency in all sorts
3940 of places. (I found it because I wanted to understand why
3941 GET-INTERNAL-RUN-TIME was consing.)
3942 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3943 PROCLAIM FTYPE information. Unfortunately this broke binary
3944 compatibility again, since now the forms output by DEFGENERIC
3945 to refer to functions which didn't exist in 0.6.3.
3946 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3947 can use the new (as of 0.6.3) transform for SOME into MAP into
3949 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3950 values to (INTEGER 0 1000000), so that the time code will no longer
3951 occasionally get blown up by Linux returning 1000000 microseconds
3952 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3953 its output conform to the ANSI spec. (Alas, this makes its output
3954 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3955 * A full call to MAP NIL with a single sequence argument no longer conses.
3956 * fixes to problems pointed out by Martin Atzmueller:
3957 * The manual page no longer talks about multiprocessing as though
3958 it were currently supported.
3959 * The ILISP support patches have been removed from the distribution,
3960 because as of version 5.10.1, ILISP now supports SBCL without us
3961 having to maintain patches.
3962 * added a modified version of Raymond Toy's recent CMU CL patch for
3963 EQUALP comparison of HASH-TABLE
3965 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3967 * The system still can't cross-compile itself with
3968 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3969 entails), but at least it can compile more of itself that way
3970 than it used to be able to, and various buglets which were uncovered
3971 by trying to cross-compile itself that way have now been fixed.
3972 * This release breaks binary compatibility again. This time
3973 at least I've incremented the FASL file format version to 2, so that the
3974 problem can be detected reliably instead of just causing weird errors.
3975 * various new style warnings:
3976 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3977 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3978 ** using the lexical binding of a variable named in the *FOO* style
3979 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3980 as specified by ANSI.
3981 * *RANDOM-STATE* is no longer automatically initialized from
3982 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3983 default behavior of the system is to repeat its behavior every time
3984 it's run. If you'd like to change this behavior, you can always
3985 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3986 old convention there was no comparably easy way to get the system to
3987 repeat its behavior every time it was run.
3988 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3989 FTYPE declarations has been removed, in favor of their ANSI
3990 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3991 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3992 the types of their sequence arguments can be determined at compile time.
3993 This is done through a new open code expansion for MAP which eliminates
3994 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3995 argument types can be determined at compile time.
3996 * The optimizer now transforms COERCE into an identity operation when it
3997 can prove that the coerced object is already of the correct type. (This
3998 can be a win for machine generated code, including the output of other
3999 optimization transforms, such as the MAP transform above.)
4000 * Credit information has been moved from source file headers into CREDITS.
4001 * Source file headers have been made more standard.
4002 * The CASE macro now compiles without complaining even when it has
4005 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4007 * (Note that the way that the PCL macroexpansions were rewritten
4008 to accommodate the change in DEFGENERIC below breaks binary
4009 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4010 not run under sbcl-0.6.2. Once we get out of alpha releases,
4011 i.e. hit release 1.0.0, we'll probably try to maintain binary
4012 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4013 and sbcl-1.4.4. Until then, however, it might be fairly common
4014 for maintenance releases to break binary compatibility.)
4015 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4017 * The reporting of unhandled signals has been changed to print some
4018 explanatory text as well as the report form. (Previously only
4019 the report form was printed.)
4020 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4021 it defines, so that the compiler no longer issues undefined function
4022 warnings for compiled-but-not-yet-loaded generic functions.
4023 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4024 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4025 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4026 section of the ANSI spec are supported.
4027 * The "" nickname for the "KEYWORD" package has been removed.
4028 The reader still handles symbol tokens which begin with a package marker
4029 as keywords, but it doesn't expose its mechanism for doing so in the
4030 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4031 * The system now issues STYLE-WARNINGs for contradictory TYPE
4032 proclamations. (Warnings for contradictory FTYPE proclamations would
4033 be nice too, but those can't be done usefully unless the type system
4034 is made smarter about FUNCTION types.)
4035 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4036 systematized, so that "*target-*.lisp is supposed to exist only on the
4037 target and imply that there's a related file which exists on the
4038 host, and *host-*.lisp is supposed to exist only on the host and imply
4039 that there's a related file which exists on the target. This involves a
4040 lot of renaming. Hopefully the acute confusion caused by the renaming
4041 will be justified by the reduction in chronic confusion..
4042 ** runtime-type.lisp -> early-target-type.lisp
4043 ** target-type.lisp -> late-target-type.lisp
4044 ** early-host-format.lisp -> early-format.lisp
4045 ** late-host-format.lisp -> late-format.lisp
4046 ** host-error.lisp -> misc-error.lisp
4047 ** early-error.lisp -> early-target-error.lisp
4048 ** late-error.lisp -> late-target-error.lisp
4049 ** host-defboot.lisp -> early-defboot.lisp
4050 ** code/misc.lisp -> code/target-misc.lisp
4051 ** code/host-misc.lisp -> code/misc.lisp
4052 ** code/numbers.lisp -> code/target-numbers.lisp
4053 ** code/early-numbers.lisp -> numbers.lisp
4054 ** early-host-type.lisp -> early-type.lisp
4055 ** late-host-type.lisp -> late-type.lisp
4056 ** host-typep.lisp -> typep.lisp
4057 ** load.lisp -> target-load.lisp
4058 ** host-load.lisp -> load.lisp
4059 ** host-disassem.lisp -> disassem.lisp
4060 ** host-insts.lisp -> insts.lisp
4061 ** byte-comp.lisp -> target-byte-comp.lisp
4062 ** host-byte-comp.lisp -> byte-comp.lisp
4063 ** host-signal.lisp -> signal.lisp
4064 ** host-defstruct.lisp -> defstruct.lisp
4065 ** late-target-type.lisp -> deftypes-for-target.lisp
4066 Furthermore, several other previously target-only files foo.lisp (e.g.
4067 hash-table.lisp and random.lisp) have been split into a target-and-host
4068 foo.lisp file and a target-only target-foo.lisp file, with their key type
4069 definitions in the target-and-host part, so that the cross-compiler will
4070 know more about target types.
4071 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4072 gone away. In their place are various *BACKEND-FOO* variables
4073 corresponding to the slots of the old structure.
4074 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4075 into the target SBCL has been fixed.
4076 * The chill.lisp system for loading cold code into a running SBCL
4078 * Support for the CMU CL "scavenger hook" extension has been removed.
4079 (It was undocumented and unused in the CMU CL sources that SBCL was
4080 derived from, and stale in sbcl-0.6.1.)
4081 * Various errors in the cross-compiler type system were detected
4082 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4083 (enabling various consistency checks). Many of them were fixed,
4084 but some hard problems remain, so the compiler is back to
4085 running without *TYPE-SYSTEM-INITIALIZED* for now.
4086 * As part of the cross-compiler type system cleanup, I implemented
4087 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4088 * I have started adding UNCROSS calls throughout the type system
4089 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4090 on cross-compiler input files will be able to go away, and various
4092 * CONSTANTP now returns true for quoted forms (as explicitly required
4095 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4097 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4098 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4099 (MAKE-LIST -1). In the longer term, it ought to become true
4100 that declarations are assertions even at SAFETY 1. For now, it's not
4101 quite true even at SAFETY 3, but it's at least more nearly true..
4102 (Note that this change seems to increases the size of the system by
4103 O(5%) and to decrease the speed of the compiler by 20% or more.)
4104 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4105 for the problem of printing dozens of lines of distracting information
4106 about low-level system machinery as part of the top stack frame
4107 on entry to the debugger when an undefined function was called.
4108 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4109 is not reset to COMMON-LISP-USER.
4110 * Compilation of stuff related to dyncount.lisp has been made conditional
4111 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4112 smaller. The various dyncount-related symbols have been moved into
4113 a new "SB-DYNCOUNT" package.
4114 * tty-inspect.lisp has been renamed to inspect.lisp.
4115 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4116 feature has gone away. (When we eventually port to other flavors of
4117 libc and/or Unix, we'll try to make the differences between flavors
4118 invisible at the user level.)
4119 * Various other *FEATURES* tags, and/or their associated conditionals,
4120 have been removed if obsolescent, or given better documentation, or
4121 sometimes given more-mnemonic names.
4123 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4125 * tidied up "make.sh" script
4126 * tidied up system directory structure
4127 * better "clean.sh" behavior
4128 * added doc/FOR-CMUCL-DEVELOPERS
4129 * many many small tweaks to output format, e.g. removing possibly-confusing
4130 trailing #\. character in DESCRIBE-INSTANCE
4131 * (EQUALP #\A 'A) no longer signals an error.
4132 * new hashing code, including EQUALP hashing
4133 * tidied up Lisp initialization and toplevel
4134 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4135 * command line argument processing
4136 * added POSIX-GETENV function to deal with Unix-ish environment variables
4137 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4138 terminating SBCL on EOF
4139 * non-verbose GC by default
4140 * There is no more "sbcl" shell script; the sbcl file is now the C
4141 runtime executable (just like CMU CL).
4142 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4144 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4145 symbol and package information as Lisp native symbol and package objects
4146 instead of strings naming symbols and strings naming packages. This way,
4147 whenever packages are renamed (as in warm init), debug information is
4148 transformed along with everything else.
4149 * tweaked the optimization policy declarations which control the building
4150 of SBCL itself. Now, among other things, the system no longer saves
4151 source location debugging information. (This helps two problems at once
4152 by reducing SBCL size and by keeping SBCL from trying to look for its
4153 sources -- which may not exist -- when reporting errors.)
4154 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4155 debugging and testing purposes
4156 * cleaned up printing, making the printer call PRINT-OBJECT for
4157 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4158 methods, giving nearly-ANSI behavior
4159 * converted almost all special variables to use *FOO* naming convention
4160 * deleted PARSE-TIME functionality, since it can be done portably
4161 * moved some files out of cold init into warm init
4162 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4164 * regularized formatting of source files
4165 * added an install.sh script
4166 * fixed ridiculous memory usage of cross-compiler by making
4167 compiler/alloc.lisp not try to do pooling unless it can hook
4168 itself into the GC of the cross-compilation host. Now the system
4169 builds nicely on my old laptop.
4170 * added :SB-ALLOC in target-features.lisp-expr
4171 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4172 was not implemented)
4173 * re-did condition handling and note reporting in the compiler. Notes
4174 are no longer handled by signalling conditions. Style warnings
4175 and warnings are handled more correctly and reported in such a way
4176 that it's easy to find one or the other in your output (so that you
4177 can e.g. figure out which of many problems caused COMPILE-FILE to
4179 * changed the severity of several compiler warnings from full WARNING
4180 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4181 compiler note reporting so that it doesn't use the condition system
4182 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4183 COMPILE-FILE command)
4184 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4185 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4186 no longer supported, since their implementation was deeply intertwingled
4187 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4188 * removed bogus "support" for compiler macros named (SETF FOO), and
4189 removed the compiler macro for SETF INFO (but only after making a fool
4190 of myself on the cmucl-imp mailing list by posting a bogus patch for
4191 DEFINE-COMPILER-MACRO..)
4192 * Compiled files containing forms which have side effects on the Lisp
4193 reader (such as DEFPACKAGE forms) are now handled more correctly.
4194 (Compiler queuing of top level lambdas has been suppressed by setting
4195 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4196 * deleted various currently-unused source files, e.g. gengc.lisp. They
4197 may be added back at some point e.g. when porting to other architectures,
4198 but until they are it's distracting to distribute them and to try to
4200 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4201 there were so many of them they're just distractions, and UNCROSS is
4202 known to be able to handle the current sources
4203 * moved PROFILE functionality into TRACE, so that it will be clear
4204 how the wrapping and unwrapping of functions when you profile them
4205 interacts with the wrapping and unwrapping of functions when you
4206 trace them. (Actually, the functionality isn't there yet, but at least
4207 the interface specification is there. Hopefully, the functionality will
4208 arrive with some future maintenance release.)
4209 * removed host-oops.lisp
4210 * changed signature of QUIT function to allow UNIX-CODE argument
4211 * fixed READ-SEQUENCE bug
4212 * tweaked verbose GC output so that it looks more like the progress
4213 output that ANSI specifies for functions like LOAD
4214 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4215 * added <http://sbcl.sourceforge.com> to the banner information printed by