1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
3 * improvement: experimental support for mach exception handling on
4 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
6 * improvement: support for GBK external format.
7 (thanks to Chun Tian (binghe))
8 * improvement: the debugger now displays variables that have been closed
9 over, in code compiled with (DEBUG 2) or higher
10 * improvement: support for executable cores on NetBSD (thanks to
12 * new feature: added a RESTART-FRAME debugger command
13 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
14 be used to provide an implementation for FILE-POSITION on Gray streams
15 (thanks to Eric Marsden)
16 * new feature: add syslog, openlog and closelog support to SB-POSIX
17 (thanks to Richard Kreuter)
18 * optimization: the function call overhead in code compiled with
19 a high DEBUG optimization setting is significantly
20 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
21 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
22 (thanks to Tony Martinez)
23 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
24 evaluated once (reported by Kevin Reid)
25 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
26 works again (thanks to Kilian Sprotte)
27 * bug fix: an error is signaled for tagbodies with duplicate tags
28 (thanks to Stephen Wilson)
29 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
30 * bug fix: Win32 port can now handle foreign code unwinding Lisp
31 stack frames from alien callbacks.
32 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
33 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
34 65k pages (thanks to David Woodhouse)
36 changes in sbcl-1.0.1 relative to sbcl-1.0:
37 * new platform: FreeBSD/x86-64, including support for threading.
38 * new feature: the compiler stores cross-referencing information
39 abount function calls (who-calls), macroexpansion (who-macroexpands)
40 and special variables (who-binds, who-sets, who-references) for code
41 compiled with (< SPACE 3). This information is available through the
42 sb-introspect contrib.
43 * new feature: users may subclass SEQUENCE, and have instances of
44 these classes interoperate with standard Common Lisp functions if
45 a number of methods are defined. (This feature is experimental
46 and the interface subject to change based on feedback from SBCL
47 users and the general community)
48 * improvement: sb-sprof traces call stacks to an arbitrary depth on
49 x86/x86-64, rather than the previous fixed depth of 8
50 * improvement: another pthread back-end of mutex "pthread-futex".
51 * bug fix: non-ascii command-line arguments are processed correctly
52 (thanks to Yaroslav Kavenchuk)
53 * bug fix: non-required arguments were not passed correctly when a method
54 defined using DEFMETHOD was called from a mop-generated method using
55 CALL-NEXT-METHOD (reported by Pascal Costanza)
56 * bug fix: recursion is now permitted in accessors through
57 SLOT-UNBOUND. (reported by Pascal Costanza)
58 * bug fix: an error was signaled at startup if the HOME environment
59 variable was defined, but had an empty value (reported by Peter Van Eynde)
60 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
61 signal a PROGRAM-ERROR, not a BUG.
62 * bug fix: SB-POSIX:READDIR works when built with large file support.
63 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
65 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
66 for better automatic indentation support. (thanks to Matt Pillsbury)
67 * optimization: loading generic functions no longer takes O(n^2) time,
68 proportional to the amount of methods in the generic function
69 (reported by Todd Sabin and Jeremy Brown)
70 * optimization: the FIND and POSITION family of sequence functions
71 are significantly faster on arrays whose element types have been
73 * improvements to the Windows port:
74 ** Intermittent heap corruption problems have been fixed. (thanks
75 to Alastair Bridgewater)
76 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
78 ** Lisp is able to unwind foreign exception frames from alien
79 callbacks. (thanks to Alastair Bridgewater)
81 changes in sbcl-1.0 relative to sbcl-0.9.18:
82 * improvement: experimental support for threading on FreeBSD/x86.
83 (thanks to NIIMI Satoshi)
84 * improvement: runtime option --dynamic-space-size can be used
85 to set the size of the dynamic space reserved on startup.
86 * improvement: floating point modes in effect are now saved in
87 core, and restored on startup.
88 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
89 startup, not time since first call to GET-INTERNAL-REAL-TIME.
90 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
91 threads are not running after *SAVE-HOOKS* have run.
92 * improvement: writes to CLOS instance slots are type-checked in code
93 compiled with (SAFETY 3)
94 * improvement: floating-point exception handling on FreeBSD (thanks to
96 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
97 (thanks to Zach Beane)
98 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
100 * improvement: added support for the Shift-JIS external format.
101 (contributed by NIIMI Satoshi)
102 * improvement: callbacks are supported on Linux/PPC. (thanks to
104 * bug fix: compiler bug triggered by a (non-standard) VALUES
105 declaration in a LET* was fixed. (reported by Kaersten Poeck)
106 * bug fix: file compiler no longer confuses validated and already
107 dumped structurres. (reported by Kaersten Poeck)
108 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
109 fill-pointer signals a type-error as required. (thanks to
111 * bug fix: disassemly of funcallable instances works.
112 * bug fix: single stepping on PPC.
113 * bug fix: fix thread-safety problems in the type system (generally
114 manifesting as nonsensical errors like "STRING is a bad type specifier
115 for sequences" or "The value 1 is not of type FIXNUM")
116 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
117 (reported by Josip Gracin)
118 * bug fix: an error is signaled for attempts to displace arrays with
119 incompatible element types (thanks to Mario Mommer)
120 * bug fix: more correct handling of wide characters in debug info
121 (bug reported by Attila Lendvai and fixed by Juho Snellman)
122 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
123 and don't cause extra consing
124 * optimization: MAP and MAP-INTO are significantly faster on vectors
125 whose elements types have been declared.
126 * Improvements to SB-SPROF:
127 ** Support for allocation profiling
128 ** Reduced profiling overhead, especially for long profiling runs
129 * Improvements to the Windows port:
130 ** floating point exceptions are now reported correctly.
131 ** stack exhaustion detection works partially.
132 ** more accurate GET-INTERNAL-REAL-TIME.
133 ** better RUN-PROGRAM behaviour in case of errors in the child
135 ** PROBE-FILE now simplifies pathnames correctly.
136 ** DIRECTORY now works correctly with :WILD-INFERIORS.
137 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
138 1st 1970 (although time timezone data will be unreliable).
139 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
141 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
142 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
143 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
145 * bug fix: two potential GC deadlocks affecting threaded builds.
146 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
147 returns T (reported by Anton Kazennikov)
148 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
149 with non-variable places
150 * bug fix: remove a race condition in the setting of
151 funcallable-instance functions, this should make threaded CLOS
152 code more stable against memory faults.
153 * bug fix: corruption of specials when unbinding is interrupted by an
154 asynchronous unwind (reported by Hannu Koivisto)
155 * improvement: the debugger will now also display local variables that
156 are only used once, for code compiled with a DEBUG optimization quality
159 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
160 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
161 * incompatible change: External-format support for FFI calls. The
162 SB-ALIEN:C-STRING no longer implies an ASCII
163 external-format. Instead the string is subject to external-format
164 conversion. Additionally return values of type SB-ALIEN:C-STRING
165 are by default of type (SIMPLE-ARRAY CHARACTER), not
166 SIMPLE-BASE-STRING. For an alien type that behaves like the old
167 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
168 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
169 * incompatible change: SB-EXT package no longer contains the
170 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
171 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
173 * incompatible change: the single-stepper is no longer available
174 on Alpha, Mips and Sparc platforms.
175 * minor incompatible change: the direct superclasses of
176 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
177 not (STANDARD-OBJECT FUNCTION). This makes the
178 class-precedence-lists of GENERIC-FUNCTION and
179 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
181 * minor incompatible change: the default stream external format on
182 non-unicode SBCL is no longer detected from the locale, but is always
184 * new feature: Add a version of evaluator that uses an interpreter instead
185 of the compiler. EVAL still uses the compiler by default, to switch it
186 to use the interpreter, set the value of the variable
187 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
188 * minor incompatible change: the single-stepper REPL has been merged
189 with the normal debugger (see the "Stepping" heading of the debugger help
190 for more details). The debugger command STEP will no longer switch
191 to the single-stepper REPL.
192 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
193 for a type now works.
194 * bug fix: loading of default sysinit file works. (thanks to Leonid
196 * bug fix: better detection of circularities in the file-compiler.
197 (reported by Marco Monteiro)
198 * bug fix: the CL pathname functions now work with files that have
199 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
200 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
201 systems with Unix98 pty semantics.
202 * bug fix: ASDF-INSTALL will now work with bsd tar.
203 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
205 * bug fix: timers expiring in dead threads no longer cause a
206 type-error (reported by Paul "Nonny Mouse").
207 * bug fix: thanks to more lightweight single-stepper instrumentation,
208 code compiled with (DEBUG 3) will compile and execute significantly faster,
209 and will have more accurate type-inferencing than before
210 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
211 whose bindings are modified
212 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
213 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
214 ** replace dummy implementations of CL:MACHINE-INSTANCE and
215 CL:SOFTWARE-VERSION with proper version
217 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
218 * feature: implemented the READER-METHOD-CLASS and
219 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
220 as specified by AMOP.
221 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
223 * optimization: faster LOGCOUNT implementation on x86 and x86-64
224 (thanks to Lutz Euler)
225 * optimization: hashing of general arrays and vectors has been
226 improved. (reported by Any Fingerhut)
227 * enhancement: SB-INTROSPECT is now able to find definitions of
228 profiled functions. (thanks to Troels Henriksen)
229 * enhancement: compiler-macro expansion applies now to FUNCALL forms
230 as well, allowing compiler-macros for SETF-functions to expand.
231 * enhancement: step-instrumentation no longer wraps known
232 single-value functions with multiple-value context, allowing
233 better type inference.
234 * fixed bug #337: use of MAKE-METHOD in method combination now works
235 even in the presence of user-defined method classes. (reported by
236 Bruno Haible and Pascal Costanza)
237 * fixed bug #339(c): if there are applicable methods not part of any
238 long-form method-combination group, call INVALID-METHOD-ERROR.
239 (reported by Bruno Haible)
240 * fixed bug #361: the :FUNCTION initarg in the protocol for
241 initialization of methods can now be used to override
242 internally-produced optimized functions. (reported by Bruno
244 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
245 system-provided lambda expression no longer cause warnings about
246 unbound #:|pv-table| symbols.
247 * bug fix: improved the handling of type declarations and the
248 detection of violations for keyword arguments with non-constant
250 * bug fix: potentially erronous calls to PATHNAME and
251 MERGE-PATHNAMES were being flushed in some cases.
252 (reported by Richard Kreuter)
253 * bug fix: compiled calls to TYPEP were mishandling obsolete
254 instances. (reported by James Bielman and Attila Lendvai)
255 * bug fix: format strings with ~> without matching ~< no longer
256 trigger an AVER, but signal an understandable error instead.
257 (reported by Antonio Martinez)
258 * bug fix: specifying an output-file without a pathname-type for
259 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
260 of input-file instead of "fasl". (reported by Robert Dodier)
261 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
262 (reported by James Y Knight).
263 * bug fix: compiler-macros-function did not consider the environment
264 argument for shadowing by local functions.
265 * bug fix: compiler-macros expansion was inhibited by local INLINE
267 * bug fix: inline expansions of known functions were subject to
268 step-instrumentation in high debug policies, leading to problems
270 * bug fix: compiler failed to differentiate between different CONS
272 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
273 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
274 * bug fix: #368: incorrect use of expressed vs. upgraded array
276 * bug fix: #306a: more precise unions of array types.
277 * thread-safety improvements:
278 ** CONDITION-WAIT could return early on Linux, if the thread was
279 interrupted and subsequently continued with SIGCONT.
280 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
282 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
283 * added support for the ucs-2 external format. (contributed by Ivan
285 * minor incompatible change: pretty printing of objects of type
286 (cons symbol) is, in the default pprint-dispatch-table, now
287 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
289 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
290 called later in a class's lifetime, possibly as late as when the
291 first instance of the class is created. Previously,
292 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
293 class became finalizeable.
294 * fixed bug: FILE-POSITION sometimes returned inconsistent results
295 for multibyte external-format streams. (thanks to "vbzoli")
296 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
297 with :ALLOCATION :CLASS inherited from superclasses of the
299 * fixed bug: anonymous classes can now be created using the :NAME
300 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
301 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
302 * fixed bug: core-files saved with :EXECUTABLE T can again be
303 executed when SBCL_HOME isn't set. (reported by James Knight)
304 * fixed bug: toplevel LOCALLY forms with declarations could
305 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
306 * fixed bug: printing from several different threads using different
307 values of *print-case* could cause invalid output, due to
308 some internal special variables of the printer not being bound
309 thread-locally (reported by Max Mikhanosha)
310 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
311 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
312 (reported by Pascal Costanza)
313 * fixed bug: FINALIZE-INHERITANCE is called from
314 REINITIALIZE-INSTANCE on classes when the class has previously
315 been finalized, as required by AMOP.
316 * minor code generation optimizations:
317 ** better register allocation in CLOS dispatching functions
318 ** overflow detection when coercing signed bytes to fixnums on x86-64
319 ** is now implemented with one IMUL instruction instead of three shifts
320 ** more efficient bit-vector access on x86 and x86-64
321 ** more efficient access to raw structure slots on x86 and x86-64
322 * fixed some bugs revealed by Paul Dietz' test suite:
323 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
326 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
327 * feature: thread support on Solaris/x86, and experimental thread support
329 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
331 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
332 default method for SLOT-UNBOUND.
333 * minor incompatible change: prevent the user from specializing the
334 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
335 somewhat counter to the intent of the protocol, I (CSR) think, and
336 additionally it just doesn't work in SBCL as currently
337 implemented, thanks to optimizations (that are always valid for
338 the other three SLOT-VALUEish functions, but not for the setter).
339 * minor incompatible change: the :SB-LDB feature is now enabled by
340 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
341 the low-level debugger.
342 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
343 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
344 down to an SB-SYS:FD-STREAM, allowing direct communication in
345 more cases. Temporary files and pipes are still used as a fallback
347 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
348 on the non-reentrant C functions localtime(3) and gmtime(3).
349 * bug fix: global optimization policy was not visible in LOCALLY and
351 * bug fix: class objects can be used as specializers in methods.
352 (reported by Pascal Costanza)
353 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
354 with a type error. (reported by blitz_ on #lisp)
355 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
356 failed with a type error. (reported by Pascal Bourguignon)
357 * bug fix: merging pathnames against defaults with :DIRECTORY
358 starting with '(:RELATIVE :BACK) should preserve the :BACK.
359 (reported by James Y Knight)
360 * bug fix: saving large (>2GB) cores on x86-64 now works
361 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
362 constant index and a (simple-array (signed-byte 32)) array
363 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
364 error (patch by Robert J. Macomber)
365 * feature: TIME now displays the amount of run-time spent in GC
366 * bug fix: The debugger now does a better job of respecting
367 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
368 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
369 workaround for bug 403.)
370 * bug fix: floating point exception handling now works on FreeBSD
371 (thanks to NIIMI Satoshi)
372 * fixed some bugs revealed by Paul Dietz' test suite:
373 ** MISC.641: LET-conversion were not supposed to work in late
375 * improvements to the Win32/x86 port:
376 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
377 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
379 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
380 * new feature: source path information is generated for macro-expansion
381 errors for use in IDE's like Slime (thanks to Helmut Eller)
382 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
383 * bug fix: compiler does not loop forever on an invalid type in
385 * improvement: compilation of most CLOS applications is significantly
387 * optimization: added a limited bytecode compiler for simple toplevel
388 forms, speeding up compilation and FASL loading
389 * bug fix: the statistical profiler now properly distinguishes anonymous
392 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
393 * minor incompatible change: in sbcl-0.9.11 (but not earlier
394 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
395 into subclasses of the finalized class. Now user calls to
396 FINALIZE-INHERITANCE finalize just the one class, and calls by the
397 system return before any subclasses are finalized.
398 * minor incompatible change: The reader no longer ignores errors
399 regarding non-existent packages in #+ and #- feature tests.
400 * new feature: command line options --no-sysinit, --no-userinit to
401 inhibit loading the corresponding init files
402 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
403 for sending data through UDP sockets (thanks to François-René Rideau)
404 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
405 error is signalled instead (thanks to François-René Rideau)
406 * minor incompatible change: Error signalling behaviour of lexical
407 operations violating package locks has changed slightly. Refer to
408 documentation on package locks for details.
409 * enhancement: EVAL can process IF-expressions without resorting to the
411 * bug fix: Some operations involving SIN, COS, or TAN weren't being
412 constant-folded properly due to missing out-of-line functions.
413 (reported by Mika Pihlajamäki)
414 * bug fix: LISTEN sometimes returned T even in cases where no data was
415 immediately available from the stream
416 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
417 were reversed. (reported by Levente Mészáros)
418 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
419 when given a vector displaced to an adjustable array. (reported
421 * bug fix: derivation of float boundaries from numbers outside the
422 appropriate float range (reported by John Wiseman)
423 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
425 * bug fix: precomputing cache entries for generic functions with
426 some subclasses of specializers as yet invalid does not attempt to
427 fill a cache line with a negative offset. (reported by Levente
429 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
430 allow condition class objects as arguments to DOCUMENTATION and
431 (SETF DOCUMENTATION); only find and set documentation for
432 structure names for the STRUCTURE doc-type. (suggested by Gary
434 * improvements to the Win32/x86 port:
435 ** (user-homedir-pathname) and default initialization file
436 locations now know about the user's "Documents and Settings"
437 directory (thanks to Yaroslav Kavenchuk)
438 ** run-program is implemented (thanks to Mike Thomas)
439 ** sockets support (thanks to Timothy Ritchey)
440 ** better backtrace support (thanks to Alastair Bridgewater)
441 ** sb-grovel supported
442 ** asdf-install and sb-posix work somewhat
443 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
444 * improvements to the Solaris/x86 port:
445 ** works on Solaris 11/Solaris Express
446 ** floating-point exception handling support
447 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
448 * fixed some bugs revealed by Paul Dietz' test suite:
449 ** REMOVE-METHOD returns its generic function argument even when
450 the method is not one of the generic functions' methods.
451 ** objects declared as MEMBER types can be admissible arguments to
453 ** printing characters should simply be printed by the FORMAT ~:C
455 ** compiler failure when compiling functions with hairy constant
456 defaults for optional parameters.
457 ** compiler produces wrong code when MAYBE-INLINE-expanding a
458 function, which is already optimized.
460 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
461 * new platform: experimental support for SBCL x86/Darwin, including
462 MacOS X 10.4.5 on Intel.
463 * new feature: Unicode character names are now known to the system
464 (through CHAR-NAME and NAME-CHAR).
465 * new feature: the filesystem location of SBCL's core file is
466 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
467 * minor incompatible change: the contrib modules SB-POSIX and
468 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
469 this change is to make it easier to distribute
470 location-independent binaries.
471 * bug fix: as implied by AMOP, standardized classes no longer have
472 slots named by external symbols of public packages. (reported by
474 * optimization: calling functions via constant symbols -- as in
475 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
476 via the function object as in (FUNCALL #'FOO).
477 * optimization: on x86 and x86-64, the calling convention for the
478 case of dealing with an unknown number of values has been altered
479 to be friendlier to the prediction heuristics implemented,
480 particularly with reference to CALL and RET pairing. (thanks to
481 Alastair Bridgewater)
482 * optimization: on x86, the code for access to thread-local symbol
483 values and binding/unbinding of thread-local symbols is smaller.
484 (thanks to Alastair Bridgewater)
485 * enhancement: CONSTANTP is now able to determine constantness of
486 more complex forms, including calls to constant-foldable standardized
487 functions and some special forms beyond QUOTE.
488 * fixed bug: occasional GC crashes on Solaris/x86
489 * optimization: x86-64 supports stack allocation of results of simple
490 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
491 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
492 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
493 (thanks to James Knight)
494 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
495 by the standard (thanks to Stephen Compall)
497 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
498 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
499 be used for bundling the runtime and the core file into one
500 executable binary. This feature is not currently supported on all SBCL
501 platforms. (thanks to James Bielman and NIIMI Satoshi)
502 * new feature: a generational or ephemeral garbage collector is now
503 the default on the PowerPC platform (both Linux and Darwin). The
504 old Cheney (stop and copy) collector is a build-time option.
505 (thanks to Cyrus Harmon, after Raymond Toy)
506 * minor incompatible change: the method by which SBCL finds its
507 contributed modules has changed; it no longer relies on symbolic
508 links from an $SBCL_HOME/systems directory, but searches directly
509 in the subdirectories of $SBCL_HOME.
510 * enhancement: the dynamic heap size on the Linux/PPC platform is
511 markedly larger, even using the older Cheney garbage collector.
512 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
513 certain complicated string types. (reported by Gary King)
514 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
515 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
516 * fixed bug: DOCUMENTATION on structures no longer signals an
517 error if no structure type of the right name exists. (reported by
519 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
520 applicable non-standard primary or :AROUND method on
521 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
522 no longer cause unbound variable errors. (reported by Kevin Reid)
523 * fixed bug: metacircle resolution in cases where methods have slots
524 added before the slots from STANDARD-METHOD. (reported by Jean
526 * fixed bug: the Power PC assembly code for calling into lisp
527 sometimes computed its return address wrongly (depending on the
528 15th bit of the address, fixed at link time). This used to
529 manifest itself in a segmentation violation while building PCL.
530 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
531 many others over the years)
532 * fixed bug: ROOM no longer reports silly numbers for stack usage if
533 the stack pointer (treated as a signed integer) is negative.
534 (thanks to Peter van Eynde)
536 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
537 * new platform: experimental support for the Windows operating
538 system has been added. (thanks to Alastair Bridgewater)
539 * fixed several bugs in and robustified the PPC FFI (including
540 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
541 * bug fix: allow non-simple string symbol names (reported by Paul
543 * bug fix: interrupt handling on NetBSD (thanks to Richard M
545 * bug fix: saving a core corrupted callbacks on x86/x86-64
546 * bug fix: closed a loophole in metacircularity detection and
547 grounding in the PCL implementation of CLOS.
548 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
549 (reported by tomppa on #lisp)
550 * bug fix: clear the direction flag on Lisp -> C transitions, as
551 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
552 (reported by Andrej Grozin and Hendrik Maryns)
553 * optimization: major improvements to GC efficiency on GENCGC platforms
554 * optimization: faster implementation of EQUAL
555 * optimization: emit more efficient opcodes for some common
556 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
557 * optimization: several other minor code-generation improvements on x86-64
558 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
560 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
561 * minor incompatible change: (SETF CLASS-NAME) and (SETF
562 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
563 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
564 by ANSI as a generic function, and for consistency (SETF
565 GENERIC-FUNCTION-NAME) is treated likewise.
566 * fixed bug #233.b: make constraint propagation notice when a variable
567 value is changed after it is referenced but before it is used
568 * fixed bug #296: no more arbitrary behaviour from filenames with
569 odd characters as --load arguments. (reported by Adam Warner)
570 * fixed bug #390: :CHARACTER-SET pathname components now work as
571 expected. (reported by Tim Daly Jr)
572 * fixed bug #391: complicated :TYPE intersections in slot
573 definitions no longer cause an error in PCL internals.
574 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
575 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
576 returning the number of octets which would be written to the
577 file-stream. (thanks to Robert J. Macomber)
578 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
579 arguments to the internal stream functions. (thanks to David
581 * bug fix: stack exhaustion is now handled on the x86/SunOS
582 platform. (thanks to Daisuke Homma)
583 * bug fix: more accurate ROOM results on GENCGC platforms
584 * optimization: improved type inference for arithmetic-for
585 index variables in LOOP
586 * optimization: faster floating-point SQRT on x86-64
587 * fixed some bugs revealed by Paul Dietz' test suite:
588 ** DOCUMENTATION returns NIL instead of "" for method combinations
589 that don't have a docstring
591 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
592 * minor incompatible change: (SETF CLASS-NAME) and (SETF
593 GENERIC-FUNCTION-NAME) are no longer generic functions, and
594 therefore are not customizeable by user code (as seems to be at
595 least permitted and maybe required by AMOP). As a consolation,
596 however, the SBCL implementation of these functions now calls
597 REINITIALIZE-INSTANCE as specified by AMOP.
598 * bug fix: slot-definition documentation is propagated as per ANSI
599 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
600 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
601 argument, as do eslotd initialization methods. (from Pascal
602 Costanza's "Closer" project)
603 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
604 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
606 * bug fix: it is now possible to have more than one subclass of
607 STANDARD-GENERIC-FUNCTION without causing stack overflow.
608 (reported by Bruno Haible, Pascal Costanza and others)
609 * bug fix: the dependent update protocol now works for generic
610 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
612 * bug fix: condition-class instances corresponding to
613 DEFINE-CONDITION forms are now created eagerly. (reported by
614 Kalle Olavi Niemitalo on comp.lang.lisp)
615 * bug fix: floating point printing is more accurate in some
616 circumstances. (thanks to Simon Alexander)
617 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
618 merged with *DEFAULT-PATHNAME-DEFAULTS*.
619 * bug fix: callbacks on OS X now preserve stack-alignment.
620 (thanks to Cyrus Harmon)
621 * enhancement: the x86-64 disassembler is much better at
622 disassembling SSE instructions. (thanks to Lutz Euler)
623 * enhancement: CHECK-TYPE on a variable now allows type-inference
624 (thanks to Vincent Arkesteijn)
625 * optimization: improved performance of EUC-JP external format.
626 (thanks to NIIMI Satoshi)
627 * optimization: performance improvements to IO on file streams of
628 :ELEMENT-TYPE CHARACTER
629 * optimization: much faster memory allocation on x86-64
630 * optimization: faster garbage collections (latency and throughput)
632 * optimization: faster bignum subtraction and division on x86 and x86-64
633 (thanks to Lutz Euler)
634 * optimization: more accurate type-inference for loops that use a
635 floating point index variable or a negative step.
637 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
638 * bug fix: add a workaround to SBCL looping infinitely at startup on
639 Linux kernels with apparently buggy implementations of personality().
640 (thanks to Svein Ove Aas)
641 * bug fix: Unicode symbols are correctly printed in LDB backtraces
642 (thanks to David Lichteblau)
643 * bug fix: local bindings shadow global symbol macros for
644 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
645 * optimization: non-open coded uses of numeric comparison operators
646 (e.g. >) no longer cons when called with more than one parameter
647 on platforms supporting dynamic-extent allocation.
648 * enhancement: saving cores with foreign code loaded is now supported
649 on MIPS/Linux in addition to the previously supported platforms.
650 * bug fix: FIND-RESTART now tests for activity, not applicability when given
651 a restart object as identifier. (reported by Helmut Eller for CMUCL)
652 * bug fix: division by zero in sb-sprof when no samples were collected
653 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
654 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
655 (thanks to Alexander Kjeldaas)
656 * new feature: ignoring the return values of destructive functions
657 like NREVERSE signals a compile-time style-warning.
658 (thanks to Kevin Reid)
659 * new feature: the alignment of alien structure fields can be
660 explicitly specified. (thanks to Cyrus Harmon)
661 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
662 and dump core on SIGQUIT
664 ** incompatible change: threads do not inherit values of specials
665 from their parents (see manual)
666 ** bug fix: threads stacks belonging to dead threads are freed by the
667 next exiting thread, no need to gc to collect thread stacks anymore
668 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
669 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
670 ** bug fix: dynamic variable and thread start related gc lossage
671 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
673 * fixed some bugs revealed by Paul Dietz' test suite:
674 ** SUBTYPEP is slightly more accurate on heinously complicated
675 CONS types where some of the members have uncertain (in the
676 NIL, NIL sense) type relationships to each other.
678 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
679 no longer purified unless :PURIFY T is explicitly specified.
680 ** Non-purified cores are significantly smaller than before
682 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
683 * new feature: timers based on Zach Beane's excellent timer package
684 * added support for the following external formats: koi8-u,
685 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
686 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
687 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
688 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
689 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
690 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
691 * incompatible change: a threaded SBCL will no longer revert to
692 non-threaded mode on non-NPTL systems, but refuse to start entirely.
693 * bug fix: interrupts are disabled until startup is complete; no
694 more sigsegvs when receiving a signal to soon
695 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
697 * bug fix: add a workaround for the memory randomization features in
698 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
699 workaround will only be in effect on systems with the proc filesystem
701 * bug fix: printing objects of type HASH-TABLE signals a
702 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
704 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
705 have been running for over 50 days. (reported by Gilbert Baumann)
706 * bug fix: the logic for getting names of functions gets less
707 confused when confronded with alternate-metaclass
708 funcallable-instances. (reported by Cyrus Harmon)
709 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
710 consistent, even on internal alternate-metaclass objects.
711 * bug fix: flush closure information collected by physical
712 environment analysis prepass before the main pass. (bug reported
714 * bug fix: compiler pack phase does not modify a hashtable, which is
715 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
716 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
717 may be expanded. (reported by Alexander <malishev> on c.l.l)
718 * bug fix: no more style-warnings when compiling code using LAST
719 under high-SPEED low-SPACE settings. (reported by David Wallin
721 * fixed some incompatibilities between SBCL's MOP and the MOP
723 ** the METAOBJECT class is now implemented;
724 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
725 STANDARD-OBJECT, as required;
726 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
727 now compatible (as required by VALIDATE-SUPERCLASS); there
728 remains a consistency requirement over the presence or absence
729 of the FUNCTION class in the superclass list of finalized
730 classes; see the manual for more details;
731 ** the :AROUND method for COMPUTE-SLOTS on
732 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
733 requested slot ordering.
735 ** bug fix: parent thread now can be gc'ed even with a live
737 ** bug fix: binding a special with PROGV to no value is not
738 the same as that symbol not having been bound (thanks to
740 ** bug fix: binding specials is thread safe (thanks to
742 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
743 and SB-SPROF do not die with 'no handler for signal XX in
744 interrupt_handle_now(..)' anymore
745 ** bug fix: WITH-TIMEOUT works with multiple running threads
746 * fixed some bugs revealed by Paul Dietz' test suite:
747 ** ENSURE-GENERIC-FUNCTION should take a method class object for
748 the :method-class keyword argument.
750 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
751 * new port: the Solaris operating system on x86 processors is now
752 mostly supported, though some rough edges in the environment
753 remain. (thanks to Daisuke Homma)
754 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
755 space for its heap. (thanks to Thiemo Seufer)
756 * optimization: SBCL on MIPS platforms now supports dynamic-extent
757 closures. (thanks to Thiemo Seufer)
758 * minor incompatible change: eof selects abort in the debugger.
759 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
760 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
762 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
763 deprecated, and will go away in a future revision of SBCL.
764 * minor incompatible change: GC-ON and GC-OFF are no longer
765 implemented with a counter, it does not matter how many times gc
766 is switched on or off
767 * bug fix: discriminating functions for generic function classes
768 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
769 make invalid assumptions about method precedence order. (reported
771 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
772 platforms. (thanks to Thiemo Seufer)
773 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
774 * bug fix: correct pathname printing with printer escaping is on.
775 (thanks to Kevin Reid)
776 * bug fix: complex VOP definitions in "user-space" no longer trigger
777 package locks. (reported by Zach Beane)
778 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
779 for funcallable-instance functions, and is no different from
781 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
782 characters in the current readtable. (reported by Nicholas Neuss)
783 * bug fix: SUBTYPEP on various CONS types returns more a more
784 accurate acknowledgment of its certainty.
785 * optimizations: REMOVE-DUPLICATES now runs in linear time on
786 lists in some cases. This partially fixes bug 384.
787 * flush all standard streams before prompting in the REPL and the
789 * bug fix: signal handling and triggering gc do not conflict
790 directly anymore, in particular a high frequency sb-sprof does
791 not prevent gc from running
792 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
793 approximation for timezone and DST information between the
794 universal time epoch and the smallest negative 32-bit time_t.
795 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
796 year 1899 so long as the encoded time is non-negative.
797 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
798 the correct answer for non-constant rotate amounts when performing
799 an inline 32-bit rotation.
801 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
802 there is only one thread in the session
803 ** bug fix: memory leak for streams created in one thread and
804 written to in another
805 ** bug fix: lockup when compiled with gcc4
806 ** bug fix: race that allows the gc to be triggered when gc is
808 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
809 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
811 ** bug fix: finished threads are gc'ed properly
812 * fixed some bugs revealed by Paul Dietz' test suite:
813 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
814 has arguments with a different set of applicable methods from
815 the orignal arguments.
816 ** The type error thrown by MAP now has a correct expected-type
818 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
819 name a compiled function.
820 ** Three MISC tests where a large bignum was improperly coerced to
821 a float that couldn't represent that bignum during type
822 derivation were fixed.
823 ** SUBTYPEP can now handle types involving the negation of a
824 list-form FUNCTION type.
825 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
826 as (COMPLEX (AND RATIO (NOT FIXNUM))).
827 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
829 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
830 * New feature: Experimental support for bivalent streams: streams
831 opened with :element-type :default now allow character and binary
832 (unsigned-byte 8) I/O
833 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
834 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
836 * bug fix: no more highly sporadic "couldn't check whether ~S is
837 readable" when reading a stream and an interrupt hits in the middle
838 of a select system call
839 * compiler better recognizes complex arrays (reported by Eduardo
841 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
842 for variables on SBCL built with linkage-tables. (reported by Luis
844 * various error reporting improvements.
845 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
846 (Thanks to Thiemo Seufer)
847 * bug fix: nested reader invokations work correctly
848 * bug fix: it is possible to have simultaneous references to foreign
849 code and foreign data with the same name.
851 ** added x86-64 support
852 ** incompatible change: the threading api now works with thread
853 objects instead of thread ids
854 ** bug fix: threads are protected from signals and interruption when
855 starting up or going down
856 ** bug fix: a race where an exiting thread could lose its stack to gc
857 ** bug fix: don't halt on infinite error in threads if possible
858 ** fixed numerous gc deadlocks introduced in the pthread merge
859 ** bug fix: fixed thread safety issues in read and print
860 ** bug fix: debugger doesn't hang on session lock if interrupted at
861 an inappropriate moment
862 ** bug fix: run-program is now thread safe(r)
863 ** bug fix: inner with-recursive-lock no longer releases the mutex
864 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
865 * fixed some bugs revealed by Paul Dietz' test suite:
866 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
867 EXPECTED-TYPE slots filled.
868 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
870 ** MAKE-CONDITION accepts classes as type-designators.
871 ** COMPILE may never return NIL.
872 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
873 range before calling Unix time functions
875 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
876 * numerous signal handling fixes to increase stability
877 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
878 * minor incompatible change: we now correctly canonize default
879 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
880 as per the MOP, rather than the historical (INITARG INITFUNCTION
881 INITFORM). (reported by Bruno Haible)
882 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
883 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
884 for more information.
885 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
886 pathname is a directory pathname.
887 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
888 * dynamic space size on PPC has been increased to 768Mb. (thanks to
890 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
891 :METACLASS argument in addition to a class name. (reported by
892 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
893 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
894 (reported by Bruno Haible; patch from Gerd Moellmann)
895 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
897 * bug fix: more cleanups to the floating point exception handling on
898 x86-64 (thanks to James Knight)
899 * bug fix: the compiler does not try to constant fold calls of
900 COERCE and BIT-* functions when they return freshly constructed
901 objects. (reported by Nikodemus Siivola and Paul Dietz)
902 * optimization: improved the MIPS versions of generic (in the
903 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
904 * optimization: direct conversion of (unsigned-byte 32) to floats on
905 the PowerPC platform.
906 * optimization: structure instances with raw slots now use less
907 memory, and probably show better memory locality. (thanks to
909 * optimization: DYNAMIC-EXTENT declarations for lists and closures
910 are treated as requests for stack allocation on the x86-64,
911 Alpha-32, PPC and SPARC platforms.
912 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
913 the wrong file descriptor; implementation of SOCKET-OPEN-P.
914 (thanks to Tony Martinez)
916 ** gcing a dead thread can no longer lead to lockups
917 ** threads block signals until they are set up properly
918 ** errno is no longer shared by threads
919 ** interrupt-thread restores the eflags register on x86
920 ** fixed some lockups due to gc/thread interaction
921 * fixed some bugs revealed by Paul Dietz' test suite:
922 ** invalid dotted lists no longer raise a read error when
924 ** signal an error if a symbol that names a declaration is used
925 as the name of a type, or vice versa
926 ** allow using the (declare (typespec var*)) abbreviation for
927 (declare (type typespec var*)) with all type specifiers
928 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
929 ** always return NIL from PPRINT-POP when OBJECT is NIL
930 ** don't signal errors when pretty-printing malformed LABELS,
931 FLET or MACROLET forms
932 ** declarations in a DOLIST body are also in scope for the
934 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
936 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
939 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
940 * fixed cross-compiler leakages that prevented building a 32-bit
941 target with a 64-bit host compiler.
942 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
943 opened with :IF-EXISTS :OVERWRITE.
944 * fixed bug 281: error for an invalid qualifier in a short-form method
945 combination method is not signalled until the faulty method is called.
946 * bug fix: iteration variable type inferrer failed to deal with open
947 intervals. (reported by Alan Shields)
948 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
949 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
950 or more runtime options were provided to the sbcl binary.
951 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
952 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
953 * optimization: there's now a fast-path for fixnum arguments in the
954 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
955 * optimization: the code generated on x86-64 is more compact thanks
956 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
957 * bug fix: floating-point exception handling now partly works on
958 x86-64. (Thanks to James Knight)
959 * improvement to the MIPS backend from Thiemo Seufer: C-style
960 64-bit long long arguments and return values to alien functions
962 * fixed some bugs revealed by Paul Dietz' test suite:
963 ** the type-error signalled from WARN has a filled-in DATUM slot.
964 ** the type-error required when a stream is not associated with
965 a file has the stream as its datum.
966 ** type-errors on single-floats on x86-64 no longer have
967 :INVALID-OBJECT as the datum
968 ** the type-errors signalled for invalid function names now have
969 a correct expected type
970 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
971 for structures defined with a :TYPE. Documentation strings for
972 typed structures are no longer immediately discarded
973 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
974 broadcast streams and synonym streams. FILE-LENGTH now also works
975 on broadcast streams.
977 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
978 * incompatible change: the --noprogrammer option, deprecated since
979 version 0.7.5, has been removed. Please use the equivalent
980 --disable-debugger option instead.
981 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
983 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
984 inoperational for a while now) has been completely removed.
985 * null lexical environments are now printed as #<NULL-LEXENV>,
986 significantly reducing the amount of clutter in typical backtraces.
987 * documentation on weak pointers, finalization, and after GC hooks
988 has been added to the manual.
989 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
990 as other specialized array types, is much faster. SUBSEQ and
991 COPY-SEQ on such arrays have also been sped up.
992 * optimization: EQL is now more efficient when at least other argument
993 is known to be of type (OR FIXNUM (NOT NUMBER)).
994 * fixed bug: the runtime is now less vulnerable to changes in the
995 size of the SBCL object on OS X, and virtual memory is reserved for
996 all spaces that need to be at a fixed address.
997 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
998 * fixed bug: finalizers and after GC hooks that cause consing are now
1000 * fixed bug: compiler error messages and summaries are now printed to
1001 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1002 * fixed inference of the upper bound of an iteration variable.
1003 (reported by Rajat Datta).
1004 * fixed bug 211e: calling local functions with duplicated constant
1005 keyword argument no longer causes a bogus style warning about an
1007 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1008 ftype declaration to be disregarded. (reported by Dave Roberts)
1009 * fixed bug 373: caused by erronous compilation of references to alien
1010 variables in the runtime on ppc/darwin.
1011 * fixed bug 376: CONJUGATE type deriver.
1012 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1013 (reported by Baughn on #lisp)
1014 * a cross-compiler bug on non-x86 platforms has been identified and
1015 fixed. (thanks to Bruno Haible)
1016 * improvements to the MIPS runtime code for increased stability.
1017 (thanks to Thiemo Seufer)
1018 * increased the maximimum compact environment size to allow
1019 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1020 * partial workaround for bug 135: don't name the function we're
1021 calling for hairy cases of EVAL, so as not to accumulate one environment
1022 entry per call to EVAL. (reported by Kevin Reid)
1023 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1024 * optimization: type testing for non-vector arrays should be faster.
1025 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1027 * bug fix: setting 31st element of a bit vector to zero did not work
1029 * bug fix: redefining a class definition which failed due to a
1030 previous accessor / function clash now works (but see BUGS entry
1031 #380 for more problems in this area). (thanks to Zach Beane)
1032 * the long form of DEFINE-METHOD-COMBINATION disables method group
1033 checking when given a single method group with pattern *. (thanks
1035 * on x86 compiler supports stack allocation of results of simple
1036 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1037 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1038 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1039 * fixed some bugs related to Unicode integration:
1040 ** the restarts for recovering from input and output encoding
1041 errors only appear when there is in fact such an error to
1043 * fixed some bugs revealed by Paul Dietz' test suite:
1044 ** MISC.549 and similar: late transformation of unsafe type
1045 assertions into derived types caused unexpected code
1047 ** SCALE-FLOAT type deriver is less wrong.
1048 ** type derivers for EXP, LOG and similar functions compute result
1049 types for complex arguments better.
1050 ** (MISC.563) CONJUGATE type deriver works for very restricted
1052 ** out-of-line type testers for character strings are available.
1053 ** EQUAL compiler transform understands specialness of objects
1055 ** accessing double-floats stored on the stack now works on x86-64.
1056 ** debugger internals could sometimes create invalid lispobjs,
1057 resulting in GC crashes.
1058 ** MISC.548: type check weakening can convert required type into
1060 ** initialization forms for bindings are not in scope of free special
1063 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1064 * incompatible change: thread support for non-NPTL systems has
1065 been removed - locking is buggy and unreliable. A threaded
1066 SBCL build will now warn at startup and refuse to create
1067 new threads, unless futex support is detected in the kernel
1068 * incompatible change: the top level REPL now has only an ABORT
1069 restart associated with it, not TOPLEVEL and ABORT as it used to.
1070 TOP and TOPLEVEL are now available as debugger commands for
1071 returning to the top level.
1072 * incompatible change: forms evaluated in the REPL now use the
1073 global optimization policy.
1074 * incompatible change: user- and system-initialization files are
1075 no longer processed with LOAD, but by READ and EVAL; hence the
1076 global optimization policy, startup package, readtable, etc,
1078 * The .fasl file format number has been incremented because of
1079 various incompatible changes.
1080 * internal entry point details and argument counts no longer appear
1081 in backtraces unless explicitly requested by setting
1082 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1083 * built-in and standard functions no longer have names like "top
1084 level local call to FOO".
1085 * fixed bug 32: functions defined in non-null lexical environments
1086 now have more legible printed representation
1087 * fixed bug 33: functions defined in non-null lexical environemnts
1088 are now more amenable to inspection by INSPECT.
1089 * workaround for bug 354: XEPs no longer appear in backtraces unless
1090 explicitly requested.
1091 * fixed bug: receiving the signal which results from attempting to
1092 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1093 SIGBUS on other *BSDs) on architectures where the C stack is also
1094 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1095 to debugger confusion or wild instability. (reported by Cheuksan
1097 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1098 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1099 (reported by Lutz Euler)
1100 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1101 compiled code. (reported by Paul Dietz)
1102 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1103 specialized parameter to a method to something that is not TYPEP
1104 the specializer is now possible.
1105 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1106 face of package deletion.
1107 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1108 pathname. (reported by David Tolpin on c.l.l)
1109 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1110 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1111 STRING-STREAM and FILE-STREAM, respectively.
1112 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1113 than previously; in addition, its test suite is now run on build.
1114 (thanks to Luke Gorrie)
1115 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1116 * optimization: added a immediate representation for single-floats
1118 * fixed some bugs revealed by Paul Dietz' test suite:
1119 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1120 ** attempting to create a package with a colliding nickname causes
1121 correctable errors to be signalled.
1122 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1123 ** several x86-64 backend bugs related to sign-extension of immediate
1126 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1127 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1128 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1130 * fixed loading of multiply forward-referenced layouts.
1131 (thanks to Cheuksan Wang)
1132 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1133 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1134 (see COMPILE-FILE documentation for details of :PRINT option)
1135 * fixed bugs 19 and 317: fixed-format floating point printing is
1136 more accurate. This also fixes a bug reported by Adam Warner
1137 related to the ~@F format directive.
1138 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1140 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1141 dispatch table if the from-char is a dispatch macro character.
1142 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1143 by the wordsize no longer ignore the last word. (reported by Lutz
1145 * fixed bug in type checking of dynamic-extent variables. (reported
1147 * optimization: sequence traversal functions use their freedom to
1148 coerce function designators to functions.
1149 * optimization: code with many calls to CLOS methods specialized on
1150 CLOS classes has had redundant type checks removed.
1151 * optimization: added declarations to speed up operations that access
1152 the internal character database (for example STRING-UPCASE)
1153 * optimization: comparison operations between floats and sufficiently small
1154 fixnums no longer create extra rationals
1155 * fixed some bugs related to Unicode integration:
1156 ** portions of multibyte characters at the end of buffers for
1157 character-based file input are correctly transferred to the
1158 start of the buffer at the next read.
1159 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1160 passing it through to OPEN.
1161 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1162 argument given to internal calls to OPEN.
1163 ** the built-in comment readers (introduced by character sequences
1164 ";" and "#|") are more forgiving to encoding errors; they will
1165 STYLE-WARN and then attempt to resync the stream at a character
1166 boundary. (thanks to Teemu Kalvas)
1167 * fixed some bugs in the x86-64 port:
1168 ** Negative short int return values from c-calls are sign-extended
1170 ** The stack is aligned to 16-bytes for c-calls, as required by
1171 the ABI. (thanks to Cheuksan Wang)
1172 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1173 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1174 for enums. (thanks to Vincent Arkesteijn)
1175 ** Multiple small optimizations and bugfixes for floating point
1177 * fixed some bugs revealed by Paul Dietz' test suite:
1178 ** Space, Tab, Linefeed, Return and Page have the invalid
1179 secondary constituent character trait.
1180 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1182 ** WITH-INPUT-FROM-STRING should only update the index place on
1184 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1185 works more reliably.
1186 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1187 with one-dimensional arrays with fill pointers.
1188 ** TYPE-OF failed on a complex with an integer realpart and a
1190 ** compiler failure during type inference for the code of form
1191 (IF (EQL X (THE ...)) ...) (MISC.535).
1193 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1194 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1195 platform. The port supports 61-bit fixnums, large memory spaces
1196 and reloading shared object files.
1197 * enhancement: saving cores with foreign code loaded is now
1198 supported on ppc/Darwin in addition to the previously supported
1200 * enhancement: the statistical profiler now walks deeper into the
1201 call stack for more meaningful call-graphs and accrued time
1202 reports (x86/x86-64 only). It also now reports time spent in
1204 * enhancement: it is now possible to trace most individual methods
1205 of a generic function in addition to tracing the generic function
1207 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1209 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1210 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1211 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1212 produces an error. (thanks to Vincent Arkesteijn)
1213 * bug fix: NAMESTRING on pathnames with :WILD components in their
1214 directories works correctly. (thanks to Artem V. Andreev)
1215 * fixed bug 125: compiler preserves identity of closures. (reported
1217 * bug fixed: functions with &REST arguments sometimes failed with
1218 "Undefined function" when compiled with (DEBUG 3). (reported by
1220 * bug fix: overflow during compiling of setting element of a bit
1221 vector with constant index and value. (reported by Timmy Douglas)
1222 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1223 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1224 * on x86 compiler partially supports stack allocation of dynamic-extent
1226 * GO and RETURN-FROM do not check the extent of their exit points
1227 when compiled with SAFETY 0.
1228 * fixed some bugs related to Unicode integration:
1229 ** encoding and decoding errors are now much more robustly
1230 handled; it should now be possible to recover even from invalid
1231 input or output to the terminal. (thanks to Teemu Kalvas)
1232 ** provided a first cut at implementing STRING-TO-OCTETS and
1233 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1234 ** altered the SB-MD5 contributed module slightly, changing the
1235 interface just enough for it to be supportable for builds where
1236 lisp characters are not eight bits.
1237 * fixed some bugs revealed by Paul Dietz' test suite:
1238 ** the FORMATTER-generated functions for ~V[ conditionals require
1239 the correct number of arguments.
1240 ** READ-FROM-STRING returns the mandated second value when applied
1241 to displaced strings.
1242 ** the #\Rubout and #\Backspace characters are treated as invalid
1243 constituent characters by the tokenizer.
1245 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1246 * new feature: reloading changed shared object files with
1247 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1248 * new feature: references to foreign variables and functions
1249 can now be compiled and loaded before the corresponding shared
1250 object file is loaded, as long as the foreign definitions are
1251 available at runtime.
1252 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1253 supported. (thanks to Dan Debertin)
1254 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1255 just keywords. (thanks to Vincent Arkesteijn)
1256 * fixed bug #331: structure-class instances corresponding to
1257 DEFSTRUCT forms are now created eagerly.
1258 * fixed bug #345: backtraces from calls to undefined functions work
1259 on x86 as well. Related bug #61 is now also partially fixed on x86
1260 (backtraces from throws to unknown catch tags.)
1261 * bug fix: lambda-list parsing is now stricter vrt. order and number
1262 of lambda-list keywords.
1263 * bug fix: as specified by AMOP, an error is signalled if a
1264 class-option appears multiple times in a DEFCLASS form. (reported
1266 * bug fix: RANDOM can be compiled when the compiler derives the type
1267 of its numeric argument as a disjoint set of small integers.
1268 (reported by Paul Dietz)
1269 * bug fix: starting a core saved with shared objects loaded when
1270 those objects are not available no longer causes threaded SBCL to
1271 hang. (reported by Sean Ross)
1272 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1274 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1275 argument, which must always be NIL. (reported by Kalle Niemitalo)
1276 * bug fix: printing 1.0d+23 no longer results in an error.
1277 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1278 * bug fix: structure slot setters preserve evaluation order. (thanks
1280 * fixed some bugs related to Unicode integration:
1281 ** RUN-PROGRAM can allow its child to take input from a Lisp
1282 stream. (reported by Stefan Scholl)
1283 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1284 locales. (reported by Ken Causey)
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** INCF, DECF and REMF evaluate their place form as specified in
1288 ** FORMATTER expands ~{ ~} iteration directives with V or #
1289 parameters correctly.
1290 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1291 consequent uses no arguments correctly.
1292 ** the system has a partial understanding of the (COMPLEX RATIO)
1294 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1297 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1298 * new feature: a build-time option (controlled by the :SB-UNICODE
1299 keyword feature, enabled by default) for building the system with
1300 support for the entire 21-bit character space defined by the
1302 * new feature: the system now has rudimentary external-format
1303 support; the primary user-visible change at this time is that
1304 characters with the high bit set (such as lower-case-e-acute) will
1305 print correctly to a terminal in a UTF-8 environment.
1306 * minor incompatible change: BASE-CHAR no longer names a class;
1307 however, CHARACTER continues to do so, as required by ANSI.
1308 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1309 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1310 has been moved to the SB-EXT package (temporarily re-exported from
1312 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1313 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1315 * fixed bug #351: better error-handlind and reporting for malformed
1317 * fixed bug #350: bignum-printing is now more memory-efficient,
1318 allowing printing of very large bignums, eg. (expt 2 10000000).
1319 (reported by Bruno Haible)
1320 * fixed bug #302: better primitive-type selection for intersection
1322 * fixed bug #308: non-graphic characters now all have names, as
1323 required. (reported by Bruno Haible)
1324 * bug fix: redefining a class with different superclasses now correctly
1325 removes it from the direct-subclasses of its previous superclasses.
1326 (reported by David Morse)
1327 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1328 new value now works. (reported by Bruno Haible)
1329 * bug fix: correct canonicalization of multiple non-standard slot
1330 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1331 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1332 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1333 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1334 now exists, an signals an error.
1335 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1336 during macroexpansion for non-iteration variables as well. (reported
1337 by Bruno Haible for CMUCL)
1338 * bug fix: Cyclic structures and unprintable objects in compiler
1339 messages no longer cause errors. (reported by Bruno Haible)
1340 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1341 and READ-FROM-STRING all now return a primary value of NIL if
1342 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1343 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1344 (reported by Bruno Haible for CMUCL)
1345 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1346 do not designate a condition. (reported by Bruno Haible for
1348 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1349 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1350 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1351 permitting resolution in favour of any of the conflicting symbols.
1352 (reported by Bruno Haible for CMUCL)
1353 * bug fix: EQUAL compiler optimizations is less aggressive on
1354 strings which can potentially compare true despite having distinct
1355 specialized array element types.
1356 * bug fix: unit enumerations can be defined without dividing by
1357 zero. (thanks to Vincent Arkesteijn)
1358 * FORMAT compile-time argument count checking has been enhanced.
1359 (report from Bruno Haible for CMUCL)
1360 * a partial workaround for the bug 262: the compiler does not try to
1361 inline-expand a local function doing RETURN-FROM from a deleted
1362 BLOCK. (thanks to Peter Denno for the bug report and to David
1363 Wragg for the simple test case)
1364 * fixed some bugs revealed by Paul Dietz' test suite:
1365 ** RENAME-PACKAGE allows all package designators as new package
1367 ** constraint propagation and lambda variable substitution are
1368 more cautious in dealing with partially deleted code.
1369 ** compiler sometimes lost reoptimization passes.
1370 ** CERROR, given a condition as condition designator, passes its
1371 remaining arguments to the continue format control without
1373 ** Case-altering FORMAT directives work correctly on non-ASCII
1375 ** The REAL type specifier handles bounds outside the floating
1376 point ranges without signalling FLOATING-POINT-OVERFLOW.
1377 ** Functions with IR1-transformations can create intercomponent
1378 references to global functions.
1379 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1381 ** FORMAT ~R treats a nil value for its first parameter correctly.
1383 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1384 * enhancement: saving cores with foreign code loaded is now
1385 supported on x86/NetBSD and sparc/Linux in addition to the previously
1386 supported platforms.
1387 * bug fix: on some platforms repeated installations caused multiple
1388 copies of HTML documentation to be installed -- should not happen
1389 any more. (reported by Stefan Scholl)
1390 * bug fix: parsing self-recursive alien record types multiple times
1391 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1392 original patch by Helmut Eller for CMUCL)
1393 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1394 (thanks to Richard Kreuter)
1395 * bug fix: defining classes whose accessors are methods on existing
1396 generic functions in other (locked) packages no longer signals
1397 bogus package lock violations. (reported by François-René Rideau)
1398 * bug fix: special variables as DEFMETHOD parameters no longer have
1399 associated bogus type declarations. (reported by David Wragg and
1401 * bug fix: read-write consistency on streams of element-type
1402 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1403 * bug fix: redefiniton of the only method of a generic function with
1404 no DEFGENERIC no longer emits a full WARNING. In addition,
1405 redefinition of generic functions with no DEFGENERIC to an
1406 incompatible lambda list now signals an error. (thanks to Zach
1408 * bug fix: DEFGENERIC now works even when there's a function of the
1409 same name in an enclosing lexical environment. (thanks to Zach
1411 * fixed compiler failure, caused by instrumenting code during
1412 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1413 * optimization: added loop analysis and improved register allocation
1414 * fixed some bugs revealed by Paul Dietz' test suite:
1415 ** POSITION on displaced vectors with non-zero displacement
1416 returns the right answer.
1417 ** (SIMPLE-STRING) is a valid type specifier for sequence
1419 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1421 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1422 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1424 ** The justification version of the FORMAT ~< directive treats
1425 non-zero minpad parameter correctly.
1427 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1428 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1429 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1430 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1431 the supported interface.
1432 * new feature: Single-stepping of code compiled with DEBUG 2 or
1433 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1434 * new feature: saving cores with foreign code loaded is now
1435 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1436 Timothy Moore's work for CMUCL)
1437 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1438 parameters to * instead of NIL if no initform is supplied.
1439 (reported by Johan Bockgaard)
1440 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1441 correctly when FUNCALL appears as the car of the form. Note:
1442 despite this FUNCALL forms are not currently subject to
1443 compiler-macro expansion. (port of Raymond Toy's fix for the
1444 same from CMUCL, reported by Johan Bockgaard)
1445 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1446 (thanks for Teemu Kalvas)
1447 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1448 case instead of terminating the iteration (thanks for Julian
1449 Squires, Sean Champ and Raymond Toy)
1450 * bug fix: incorrect expansion of defgeneric that caused a style
1451 warning. (thanks for Zach Beane)
1452 * bug fix: slot accessor effective method computation works properly
1453 for classes with multiple non-standard applicable methods on
1454 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1455 * on x86 compiler supports stack allocation of results of LIST and
1456 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1457 CMUCL implementation by Gerd Moellmann)
1458 * fixed some bugs revealed by Paul Dietz' test suite:
1459 ** FORMAT strings with both the ~<~:;~> form of the justification
1460 directive and pretty-printing directives cause an error.
1462 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1463 * incompatible change: the internal functions
1464 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1465 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1466 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1467 instead of the old functions.
1468 * new feature: on platforms where "dladdr" is available foreign
1469 function names now appear in backtraces. (based on Helmut Eller's
1471 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1472 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1474 * documentation: ASDF manual is now distributed with SBCL.
1475 * bug fix: SBCL can now load its contributed modules with REQUIRE
1476 even if the system-provided entries have been removed from
1477 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1479 * bug fix: backtraces involving undefined functions or assembly
1480 routines are more informative. (thanks to Brian Downing)
1481 * bug fix: mutually referent alien structures now work correctly.
1482 (reported by Rick Taube)
1483 * bug fix: structures defined by WITH-ALIEN can be referred to
1484 within other definitions in the same WITH-ALIEN.
1485 * bug fix: division operators (MOD, TRUNCATE and the like) with
1486 constant zero divisors and integer dividends no longer generate
1488 * bug fix: provide default methods for INPUT-STREAM-P and
1489 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1490 * bug fix: improve the detection and resolution of MOP metacycles
1491 (where effective method computation depends on the generic function
1492 having its effective method computed). (reported by Bruno Haible)
1493 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1494 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1495 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1496 * bug fix: pathnames with relative directory components can be
1497 represented relative to default pathnames.
1498 * optimization: in taking the GCD of bignums, reduce the two bignums
1499 to approximately the same size (using Euclid's algorithm) before
1500 applying the more sophisticated binary GCD. (thanks to Juho
1502 * optimization: COUNT on bitvectors now operates word-at-a-time.
1503 * optimization: ASH with a positive, but not necessarily constant,
1504 (leftwards) shift, when in a modular context, is compiled to a
1506 * fixed some bugs revealed by Paul Dietz' test suite:
1507 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1508 if the corresponding argument is NIL.
1509 ** FORMAT directives accepting a minpad parameter treat negative
1511 ** literal commas as character parameters to FORMAT directives are
1513 ** literal spaces directly after ~<Newline> directives within a
1514 format-logical-block (~:< ~@:>) do not induce :FILL-style
1515 conditional newlines.
1516 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1517 as specified: it no longer includes conditional newlines).
1518 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1520 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1521 :PER-LINE-PREFIX argument does not evaluate to a string.
1523 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1524 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1525 the manual for details; for now, package locks can be disabled by
1526 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1527 if no major problems are found then it is likely that they will be
1528 compiled in unconditionally.
1529 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1530 unsupported operators on all platforms. To load a shared library
1531 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1532 object file, link it into a shared library outside of SBCL and
1533 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1535 * bug workaround: ROOM T and the GC were not working together
1536 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1537 reporting facility aren't preserved. That reporting has been
1538 disabled, so now until and unless someone figures out how to make it
1539 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1540 * minor incompatible change: as threatened around sbcl-0.8.0, the
1541 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1542 The symbols are also exported from SB-PCL for backwards
1543 compatibility, but more so than before SB-PCL should be treated as
1544 an implementation-internal package.
1545 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1547 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1548 It is known as of this release not to work on the Alpha, however.
1549 * fixed bug #167: errors signalled due to illegal syntax in method
1550 bodies are now more legible.
1551 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1552 designators and can hence be used with TYPEP.
1553 * fixed bug #333: CHECK-TYPE now ensures that the type error
1554 signalled, if any, has the right object to be accessed by
1555 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1556 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1557 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1559 * fixed bug #334: programmatic addition of slots using specialized
1560 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1561 and :ALLOCATION :CLASS slots.
1562 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1563 float exponents. (rereported by Peter Seibel)
1564 * fixed a bug: #\Space (and other whitespace characters) are no
1565 longer considered to be macro characters in standard syntax by
1566 GET-MACRO-CHARACTER.
1567 * fixed bug: initialization of condition class metaobjects no longer
1568 causes an instance of the condition to be created. (reported by
1570 * fixed bug: it is now possible to have slots such that
1571 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1572 description is neither :INSTANCE nor :CLASS.
1573 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1574 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1575 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1576 non-standard SLOT-DEFINITION-ALLOCATION values.
1577 * fixed bugs: various race conditions handling exiting threads.
1578 CL-PPCRE's thread torture test now passes.
1579 * fixed arguably-a-bug: GC time proportional to number of threads in
1580 system even when most of them are idle
1581 * optimization: improved performance of BIT and SBIT on bit-vectors.
1582 * .fasl file incompatibility: The fasl file version number has
1583 been incremented because of changes associated with package locks.
1585 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1586 * minor incompatible change: the system no longer provides
1587 optimization hints (conditions which are TYPEP
1588 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1589 mode; these hints are still emitted when higher SPEED optimization
1591 * new contrib module: a sampling profiler (profiling by statistical
1592 sampling, rather than by instrumenting functions) is available as
1593 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1594 * the behaviour of the standard function ED is now customizeable by
1595 third parties through a hook variable: see ED's documentation
1596 string for information on the protocol.
1597 * the compiler no longer emits efficiency notes for (FUNCALL X)
1598 when the type of X is uncertain under default optimization
1600 * fixed bug 276: mutating a binding of a specialized parameter to a
1601 method to something that is not TYPEP the specializer is now
1603 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1604 exhibit the required behaviour on the x86 platform. (thanks to
1605 Peter van Eynde, Eric Marsden and Bruno Haible)
1606 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1607 even for difficult arguments. (reported by Peter Graves)
1608 * fixed bug 141a: the backquote printer now descends quoted
1610 * fixed another bug in backquote printing: no more destructive
1611 modification of the form's list structure. (reported by Brian
1613 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1614 move between its address being taken and the call to
1615 interrupt_thread, fixing a crashing race condition.
1616 * the SB-POSIX contrib implementation has been adjusted so that it
1617 no longer exhibits ridiculously poor performance when constructing
1618 instances corresponding to C structs.
1620 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1621 * minor incompatible change: the sb-grovel contrib now treats C
1622 structures as alien (in the sense of SB-ALIEN) objects rather than
1623 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1624 has implications for memory management of client code
1625 (sb-grovel-returned objects must now be manually managed) and for
1626 type safety (alien objects now have full types).
1627 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1628 used to control emission of compiler diagnostics, rather than the
1629 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1630 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1631 quality should be considered deprecated.
1632 * install.sh now installs the user manual as well
1633 * (not quite a new documentable feature, but worth considering in
1634 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1635 beginnings of a semantically meaningful condition hierarchy is
1636 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1638 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1639 designator as the defaults argument.
1640 * fixed bug: Displaced arrays whose displaced-to array has become
1641 too small now cause ARRAY-DIMENSION to signal an error, providing
1642 for safer bounds-checking. (reported by Bruno Haible)
1643 * fixed bug: DEFCLASS slot definitions with identical :READER and
1644 :WRITER names now signal a reasonable error. (reported by Thomas
1646 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1648 * fixed bug: Invalid slot specification errors now print correctly.
1649 (thanks to Zach Beane)
1650 * fixed bug 320: Shared to local slot value transfers in class
1651 redefinitions now happen corrently from superclasses as
1652 well. (reported by Bruno Haible)
1653 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1655 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1656 improper lists correctly. (reported by Bruno Haible)
1657 * fixed bug 313: source-transform for <fun-name> was erroneously
1658 applied to a call of a value of a variable with name <fun-name>.
1659 (reported by Antonio Menezes Leitao)
1660 * fixed bug 307: The obsolete instance protocol ensures that
1661 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1662 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1663 remove dead unknown-values globs from the middle of the stack.
1664 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1665 easier to resume long computations after using *BREAK-ON-SIGNALS*
1666 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1667 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1668 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1669 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1670 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1671 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1673 * fixed a (fairly theoretical) bug in string printing: if
1674 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1675 does not have array-element-type equal to the most general string
1677 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1678 function name. (thanks to Nikodemus Siivola)
1679 * fixed bug: compile-time format string checker failed on
1680 non-closed ~{. (reported by Thomas F Burdick)
1681 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1682 don't warn on legal loop code involving more than one aggregate
1683 boolean. (thanks to Nikodemus Siivola)
1684 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1685 in keyword position to MAKE-PACKAGE was not being reported as
1687 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1688 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1690 * fixed bugs 280 and 312: the checking for multiple definitions in a
1691 file is less likely to become confused by uses of inline
1693 * fixed bug: the #S reader macro performs the keyword coercion
1694 specified for slot names. (reported by Kalle Niemitalo)
1695 * fixed bug: lambda lists may contain symbols whose names start with
1696 & but are not lambda-list-keywords; their occurrence triggers a
1698 * fixed bug 321: define-method-combination argument lambda lists do
1699 not cause invalid code to be generated when &optional or &aux
1700 variables have default values. (reported by Bruno Haible)
1701 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1702 have CLOS classes; this appears also to have fixed reported
1703 instability in stack exhaustion detection.
1704 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1705 MUFFLE-WARNING when no associated restart is present is now
1707 * optimization: rearranged the expansion of various defining macros
1708 so that each expands into only one top-level form in a
1709 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1711 * optimization: used a previously little-used slot in symbols to
1712 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1714 * fixed some bugs revealed by Paul Dietz' test suite:
1715 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1717 ** Fixed an optimization bug related to inheritance of initforms
1718 from local to shared slots.
1719 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1720 ** CAST optimizer forgot to flush argument derived type.
1721 ** print/read consistency on floats is now orders of magnitude
1722 more likely. (thanks also to Bruno Haible for a similar report
1724 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1725 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1726 for objects of type REAL. Make it so.
1727 ** SXHASH is specified (infelicitously) to respect similarity,
1728 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1729 it so. (thanks to Markus Ziegler)
1730 ** on the Alpha, the compiler succeeds in compiling functions
1731 returning a known number of arguments greater than 63.
1732 ** fixed handling of invalid NIL arguments in keyword position
1734 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1736 ** provided a readably-printable representation for RANDOM-STATE
1738 ** ensured that pathnames, where they have a namestring, always
1739 print using #P"..." syntax.
1741 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1742 * Support for the forthcoming 2.0 version of the NetBSD kernel
1743 running on x86 hardware has been added. (thanks to Perry
1744 E. Metzger most immediately, and others for their past work)
1745 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1746 current release version; SBCL's OpenBSD support had been broken
1747 since about the time of OpenBSD's switch to ELF binary format.)
1748 * [placeholder for DX summary]
1749 ** user code with &REST lists declared dynamic-extent, under high
1750 speed or space and low safety and debug optimization policy.
1751 * The manual has been converted to Texinfo format and the debugger
1752 chapter from the cmucl manual has been added.
1753 * A facility has been added to extract documentation strings from
1754 sbcl and store them as Texinfo-formatted snippets for inclusion in
1755 the manual (via Texinfo's @include directive)
1756 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1757 the test case to Dave Roberts)
1758 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1759 pointers. (reported by Sean Ross)
1760 * bug fix: PROFILE output is printed nicely even for large numerical
1761 values. (thanks to Zach Beane)
1762 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1763 greater than 32 handle EOF correctly.
1764 * bug fix: on X86 an immediate argument of the IMUL instruction is
1765 correctly printed in disassembly. (thanks to Lutz Euler)
1766 * bug fix: class slots in redefined classes preserve their old
1767 values. (thanks to Bruno Haible and Nikodemus Siivola)
1768 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1769 succeeds. (reported by Marco Baringer)
1770 * fixed some bugs revealed by Paul Dietz' test suite:
1771 ** READ-SEQUENCE now works on ECHO-STREAMs.
1772 ** RATIONALIZE works more according to its specification. (thanks
1775 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1776 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1777 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1778 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1779 should matter to you only if you rebind the printer control
1780 variables and then find you want different bindings in the
1781 debugger than in the ordinary execution of your program.)
1782 * The runtime build system has been tweaked to support building
1783 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1784 assembler and linker. (thanks to Nikodemus Siivola)
1785 * Unbound, undefined, undeclared variables now trigger full
1786 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1787 more often programmer error than deliberate exploitation of undefined
1789 * optimization: the hash algorithm for strings has changed to one
1790 that is less vulnerable to spurious collisions. (thanks to Juho
1792 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1793 needless bounds checking. (thanks to Juho Snellman)
1794 * optimization: implemented multiplication as a modular
1795 (UNSIGNED-BYTE 32) operation on the PPC backend.
1796 * fixed some bugs revealed by Paul Dietz' test suite:
1797 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1799 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1800 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1801 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1803 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1805 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1807 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1808 their output stream on EOF from read.
1809 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1810 have been read to end-of-file.
1811 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1813 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1814 description of determination of which consecutive characters
1816 ** Printing the "Space" character with escaping on now yields "#\\ ",
1817 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1818 ** Reading floating-point numbers with *READ-BASE* set to a number
1819 less than 10 works correctly.
1820 ** Reading floating-point numbers with *READ-BASE* set to a number
1821 more than 10 works correctly.
1822 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1823 the readtable currently in effect.
1825 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1826 * minor incompatible change: parsing of namestrings on a physical
1827 (Unix) host has changed; numbers after the final #\. in a
1828 namestring are no longer interpreted as a version field. This is
1829 intented to be largely invisible to the user, except that the
1830 meaning of the namestring "*.*.*" has changed: it now refers to a
1831 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1832 should usually be replaced by
1833 (make-pathname :name :wild :type :wild :version :wild)
1834 with the added benefit that this is more likely to be portable.
1835 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1836 option to OPEN now signals an error if the file being opened
1837 exists; this may have an impact on existing code.
1838 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1839 signals received, on the PowerPC platforms (both Linux and
1840 Darwin). (thanks to Pierre Mai for pointing out the location of
1842 * several fixes on OS X: The system now builds and runs cleanly on
1843 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1844 (thanks to Brian Mastenbrook)
1845 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1846 second-resolution: integer multiples of 1/3600 between -24 and 24.
1847 (thanks to Vincent Arkesteijn)
1848 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1849 type when called with 1 argument; PEEK-CHAR checked type of
1850 PEEK-TYPE only after having read first character from a
1851 stream. (reported by Peter Graves)
1852 * bug fix: the garbage collector now has much better locality
1853 behaviour, and in particular no longer treats all memory as being
1854 exhausted when many small objects point to each other in a deeply
1856 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1857 recognized as being TYPEP their class.
1858 * bug fix: the PUSHNEW documentation string has been corrected.
1859 (thanks to Vincent Arkesteijn)
1860 * bug fix: defaulting of the value for the last of an atypically
1861 large number of multiple values being bound was not being
1862 performed correctly on the Alpha or PPC platforms
1863 * optimization: implemented multiplication as a modular
1864 (UNSIGNED-BYTE 32) operation on the x86 backend.
1865 * optimization: SEARCH on simple-base-strings can now be open-coded.
1866 (see also contrib/compiler-extras.lisp for inspiration for
1867 teaching the compiler about the Boyer-Moore algorithm).
1868 * value, returned by MAX (and MIN) called with several EQUALP, but
1869 not EQL, arguments now does not depend on compiler settings.
1870 * fixed some bugs revealed by Paul Dietz' test suite:
1871 ** in stack analysis liveness information is propagated from
1872 non-local entry points.
1873 ** pathwise CAST removing failed when the CAST node did not start
1875 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1876 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1878 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1879 host is already defined.
1880 ** RENAME-FILE works on streams instead of signalling an internal
1882 ** PEEK-CHAR uses the current readtable when determining whether
1883 or not a character is whitespace.
1884 ** MERGE-PATHNAMES handles the case when the pathname does not
1885 specify a name while the default-pathname specifies a version.
1886 ** Pathnames now stand a better chance of respecting print/read
1888 ** Attempting to use standardized file system operators with a
1889 pathname with invalid :DIRECTORY components signals a
1891 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1893 ** DIRECTORY on logical pathnames is more correct.
1894 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1895 signal a TYPE-ERROR if their argument is not a stream.
1896 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1897 designator argument does not designate a stream.
1898 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1899 examining the synonym.
1900 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1902 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1903 element-type larger than ([UN]SIGNED-BYTE 32).
1905 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1906 * When built with the :SB-FUTEX feature, threaded builds now take
1907 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1908 for faster/more reliable mutex and condition variable support.
1909 * Incompatible change (but one you probably shouldn't have been using
1910 anyway): the interface and code for arbitrating between multiple
1911 threads in the same user session has been redesigned.
1912 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1913 error when called without an explicit environment argument.
1914 (thanks to Nikodemus Siivola)
1915 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1916 confused about their position. (thanks to Adam Warner and Gerd
1918 * bug fix: Pretty printing backquoted forms with unquotations in the
1919 argument list position of various code constructs such as LAMBDA
1920 now works correctly. (reported by Paul Dietz)
1921 * bug fix: Pretty printing unquotations no longer loses all
1922 stream position information.
1923 * optimization: performance of string output streams is now less
1924 poor for multiple small sequence writes.
1925 * optimization: performance of CSUBTYPEP in the presence of complex
1926 expressions involving CONS and NOT many times has been improved.
1927 (reported by Paul Dietz)
1928 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1930 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1931 streams for the REPL, for the debugger, and so on. (thanks to
1933 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1934 * fixed some bugs revealed by Paul Dietz' test suite:
1935 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1936 is ignored if it is too small, rather than propagating through
1937 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1938 ** extremely complex negations of CONS types were not being
1939 sufficiently canonicalized, leading to inconsistencies in
1941 ** VALUES tranformer lost derived type.
1943 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1944 * fixed a bootstrapping bug: the build process no longer assumes
1945 that the various BOOLE-related constants have the same value in
1946 host and target lisps. (noted by Paul Dietz' test suite on an
1947 SBCL binary built from CLISP)
1948 * The system can now be dynamically linked on the MIPS platform,
1949 which enables dynamic loading of foreign code from Lisp. (thanks
1950 to Ralf Baechle for discussions on the MIPS ABI)
1951 * The system now records debugging information for its own source
1952 files in a filesystem-position-independent manner, relative to
1953 the "SYS" logical host.
1954 * fixed a compiler bug: MV-LET convertion did not check references
1955 to the "max args" entry point. (reported by Brian Downing)
1956 * tweaked disassembly notes to be less confident about proclaiming
1957 some instruction as an LRA. (thanks to Brian Downing)
1958 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1959 now each have their own history, command character, and other
1960 characteristics. (thanks to David Lichteblau)
1961 * fixed some bugs revealed by Paul Dietz' test suite:
1962 ** compiler failure in compiling LOGAND expressions including a
1964 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1965 shift greater than 32.
1966 ** FLUSH-DEST did not mark blocks for type check regeneration.
1967 ** HANDLER-CASE failed to accept declarations in handler clauses
1968 in some circumstances.
1970 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1971 * New code in contrib/sb-introspect (still probably not entirely
1972 stable yet) provides some support for smart Lisp development
1973 environments like SLIME.
1974 * The conditions signalled for errors occurring when loading .fasl
1975 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1976 in a way which should help ASDF recover gracefully.
1977 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1978 stuff has been cleaned up. If you code contrib/ stuff, this might
1979 affect you, and you can look at contrib/README, contrib/STANDARDS,
1980 and/or the 0.8.4.27 diff to check.
1981 * In full calls the compiler now does not generate checks for declared
1982 argument types for all arguments.
1983 * various threading fixes
1984 ** and some experimental patches which didn't make it into
1985 the main tree for this release, but which are shipped in
1986 contrib/experimental-thread.patch as a possible fix for some
1987 failures (deadlock, spinning...) in GC-intensive multithreaded
1989 * fixed PPC build problem (source code incompatibility of different
1990 library versions): added offsetof() hackery which attempts to divine
1991 where glibc maintainers put uc_mcontext today
1992 * fixed bug 282: compiler does not trust type assertions while passing
1993 arguments to a full call.
1994 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1995 &OPTIONAL VALUES type parameter.
1996 * fix bug 214: algorithm for noting rejected templates is now more
1997 similar to that of template seletion. (also reported by rydis on
1999 * fixed bug 141b: printing backquoted information readably and prettily
2000 inserts a space where necessary.
2001 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2002 (CHARACTER 10) are now reported as errors, rather than propagated
2003 as unknown types. (reported by piso on #lisp)
2004 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2005 with values NIL and :ERROR. (thanks to Milan Zamazal)
2006 * fixed bug 191c: CLOS now does proper keyword argument checking as
2007 described in CLHS 7.6.5 and 7.6.5.1.
2008 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2009 longer raise an error; further, using a list as a for-as-arithmetic
2010 counter now raises a meaningful error.
2011 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2012 now understood by sequence creation functions such as MAKE-SEQUENCE
2014 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2015 a STRING-INPUT-STREAM.
2016 * compiler enhancement: SIGNUM is now better able to derive the type
2018 * type declarations inside WITH-SLOTS are checked. (reported by
2020 * fixed some bugs revealed by Paul Dietz' test suite:
2021 ** incorrect optimization of TRUNCATE for a positive first
2022 argument and negative second.
2023 ** compiler failure in let-convertion during flushing dead code.
2024 ** compiler failure while deriving type of TRUNCATE on an
2025 interval, containing 0.
2026 ** ASH of a negative bignum by a negative bignum count now returns
2028 ** intersection of CONS types now canonicalizes properly, fixing
2029 inconsistencies in SUBTYPEP.
2031 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2032 * incompatible change: The --disable-debugger command line
2033 option now clobbers the debugger at a more fundamental
2034 level, by redefining #'INVOKE-DEBUGGER instead of by
2035 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2036 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2037 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2038 enter the debugger REPL and then suffer recursive errors
2039 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2040 while under the new system, BREAK in --disable-debugger mode
2041 terminates the system just as an unhandled error would.
2042 * fixed compiler performance when processing loops with a step >1;
2043 * bug fix: DOCUMENTATION now retrieves generic function
2044 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2045 support has been systematized, and now supports the methods
2046 specified by ANSI, along with a default method and a method for
2047 slot documentation. (reported by Nathan Froyd)
2048 * bug fix: effective methods associated with a generic function are
2049 no longer cached over a change of that generic function's method
2050 combination. (reported by Andreas Fuchs)
2051 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2052 names a non-existent directory. (thanks to Andreas Fuchs)
2053 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2054 the PPC platform, lead to stack corruption; this has been fixed.
2055 (reported by Rainer Joswig)
2056 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2057 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2058 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2059 arithmetic in general, is now much more reliable on the PPC
2061 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2062 platform now returns the right answer.
2063 * optimization: restored some effective method precomputation in
2064 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2065 precomputation is now tunable.
2066 * optimization: compiler-internal data structure use has been
2067 reviewed, and changes have been made that should improve the
2068 performance of the compiler by about 20%.
2069 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2070 simple-base-strings and simple-bit-vectors is improved.
2071 * optimization: the optimization of 32-bit logical and arithmetic
2072 functions introduced in version 0.8.3 on the x86 has been
2073 implemented on the mips, ppc and sparc platforms; an
2074 implementation of the same facility, but for 64-bit arithmetic,
2075 has been added for the alpha.
2076 * microoptimization: the compiler is better able to make use of the
2077 x86 LEA instruction for multiplication by constants.
2078 * bug fix: in some situations compiler did not report usage of
2079 generic arithmetic in (SPEED 3) policy.
2080 * bug 145b fix: compiler used wrong type specifier while converting
2081 MEMBER-types to numeric.
2082 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2084 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2085 index argument now works properly on the Alpha platform.
2086 * bug fix: floating point exception treatment on the Alpha platform
2088 * bug fix: FILE-POSITION works much better on string input and
2089 output streams. (thanks to Nikodemus Siivola)
2090 * bug fix: many threading/garbage collection symptoms sorted.
2091 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2092 might be pseudo-atomic.
2093 * internal change: Stopping for GC is now done with signals not
2094 ptrace. GC is now done in whichever thread wanted it, instead of
2096 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2098 * bug fix: result form in DO is not contained in the implicit
2100 * incompatible change: ICR structure is changed; the value part of
2101 CONTINUATION is now called LVAR; corresponding functions are
2102 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2103 * added type deriver for ISQRT (thanks to Robert E. Brown).
2104 * bug fix: better support for loading from the command line when an
2105 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2107 * fixed some bugs revealed by Paul Dietz' test suite:
2108 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2109 ** ROUND and FROUND now give the right answer when given very
2110 small float arguments.
2111 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2113 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2114 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2115 ** LCM should return a non-negative integer.
2116 ** PARSE-INTEGER returned the index of a terminator instead of the
2117 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2118 ** PARSE-INTEGER returned an incorrect index being applied to a
2120 ** LCM with two arguments of 0 returns 0 rather than signalling
2122 ** unsigned addition of a 32-bit constant with the high bit set no
2123 longer causes an internal compiler error on the x86.
2124 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2125 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2126 with negative last argument.
2127 ** byte specifiers with zero size and position no longer cause
2128 an error during type derivation.
2129 ** bignum multiplication on the Alpha platform now returns the
2131 * porting: The system now builds on SuSE AMD64, although it still
2132 generates a 32-bit binary.
2133 * .fasl file incompatibility: The fasl file version number has
2134 been incremented (because of the changes to internal compiler
2135 data structures referred to above).
2137 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2138 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2139 more accurately, on the Darwin kernel running on PowerPC hardware.
2140 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2141 * Compiler code deletion notes now signal a condition of type
2142 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2143 an associated MUFFLE-WARNING restart.
2144 * The compiler now performs limited argument count validation of
2145 constant format strings in FORMAT, and where appropriate in ERROR,
2146 CERROR and WARN. (thanks to Gerd Moellmann)
2147 * New ASDF-INSTALL contrib can be used for automatic download and
2148 installation of third-party Lisp code from CCLAN or other sites
2150 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2151 which forces another thread to execute a function supplied by the
2153 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2154 now accept and act upon their :ELEMENT-TYPE keyword argument.
2155 (reported by Martin Atzmueller, Edi Weitz)
2156 * bug fix: FILE-POSITION now accepts position designators up to
2157 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2158 is the greater. (thanks to Patrik Nordebo)
2159 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2160 Kalle Olavi Niemitalo)
2161 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2162 printable. (reported by Eric Marsden)
2163 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2164 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2165 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2166 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2167 (thanks to Stig Sandoe)
2168 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2169 UNREAD-CHAR now correctly includes the unread character in the
2170 target sequence. (thanks to Gerd Moellmann)
2171 * bug fix in threaded builds: the system can now be suspended and
2172 resumed by shell job control with minimal disruption.
2173 * bug fixes in times and timezones >2038AD
2174 * better handling of "where is GNU make?" problem in build scripts
2175 (thanks to Nikodemus Siivola)
2176 * new optimization: inside a named function any reference to a
2177 function with the same name is considered to be a self-reference;
2178 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2179 optimization quality.
2180 * new optimization on x86: logical functions and + now have
2181 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2182 used when the result is truncated to 32 bits.
2183 * VALUES declaration is partially enabled.
2184 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2185 * fixed some bugs revealed by Paul Dietz' test suite:
2186 ** The system now obeys the constraint imposed by
2187 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2188 types form a lattice under type intersection.
2189 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2190 ** ASSOC now ignores NIL elements in an alist.
2191 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2192 and (1+ MOST-POSITIVE-FIXNUM) answers.
2193 ** The addition of a method with invalid qualifiers to a generic
2194 function does not cause an error to be signalled immediately;
2195 a warning is signalled, and the error is generated only on
2196 calling the generic function.
2197 * changed .fasl file version number, in part to add type codes for
2198 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2199 obscure ANSI requirements
2201 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2202 * fixed bug 148: failure to inline-expand a local function left
2203 garbage, confusing the compiler.
2204 * fixed bugs 3cd: structure slot readers perform type check if the
2205 slot can have an invalid value (i.e. it is either not initialized
2206 or can be written with a less specific slot writer).
2207 * bug fix: the compiler now traps array references to elements off
2208 the end of an array; previously, the bounds checking in some
2209 circumstances could go off-by-one.
2210 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2212 * type declarations for array element types now obey the description
2213 on the CLHS page "Declaration TYPE", as per discussions on
2214 sbcl-help around 2003-05-08. This means that a declaration
2215 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2216 declaration, all references to BAR will be asserted or assumed
2217 (with THE, so dependent on compiler policy) to involve objects of
2218 type FOO. Note that no such declaration is implied in
2219 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2220 * declared types of functions from the "Conditions"
2221 chapter. (reported by Paul Dietz)
2222 * bug fix: CERROR accepts a function as its first argument.
2223 * bug fix: NTH an NTHCDR accept a bignum as index
2224 arguments. (reported by Adam Warner)
2225 * optimization: character compare routines now optimize comparing
2226 against a constant character. (reported by Gilbert Baumann)
2227 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2228 argument now works properly on the MIPS platform.
2229 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2230 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2231 declared to have a functional type, does not cause an error
2233 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2234 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2235 * fixed bug: if last continuation of a deleted block has a
2236 destination, this destination should be deleted too. (reported by
2238 * fixed a bug in the bootstrap process: the host compiler's values
2239 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2240 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2241 test case from Patrik Nordebo)
2242 * improved the ability of the disassembler on the PPC platform to
2243 provide helpful disassembly notes.
2244 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2245 the class in more cases than previously.
2246 * bug fix: FILE-POSITION now understands :START and :END for
2247 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2248 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2249 8) data. (thanks to David Lichteblau)
2250 * bug fix: it is possible to add a method to a generic function
2251 without lambda list.
2252 * bug fix: reader failed to signal END-OF-FILE inside an
2253 object representation. (reported by Nikodemus Siivola)
2254 * fixed some bugs revealed by Paul Dietz' test suite:
2255 ** LAST and [N]BUTLAST should accept a bignum.
2256 ** condition slot accessors are methods.
2257 ** (VECTOR NIL) is a subtype of STRING.
2259 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2260 * minor incompatible change: some nonsensical specialized lambda
2261 lists (used in DEFMETHOD) which were previously ignored now signal
2263 * minor incompatible change: the system is now aware of the types of
2264 variables in the COMMON-LISP package, and will signal errors for
2265 most violations of these type constraints (where previously they
2266 were silently accepted).
2267 * minor incompatible change: COMPILE-FILE now uses the freedom
2268 afforded (ANSI 3.2.2.3) to use derived function types for
2269 functions defined in the same file. This also permits the system
2270 to warn on static type mismatches and function
2271 redefinition. (Currently it does not work with high DEBUG level.)
2272 * minor incompatible change: VALUES declaration is disabled.
2273 * When issuing notes, the compiler now signals a condition of type
2274 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2275 restart for use in user handlers. It is expected that the
2276 COMPILER-NOTE condition will eventually become a condition
2277 supertype to a hierarchy of note types, which will then be
2278 handleable in a similar fashion. However, at the moment, no such
2279 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2280 but it's an implementation detail, not a classification for the
2282 * Changes in type checking closed the following bugs:
2283 ** type checking of unused values (192b, 194d, 203);
2284 ** template selection based on unsafe type assertions (192c, 236);
2285 ** type checking in branches (194bc).
2286 * A short form of VALUES type specifier has ANSI meaning (it has
2287 increased the number of situations when SBCL cannot perform type
2289 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2290 keywords or constants is permissible.
2291 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2292 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2293 classes. (thanks to Antonio Martinez)
2294 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2295 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2296 argument to a pretty-print stream. Instead, it leaves any such
2297 operations to DESCRIBE-OBJECT methods.
2298 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2300 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2301 lambda lists. (thanks to David Lichteblau)
2302 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2303 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2304 (thanks to Gerd Moellmann)
2305 * fixed bug 246: increased compilation speed of long
2306 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2308 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2310 * DEFCONSTANT now throws a condition of type
2311 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2312 constant to a non-EQL value; CONTINUE and ABORT restarts
2313 respectively change and preserve the value.
2314 * fixed bug 63: The code walker, part of the implementation of CLOS,
2315 is now better at handling symbol macros.
2316 * bug fix: There is no longer an internal implementation type named
2317 CL:LENGTH. (reported by Raymond Toy)
2318 * bug fix: In macro-like defining macros/special operators the
2319 implicit block does not enclose the lambda list.
2320 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2321 enclosed in parentheses) are not suitable as type specifiers, and
2322 their use properly signals an error now.
2323 * bug fix: An argument count mismatch for a type specifier in code
2324 being compiled no longer causes an unhandled error at compile
2325 time, but signals a compile-time warning.
2326 * fixed simple vector readable printing
2327 * bug fix: DESCRIBE takes more care over whether the class
2328 precedence list slot of a class is bound before accessing it.
2329 (reported by Markus Krummenacker)
2330 * bug fix: FORMATTER can successfully compile pretty-printer format
2331 strings which use variants of the ~* directive inside.
2332 * bug fix: SEARCH now applies its TEST predicate to the elements of
2333 the arguments in the correct order. (thanks to Wolfhard Buss)
2334 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2335 in type inference. (thanks to Robert E. Brown)
2336 * bug fix: Reading in symbols with an explicit package name of ""
2337 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2338 * fixed some bugs revealed by Paul Dietz' test suite:
2339 ** NIL is now allowed as a structure slot name.
2340 ** Arbitrary numbers, not just REALs, are allowed in certain
2341 circumstances in LOOP for-as-arithmetic clauses.
2342 ** Multiple class redefinitions before slot access no longer
2343 causes a type error.
2344 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2345 association between the name and a class.
2346 ** Generic functions with non-standard method-combination and over
2347 five methods all of which return constants no longer return NIL
2348 after the first few invocations. (thanks to Gerd Moellmann)
2349 ** CALL-NEXT-METHOD with no arguments now passes the original
2350 values of the arguments, even in the presence of assignment.
2351 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2353 ** DISASSEMBLE works with closures and funcallable instances.
2354 ** ADD-METHOD now returns the generic function, not the new method.
2355 ** FIND-METHOD signals an error if the lengths of the specializers
2356 is incompatible with the generic function, even if the ERRORP
2358 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2359 which its argument is a member.
2360 ** DEFCLASS only redefines the class named by its class-name
2361 argument if that name is the proper name of the class;
2362 otherwise, it creates a new class.
2363 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2364 of the UNBOUND-SLOT condition to the name of the slot.
2365 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2366 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2367 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2368 SLOT-MAKUNBOUND in the specified fashion.
2370 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2371 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2372 cross-compilation host. As a consequence, we can now bootstrap our
2373 way up to SBCL starting with a bare gcc toolchain and human-readable
2374 source code (first the source to CLISP, then the source to SBCL).
2375 * A contributed module containing a partial implementation of the
2376 simple-streams interface has been included. (thanks to Rudi
2378 * A contributed module implementing the RFC1321 Message Digest
2379 Algorithm, known as MD5, has been included.
2380 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2381 no longer has any effect, as the code controlled by this feature
2382 has been deleted. (As far as we know, no-one has ever built using
2383 this feature, and its semantics were confused in any case).
2384 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2385 work on conditions (as required by the ANSI specification),
2386 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2387 expected behaviour on conditions. Users should note, however,
2388 that such behaviour is not required by the ANSI specification,
2389 and so use of this behaviour may render their code unportable.
2390 * fixed some bugs revealed by Paul Dietz' test suite:
2391 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2393 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2394 ** SXHASH on condition objects no longer returns NIL.
2395 ** :ALLOCATION :CLASS slots are better treated; their values are
2396 updated on class redefinition, and initforms inherited from
2397 superclasses are applied.
2398 ** REMOVE-METHOD returns its generic function argument even when
2399 no method was removed.
2400 ** SHARED-INITIALIZE now initializes the values of the requested
2401 slots, including those with :ALLOCATION :CLASS.
2402 ** ALLOCATE-INSTANCE now works on structure classes defined via
2403 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2405 ** SLOT-EXISTS-P now works on conditions, as well as structures
2407 ** MAKE-LOAD-FORM now has the required methods on
2408 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2409 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2410 keyword, and now implements the SLOT-NAMES argument.
2411 ** methods with &OPTIONAL arguments no longer allow too many
2412 arguments to be passed in the call without error.
2413 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2414 option is consistent with the required arguments of the generic
2415 function lambda list.
2416 * bug fix: REQUIRE accepts a string designator. (Thanks to
2418 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2419 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2420 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2421 * bug fix: The compiler now checks for duplicated variables in macro
2423 * bug fix: SETQ on globals returns the correct value.
2424 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2425 not a COMPILER-ERROR (followed by some other strange error on
2426 choosing the CONTINUE restart).
2427 * bug fix: make.sh and friends are now more consistent in the way that
2428 they look for GNU "make".
2430 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2431 * experimental native threads support (on x86 Linux >=2.4 only).
2432 This is not compiled in by default: you need to add :SB-THREAD to
2433 the target features. See the "Beyond ANSI" chapter of the manual
2435 * fix for longstanding nonANSIism: The old distinction between
2436 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2437 The return value from CL:FIND-CLASS is now a CLOS class, and
2438 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2439 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2440 * An interface to the MetaObject Protocol, as described in Kiczales,
2441 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2442 MIT Press, 1991, is available from the SB-MOP package.
2443 * incompatible change: the SB-PCL package should now be considered
2444 a private implementation detail, and no longer a semi-private MOP
2446 * minor incompatible change: due to rearrangement for threads, the
2447 control stack and binding stack are now allocated at arbitrary
2448 addresses instead of being hardcoded per-port. Users affected by
2449 this probably have to be doing advanced things with shared
2450 libraries, and will know who they are.
2451 * minor incompatible change: Previously, all --eval forms used were
2452 processed with READ before any of them were processed with EVAL.
2453 Now each --eval form is processed with both READ and EVAL before
2454 the next --eval form is processed. (Thus package operations like
2455 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2456 work as the user might reasonably expect.)
2457 * minor incompatible change: *STANDARD-INPUT* is now only an
2458 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2460 * minor incompatible change: Y-OR-N-P is now character-oriented,
2461 not line oriented. Also, YES-OR-NO-P now works without errors.
2462 (thanks to Antonio Martinez)
2463 * sb-aclrepl module improvements: an integrated inspector, added
2464 repl features, and a bug fix to :trace command.
2465 * Known functions, which cannot be open coded by the backend, are
2466 considered to be able to check types of their arguments. (fixing
2467 a bug report by Nathan J. Froyd)
2468 * fixed a bug in computing method discriminating functions: It is
2469 now possible to define methods specialized on classes which have
2470 forward-referenced superclasses. (thanks to Gerd Moellmann)
2471 * fixed evaluation order in optional entries (reported by Gilbert
2473 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2474 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2475 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2476 required-by-AMOP NAME argument, as well as CLASS and
2477 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2478 * fixed bug 20: DEFMETHOD can define methods using names that are
2479 not the proper names of classes to designate class specializers.
2480 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2481 Unix file descriptors, instead of blowing up. (thanks to Antonio
2483 * Garbage collection refactoring: user-visible change is that a
2484 call to the GC function during WITHOUT-GCING will not do garbage
2485 collection until the end of the WITHOUT-GCING. If you were doing
2486 this you were probably losing anyway.
2487 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2488 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2489 TYPEP the latter but not the former.
2490 * The compiler issues a full WARNING for calls to undefined functions
2491 with names from the CL package.
2492 * MAP-INTO for a vector destination is open coded. (reported by
2493 Brian Downing on c.l.l)
2494 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2495 documentation string.
2496 * fixed some bugs revealed by Paul Dietz' test suite:
2497 ** COPY-ALIST now signals an error if its argument is a dotted
2499 ** Condition slots are now accessed more correctly in the presence
2500 of multiple initargs for a given slot.
2501 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2502 exclude restarts of the same name associated with a different
2504 ** DEFCLASS of forward-referenced classes with another
2505 forward-referenced class in the superclasses list no longer
2507 ** Condition slots are now initialized once each, not multiple
2508 times. (thanks to Gerd Moellmann)
2509 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2511 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2512 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2513 arguments contain duplicated elements.
2514 ** RESTART-CASE understands local macros.
2515 ** RESTART-CASE associates exactly its own restarts with a condition.
2516 ** ENDP in safe mode checks its argument to be of type LIST.
2517 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2518 in question is unbound.
2519 ** Optimized MAKE-INSTANCE functions no longer cause internal
2520 assertion failures in the presence of duplicate initargs.
2521 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2522 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2523 propagated even in the ctor optimized implementation of
2525 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2527 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2528 * a better implementation of SXHASH on (simple) bit vectors,
2529 measured both in execution speed and in distribution of results
2530 over the positive fixnums, has been installed. Likewise, a better
2531 implementation of EQUAL for simple bit vectors is now available.
2532 * fixed CEILING optimization for a divisor of form 2^k.
2533 * fixed bug 240 (emitting extra style warnings "using the lexical
2534 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2535 by Antonio Martinez)
2536 * fixed SXHASH, giving different results for NIL depending on type
2537 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2538 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2539 variable. (found by Rolf Wester)
2540 * SBCL does not ignore type declarations for special
2541 variables. (reported by rif on c.l.l 2003-03-05)
2542 * some bug fixes in contrib/sb-aclrepl/
2543 * fixed some bugs revealed by Paul Dietz' test suite:
2544 ** a bug in the CONS type specifier, whereby the CAR and CDR
2545 types got intertwined, has been fixed;
2546 ** the type system is now able to reason about the interaction
2547 between INTEGER and RATIO types more completely;
2548 ** APPEND, [N]REVERSE and NRECONC check that those their
2549 arguments, which must be proper lists, are really so;
2550 ** An array specialized to be unable to hold elements has been
2551 implemented, as required -- yes, really -- by ANSI;
2552 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2553 on malformed property lists;
2555 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2556 * incompatible packaging change: in line with Unix convention,
2557 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2558 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2559 * REQUIRE and PROVIDE are now optionally capable of doing something
2560 useful. See the documentation string for REQUIRE.
2561 * infrastructure for a managed SBCL contrib system: contributed
2562 modules in this release include:
2563 ** the ASDF system definition facility;
2564 ** an interface to the BSD Sockets API;
2565 ** an ACL-like convenience interface to the repl;
2566 (thanks to Kevin Rosenberg)
2567 ** an implementation of ROTATE-BYTE, with efficient implementation
2569 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2570 gives a closer approximation to the right answer than previously.
2571 (thanks to Raymond Toy)
2572 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2573 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2574 optional environment arguments, as required by ANSI.
2575 * fixed bugs in other functions taking environment objects, allowing
2576 calls with an explicit NIL environment argument to be compiled
2578 * fixed bug 228: primary return values from
2579 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2580 COMPILE or FUNCTION.
2581 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2582 :TYPE will no longer signal errors on innocuous objects.
2583 * fixed bug 231b: SETQ is better at respecting type declarations in
2584 the lexical environment.
2585 * fixed a bug in DEFCLASS: classes named by symbols with no or
2586 unprintable packages can now be defined.
2587 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2588 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2589 * fixed compiler failure related to checking types of functions.
2590 (reported by Robert E. Brown)
2591 * the compiler is now much more consistent in its error-checking
2592 treatment of bounding index arguments to sequence functions: in
2593 (SAFETY 3) code, errors will be signalled in almost all cases if
2594 invalid sequence bounding indices are passed to functions defined
2595 by ANSI to operate on sequences.
2596 * fixed a bug in the build procedure: documentation of SBCL-specific
2597 packages is now preserved and available in the final Lisp image.
2598 * lifted FDEFINITION lookup out of loops in the implementation of
2599 many list operations. (thanks to Robert E. Brown)
2600 * fixed a bug in the reader: the #n# reader macro now works for
2601 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2602 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2603 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2604 E. Brown and Tony Martinez respectively)
2605 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2606 SLOT-BOUNDP: in particular, it is now able to optimize them much
2607 better, and is now not vulnerable to having packages renamed.
2608 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2609 always called when a slot is not present in an instance. (thanks
2611 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2612 longer signal an error on unreading a character following EOF on
2613 the previous constituent stream. (thanks to Tony Martinez)
2614 * fixed some bugs revealed by Paul Dietz' test suite:
2615 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2616 not just nonnegative fixnums;
2617 ** the logical bit-array operators such as BIT-AND now accept an
2618 explicit NIL for their "opt-arg" argument (to indicate a
2619 freshly-consed result bit-array);
2620 ** ELT now signals an error on an invalid sequence index in safe
2622 ** the type system is now cleverer about negations of numeric
2623 types, and consequently understands the BIGNUM and RATIO types
2625 ** the type system is now cleverer about the interaction between
2626 INTEGER and RATIO types: while bugs still remain, many more
2627 cases are accurately computed;
2628 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2629 if it is in the last clause;
2630 ** CONSTANTLY now correctly returns a side-effect-free function in
2632 ** DECLARE is no longer treated as a special-operator; in
2633 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2634 * incremented fasl file version number due to the change in the
2635 DEFSTRUCT-SLOT-DESCRIPTION structure.
2637 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2638 * minor incompatible change: code processed by the "interpreter" or
2639 EVAL now has a compilation optimization policy of (DEBUG 2)
2640 (changed from (DEBUG 1)) to improve debuggability of interactive
2641 development, and to allow the use of the debug RETURN command in
2643 * an experimental implementation of the RETURN command for the
2644 debugger has been included. (thanks to Frederik Kuivinen)
2645 * fixed bug 62: constraints were not propagated into a loop.
2646 * fixed bug in embedded calls of SORT (reported and investigated by
2648 * fixed some bugs revealed by Paul Dietz' test suite:
2649 ** printing and reading of arrays with some dimensions having
2650 length 0 (thanks to Gerd Moellmann);
2651 ** BOA constructor with &AUX argument without a default value does
2652 not cause a type error;
2653 ** CONSTANTP now returns true for all self-evaluating objects.
2655 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2656 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2657 accessors that are related by inheritance, as specified in the
2658 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2659 Valtteri Vuorikoski)
2660 * The compiler is now able to inline functions that were defined in
2661 a complex lexical environment (e.g. inside a MACROLET).
2662 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2664 * Support for the upcoming FreeBSD-5.0 release has been included.
2665 (thanks to Dag-Erling Smorgrav)
2666 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2667 effect when it is not in a toplevel context.
2668 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2669 stand a better chance of being correct. (thanks to Gerd
2671 * fixed bug in COERCE, which now signals an error on coercing a
2672 rational to a bounded real type which excludes the expected
2674 * The compiler is now able to derive types more accurately from the
2675 COERCE and COMPILE functions.
2676 * fixed bug 223: functional binding is considered to be constant
2677 only for symbols in the CL package.
2678 * fixed bug 231: SETQ did not check the type of a variable being set
2679 (reported by Robert E. Brown)
2680 * A new optimization for MAKE-INSTANCE has been included, fixing
2681 various bugs (including relating to :ALLOCATION :CLASS slots and
2682 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2684 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2685 ** As required by ANSI, LOOP now disallows anonymous collection
2686 clauses such as COLLECT I in conjunction with aggregate boolean
2687 clauses such as THEREIS (= I 1);
2688 ** LOOP now signals an error when any variable is reused in the
2689 same loop (including the potentially useful construct analogous
2690 to WITH A = 1 WITH A = (1+ A);
2691 ** IT is only a special loop symbol within the first clause of a
2692 conditional loop clause;
2693 ** LOOP with a typed iteration variable over a hashtable now
2694 signals a type error iff it should.
2695 * fixed some other bugs revealed by Paul Dietz' test suite:
2696 ** FILE-STREAM now names the class previously known as FD-STREAM;
2697 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2698 argument) no longer signals an error;
2699 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2700 of the slot symbol, rather than using the current package
2701 ((:CONC-NAME "") continues to intern the slot's name in the
2703 * incremented fasl file version number, because of the incompatible
2704 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2705 of the new implementation of DEFINE-COMPILER-MACRO.
2707 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2708 * Support for building SBCL for MIPS platforms running in
2709 little-endian mode has now been checked in, and basic
2710 functionality on said platforms verified.
2711 * minor incompatible change: PCL now records the pathname of a file
2712 in which methods and the like are defined, rather than its
2714 * minor incompatible change: TRUENAME now considers the truename of
2715 a file naming a directory to be the pathname with :DIRECTORY
2716 component indicating that directory.
2717 * minor incompatible change: a NAMED clause in the extended form of
2718 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2719 reason for the previous behaviour is unclear.
2720 * more systematization and improvement of CLOS and MOP conformance
2721 in PCL (thanks to Gerd Moellman and Pierre Mai):
2722 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2724 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2725 primary methods with no specializers;
2726 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2728 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2729 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2730 FORWARD-REFERENCED-CLASSes; error reporting on
2731 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2733 ** SXHASH on CLOS instances now uses a slot internal to the
2734 instance to return different numbers on distinct instances,
2735 while preserving the same return value through invocations of
2737 ** DEFMETHOD signals errors when methods with longer incongruent
2738 lambda lists are added to generic functions;
2739 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2740 CLASS, as specified in AMOP;
2741 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2742 on the order returned by the primary method for classes of
2743 class STANDARD-CLASS;
2744 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2745 * fixed some bugs shown by Paul Dietz' test suite:
2746 ** DOLIST puts its body in TAGBODY;
2747 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2749 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2750 value producing form;
2751 ** if more variables are given to PROGV than values, extra
2752 variables are bound and made to have no value;
2753 ** NSUBSTITUTE on list arguments gets the right answer with
2755 ** ELT signals an error of type TYPE-ERROR when the index argument
2756 is not a valid sequence index;
2757 ** LOOP signals (at macroexpansion time) an error of type
2758 PROGRAM-ERROR when duplicate variable names are found;
2759 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2760 ** FUNCALL of special-operators now cause an error of type
2762 ** PSETQ now works as required in the presence of side-effecting
2763 symbol-macro places;
2764 ** NCONC accepts any object as its last argument;
2765 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2767 ** loop-for-as-package does not require a package to be explicitely
2769 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2771 * fixed bug 166: compiler preserves "there is a way to go"
2772 invariant when deleting code.
2773 * fixed bug 172: macro lambda lists with required arguments after
2774 &REST arguments now cause an error to be signalled. (thanks to
2776 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2777 bug. (thanks to Matthew Danish)
2778 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2780 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2781 arguments are lexically rebound. (thanks to Gerd Moellmann and
2783 * fixed bug 194: error messages are now more informative when there
2784 is no primary method applicable in a call to a generic
2785 function. (thanks to Gerd Moellmann)
2786 * fixed bug in command line argument checking (thanks to Julian
2788 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2789 (thanks to Matthew Danish)
2790 * incremented fasl file version number, because of the
2791 SXHASH-related changes in the layout of CLOS data structures
2793 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2794 * minor incompatible change: The runtime (the Unix executable named
2795 "sbcl") is now much pickier about the .core files it will load.
2796 Essentially it now requires .core files to descend from the same
2797 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2798 as the runtime does. (The intent is to prevent the crashes which
2799 can occur, and which can even be reported as mysterious failures,
2800 when people patch the sources or change the build parameters
2801 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2802 sbcl and .core files.)
2803 * fixed bug: VALUES-LIST is no longer optimized away.
2804 * fixed bug 142: The FFI conversion of C string values to Lisp
2805 string values no longer conses excessively. (thanks to Nathan
2806 Froyd porting Raymond Toy's fix to CMU CL)
2807 * began to systematize and improve MOP conformance in PCL (thanks to
2808 Nathan Froyd, Gerd Moellman and Pierre Mai):
2809 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2811 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2812 ** FINALIZE-INHERITANCE is now called on class finalization;
2813 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2814 argument precedence order.
2815 * fixed bug 202: The compiler no longer fails on functions whose
2816 derived types contradict their declared type.
2817 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2818 so it can be non-toplevel.
2819 * The fasl file version number has changed (because of the new
2820 implementation of DEFMACRO).
2821 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2822 safe code, that any length requirement by their type-specifier
2823 argument is valid. The exceptions to this are described in bug
2825 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2826 functions now check the types of their inputs as required by ANSI.
2827 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2828 introduced symbol is DECLAREd to be SPECIAL.
2829 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2830 * fixed bug: :COUNT argument to sequence functions may be negative
2831 * fixed bug: body of DO-SYMBOLS may contain declarations
2832 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2833 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2834 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2835 symbol macro only once
2836 * fixed printing of call frame when argument list is unavailable
2837 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2838 * compiler no longer signals WARNING on unknown keyword
2841 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2842 * A beta-quality port to the mips architecture running Linux,
2843 based on the old CMUCL backend, has been made. It has been tested
2844 on a big-endian kernel, and works sufficiently well to be able to
2845 rebuild itself; it has not been tested in little-endian mode.
2846 * fixed an inconsistency between gencgc.c and purify.c which made
2847 dumping/loading .core files unreliable
2848 * fixed bug 120a: The compiler now deals correctly with IFs where
2849 the consequent is the same as the alternative, instead of
2850 misderiving the return type. (thanks to Alexey Dejneka)
2851 * fixed bug 113: Logical pathnames are now dumpable (the logical
2852 host is resolved at load-time, throwing an error if it is not
2854 * fixed bug 174: FORMAT's error message is slightly clearer when a
2855 non-printing character is used in a format directive.
2856 * fixed several bugs in compiler checking of type declarations, i.e.
2857 violations of the Python "declarations are assertions" principle
2858 (thanks to Alexey Dejneka)
2859 * fixed several bugs in PCL's error checking (thanks to Gerd
2861 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2863 * fixed bug in compilation of functions as first class values
2864 (thanks to Antonio Martinez-Shotton)
2865 * The compiler's handling TYPE-ERRORs which it can prove will
2866 inevitably happen at runtime has been cleaned up and corrected
2867 in several ways. (thanks to Alexey Dejneka)
2868 * improved argument type checking for various basic arithmetic
2869 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2870 lavished on them in the past that they can be compiled in many
2871 ways in different special cases
2872 * fixed bug 181: compiler checks validity of user supplied type
2874 * cleaned up code flushing in optimization: Function calls which
2875 should signal errors for safety purposes (e.g. which ANSI says
2876 should signal errors when their arguments are of incorrect type)
2877 are no longer optimized away.
2878 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2879 * incremented fasl file version number, because changes in the
2880 implementation of sequence functions like COERCE caused
2881 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2882 in old inline expansions) to become undefined. (Actually these
2883 changes were later undone, so we might very well be binary
2884 compatible with 0.7.7 after all, but leaving the version number
2885 incremented seemed like the simplest and most conservative
2888 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2889 * An alpha-quality port to the parisc architecture running Linux,
2890 based on the old CMUCL backend, has been made. This, even more so
2891 than the other backends, should be considered still a work in
2892 progress; known problems include that the Linux kernel in 64-bit
2893 mode does not propagate the correct sigcontext structure to
2894 userspace, and consequently SBCL on a parisc64 kernel will not
2896 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2897 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2898 suppressed.) Also now that the compiler is looking at declarations
2899 in the environment, it checks optimization declarations as well,
2900 and suppresses inlining when (> DEBUG SPEED).
2901 * More fixes have been made to treatment of floating point exception
2902 treatment and other Unix signals. In particular, floating point
2903 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2904 * The detection and handling of control stack exhaustion (infinite
2905 or very deeply nested recursion) has changed. Stack exhaustion
2906 detection is now done by write-protecting pages at the OS level
2907 and applies at all optimization settings; when found, a
2908 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2909 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2910 be caught using IGNORE-ERRORS.
2911 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2912 correctly when transforming recursive LABELS functions to LETs.
2913 (thanks to Alexey Dejneka)
2914 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2915 that are names of constants or global variables.
2916 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2917 alien routines with docstrings.
2918 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2919 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2921 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2922 (thanks to Eric Marsden)
2923 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2924 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2925 * Bug fix: The SPARC backend can now compile functions involving
2926 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2927 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2928 to a functions expecting a PATHNAME-DESIGNATOR.
2929 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2930 lambda lists. (thanks to Alexey Dejneka)
2931 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2932 functions so that EXPORTing the name of the class doesn't cause
2933 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2934 to Antonio Martinez for reporting this)
2935 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2936 constant-folded without failing an assertion. (thanks to Einar
2937 Floystad Dorum for reporting this)
2938 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2939 types are now dealt with more correctly by the compiler.
2940 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2941 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2942 behaviour analogous to RENAME-FILE. This puts its behaviour more
2943 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2945 * The fasl file version number has changed again. (because of the
2946 bug fix involving the names of PCL MAKE-INSTANCE functions)
2948 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2949 * bug fix: Floating point exceptions are treated much more
2950 consistently on the x86/Linux and PPC/Linux platforms.
2951 * Array initialization with :INITIAL-ELEMENT is now much faster for
2952 cases when the compiler cannot open code the array creation, but
2953 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2954 array accesses have also seen a speed increase.
2955 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2956 specified. (This was at the root of some bad interactions between
2957 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2959 * bug fix: Internal error arguments for undefined functions are now
2960 computed correctly on the PPC/Linux platform.
2961 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2962 Raymond Toy's patch for CMU CL)
2963 * Support for the Solaris 9 operating environment has been included
2964 (thanks to Daniel Merritt)
2965 * A very ugly but hopefully complete draft of the missing FFI chapter
2966 of the manual has been created by reformatting the corresponding
2967 CMU CL manual chapter into (currently very ugly and incoherent)
2968 DocBook and bringing it up to date for SBCL behavior. Thus, the
2969 manual is now essentially complete, at least by my extreme
2970 once-and-only-once standards, whereby it's acceptable to refer to
2971 the doc strings of SB-EXT functions as the primary documentation.
2972 * The fasl file version number has changed again, due to cleanup of
2973 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2974 is no longer a static symbol.)
2976 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2977 * SBCL now builds with OpenMCL (version 0.12) as the
2978 cross-compilation host; also, more progress has been made toward
2979 bootstrapping under CLISP.
2980 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2982 * bug 158 fixed: The compiler can now deal with integer loop
2983 increments different from 1; fixing this turned out also to fix
2985 * bug 169 fixed: no more bogus warnings about using lexical bindings
2986 despite the presence of perfectly good SPECIAL declarations (thanks
2987 to David Lichteblau)
2988 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2989 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2990 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2992 * bug 180 fixed: Method combination specifications no longer ignore
2993 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2994 * bug fix: Structure type predicate functions now check their argument
2995 count as they should.
2996 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2997 correctly. (thanks to Pierre Mai)
2998 * minor incompatible change: The --noprogrammer option is deprecated
2999 in favor of the new --disable-debugger option, which is very similar.
3000 (The major difference is that it takes effect at a slightly different
3001 time at startup, causing handling of errors in --sysinit and
3002 --userinit files will be affected differently.) The
3003 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3004 been added to allow this functionality to be controlled from ordinary
3005 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3006 maintainers, who might want to run non-interactive scripts to
3007 build SBCL cores which will later be used interactively.)
3008 * minor incompatible change: The LOAD function no longer, when given
3009 a wild pathname to load, loads all files matching that pathname.
3010 Instead, an error of type FILE-ERROR is signalled.
3012 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3013 * bug 147 fixed: The compiler preserves its block link/count
3014 invariants more correctly now so that it doesn't crash. (thanks
3016 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3018 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3019 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3020 no-value-for-this-character as NIL (as specified by ANSI).
3021 * HOST-NAMESTRING on physical pathnames now returns a string that is
3022 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3023 (thanks to Christophe Rhodes)
3024 * The Alpha port handles icache flushing more correctly. (thanks to
3026 * More progress has been made toward bootstrapping under CLISP. (thanks
3027 to Christophe Rhodes)
3028 * The fasl file format has changed again, because dynamic loading
3029 on OpenBSD (which has non-ELF object files) motivated some cleanups
3030 in the way that foreign symbols are transformed and passed around.
3031 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3032 is no longer treated as whitespace by the reader, but instead as
3033 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\x7fB") returns
3034 |A
\x7fB|, instead of A as it used to.
3036 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3037 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3038 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3039 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3040 this as of 0.7.1.45, but was left out of the previous news section
3041 (thanks to Dan Barlow)
3042 * SBCL now runs on the Solaris operating system on SPARC architectures
3043 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3044 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3045 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3046 * SPARC backend cleanups, allowing builds of cores optimized for V8
3047 and V9 SPARCS, and also emission of code targeted to a particular
3048 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3050 * SBCL is closer to bootstrapping under CLISP, thanks to various
3051 fixes by Christophe Rhodes.
3052 * The fasl file format has changed again, to allow the compiler's
3053 INFO database to support symbol macros.
3054 * The user manual (in doc/) is formatted into HTML more nicely.
3055 (thanks to coreythomas)
3056 * The system is smarter about SUBTYPEP relationships, especially
3057 those involving NOT types (including types such as ATOM which are
3058 represented internally using NOT types). Thus SUBTYPEP is less
3059 likely to return (VALUES NIL NIL) in general, and in particular
3060 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3062 * The fasl file format has changed again, because the internal
3063 representation of types now includes a new slot to support the new
3064 SUBTYPEP-of-NOT-types logic.
3065 * (not a change in the main branch of SBCL, but a related prototype
3066 which can hopefully be merged into the main branch of SBCL in the
3067 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3068 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3069 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3070 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3071 dispatch macro character combinations. (thanks to Alexey Dejneka)
3072 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3073 that superficially look like logical namestrings correctly.
3074 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3076 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3077 default supplied-p) for &optional and &key arguments. (thanks to
3080 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3081 * incompatible change: The compiler is now less aggressive about
3082 tail call optimization, doing it only when (> SPACE DEBUG) or
3083 (> SPEED DEBUG). (This is an incompatible change because there are
3084 programs which relied on the old CMU-CL-style behavior to optimize
3085 away their unbounded recursion which will now die of stack overflow.)
3086 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3087 for non-GENCGC systems has been increased to 20M (since that
3088 seems much closer to the likely performance optimum for modern
3089 systems than the old 4M value was)
3090 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3091 and *DEBUG-PRINT-LEVEL*
3092 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3093 of CMU CL's support for SPARC, and various endianness and other
3094 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3095 * new syntactic sugar for the Unix command line: --load foo.bar is now
3096 an alternate notation for --eval '(load "foo.bar")'.
3098 ** The system now detects stack overflow and handles it gracefully,
3099 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3100 optimization settings. (This is a good thing in general, and
3101 its introduction in this version should be particularly timely
3102 for anyone whose code fails because of suppression of tail
3104 ** The system now hunts for the C variable "environ" in a more
3105 devious way, to avoid segfaults when the C library version differs
3106 between compile time and run time. (thanks to Christophe Rhodes)
3107 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3108 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3109 ** The compiler no longer issues bogus style warnings for undefined
3110 classes in the same source file as the DEFCLASSes which defined
3111 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3113 ** fixes in CONDITION class precedence list for undefined function
3114 errors (thanks to Alexei Dejneka)
3115 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3116 correctly. (thanks to Dan Barlow)
3117 ** portability fixes aiming at bootstrapping under CLISP (thanks
3118 to Dave McDonald and Christophe Rhodes)
3119 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3120 ** fixes in type translation and and type inference (thanks to
3122 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3123 ** various fixes in the new ports (thanks to Dan Barlow)
3124 * several changes related to debugging:
3125 ** suppression of tail recursion, as noted above
3126 ** stack overflow detection, as noted above
3127 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3128 is now the default. (For some time encapsulation has been more
3129 reliable than the breakpoint-based :ENCAPSULATE NIL
3130 implementation, at least on X86 systems; and I just noticed that
3131 encapsulation also seems closer to the spirit of the ANSI
3134 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3136 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3137 up properly again. (There was a packaging bug in 0.7.0 which
3138 left their definitions in SB-SYS::LOAD-FOREIGN and
3139 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3140 vital for most things which interface to C-level interfaces,
3141 like extensions working with sockets or databases or
3142 Perl-compatible regexes or whatever, and the need to fix
3143 this bug is the main reason that 0.7.1 was released so
3145 ** DEFGENERIC is now choosier about the methods it redefines, so that
3146 reLOADing a previously-LOADed file containing DEFGENERICs does
3147 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3148 cycle now works as it should. (thanks to Alexey Dejneka)
3149 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3150 fixed by Christophe Rhodes. (He actually submitted this patch
3151 months ago, and I delayed until after 0.7.0.)
3152 ** Bug 111 (internal compiler confusion about runtime checks on
3153 FUNCTION types) was fixed by Alexey Dejneka.
3154 * Some internal cleanups (getting rid of variables which aren't
3155 needed now that the byte interpreter is gone) caused the fasl
3156 file format number to change again.
3158 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3159 * major incompatible change: The default fasl file extension, i.e. the
3160 default extension for files produced by COMPILE-FILE, has changed
3161 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3163 ** There are many changes in the implementation of the compiler.
3164 SBCL is now essentially a compiler-only implementation of ANSI
3165 Common Lisp. EVAL still "interprets" a few special cases, but
3166 almost all the interesting cases are handled by creating
3167 a LAMBDA expression, calling COMPILE on it, then calling
3168 FUNCALL on the result.
3169 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3170 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3171 Since the code is newer, there might still be some new bugs
3172 (though not as many as before Martin Atzmueller's fixes:-). But
3173 the new code is substantially simpler and clearer, and hopefully
3174 any remaining bugs will be simpler, less fundamental, and more
3175 fixable then the bugs in the old code.
3176 ** The revised compiler is still a little unsteady on its feet.
3178 *** The debugging information it produces (particularly the names
3179 of FUNCTION objects) is sometimes much less useful than what
3180 the old compiler produced.
3181 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3182 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3183 within a MACROLET) has been temporarily weakened.
3184 ** There are new compiler optimizations for various functions:
3185 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3186 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3187 *** the math functions TRUNCATE, FLOOR, and CEILING
3188 *** the function-of-all-trades COERCE
3189 Mostly these should be transparent, but there's one
3190 potentially-annoying problem (bug 117): when the compiler
3191 inline-expands a function and does type analysis on the result,
3192 it can create control paths which have type mismatches, and
3193 when it can't prove that those control paths aren't taken,
3194 it will issue WARNINGs about the type mismatches. This is
3195 a particular problem in practice for the new sequence functions.
3196 It's not clear how this should be fixed, and for now, a
3197 workaround is given in the entry for 117 in the BUGS file.
3198 ** (Because of the interaction between the two previous items --
3199 occasional inlining problems and new inline expansions -- some
3200 of the new sequence function optimizations won't really kick in
3201 completely until debugging information, and then inlining, are
3202 straightened out in some future version.)
3203 * minor incompatible changes:
3204 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3205 DIRECTORY no longer implicitly promotes NIL slots of its
3206 pathname argument to :WILD. In particular, when you ask for the
3207 contents of a directory (which you used to be able to do without
3208 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3209 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3210 ** changes in behavior that ANSI explicitly defines to be
3211 implementation dependent:
3212 *** The new compiler-only implementation still conforms with ANSI,
3213 but acts a little different than before. Besides the obvious
3214 changes in performance tradeoffs (that the cost per form passed
3215 to EVAL has gone up, and the cost per form executed by EVAL
3216 has gone down), the behavior of the system changes a little
3217 because there are no longer any interpreted function objects.
3218 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3219 e.g. doing COMPILE on the output of interactive DEFUN is
3221 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3223 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3224 has changed from (:CL) to NIL.
3225 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3226 e.g. #\Newline and #\Tab, don't have names specified in the
3227 ANSI Common Lisp standard, is now based on their ASCII symbolic
3228 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3229 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3230 are no longer used for output.
3231 ** changes in internal implementation constants:
3232 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3233 4 million. (If your application spends a lot of time GCing and
3234 you have a lot of RAM, you might want to experiment with
3235 increasing it even more.)
3236 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3237 However, almost all old code should still continue to work without
3238 immediate update, as SB-C-CALL is now a (deprecated) nickname
3240 ** Old operator names in the style DEF-FOO are now deprecated in
3241 favor of new corresponding names DEFINE-FOO, for consistency with
3242 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3243 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3244 internal symbols, but a few supported extensions like
3245 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3246 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3247 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3248 etc. as you get deeper into recursive calls to the debugger
3249 command loop, instead of the old "5]", "5]]", "5]]]"
3250 sequence. (I was motivated to do this when squabbles between
3251 ILISP and SBCL left me very deeply nested in the debugger. In the
3252 short term, this change will probably provoke more ILISP/SBCL
3253 squabbles, but hopefully it will be an improvement in the long run.)
3254 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3255 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3256 function, since declaiming return types involving aliens is
3257 (1) annoyingly messy to do by hand and (2) vital to efficient
3258 compilation of code which calls such functions.
3259 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3260 longer reexported by the SB-EXT package. They're solely useful
3261 for alien code, so it seems more logical that you should get
3262 them from the SB-ALIEN package, not in SB-EXT.
3263 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3264 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3265 features. Instead, the code that they used to control is always
3266 built into the system.
3267 * many other bug fixes
3268 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3269 advantage of the new EVAL-WHEN stuff and to clean them up in
3270 general, and they are now more ANSI-compliant in a number of
3271 ways. Martin Atzmueller is responsible for a lot of this.
3272 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3274 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3275 *** correct ERROR type for various file operations
3276 *** some fixes for Lisp streams
3277 *** DEFMETHOD syntax checking
3278 *** changing old weird representation of debug information as
3279 strings (which, among their other deficiencies, don't transform
3280 correctly when you rename packages, and don't change their
3281 print representation when you change things like *PACKAGE*
3282 and *PRINT-LENGTH*) to symbols and lists of symbols
3283 He also made several improvements and fixed several bugs in DESCRIBE.
3284 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3286 *** misbehavior of WRITE-STRING/WRITE-LINE
3287 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3288 and several other LOOP problems as well
3289 *** DIRECTORY when similar filenames are present
3290 *** DEFGENERIC with :METHOD options
3291 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3292 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3293 *** argument ordering in FIND with :TEST option
3294 *** mishandled package designator argument in APROPOS-LIST
3295 *** various problems in the backquote readmacro
3296 *** a bug in APROPOS
3297 *** probably some others that I'm not describing very well here,
3298 since the CVS log documents them by reference to sbcl-devel
3299 messages, and the SourceForge archives aren't working well.:-(
3300 ** Dan Barlow improved the Alpha port (and is making progress on the
3301 PPC port, for those of you who think different).
3302 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3303 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3304 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3305 continues to work on the SPARC port (for those of you in a position
3306 to look down upon our little PC-compatible boxes from a great height).
3307 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3308 and uses that copy, rather than the current dynamic value, when
3309 it's trying to decide whether to truncate output. Thus e.g.
3310 (let ((*print-lines* 50))
3311 (pprint-logical-block (stream nil)
3313 (let ((*print-lines* 8))
3314 (print (aref possiblybigthings i) stream)))))
3315 should now truncate the logical block only at 50 lines, instead of
3316 often truncating it at 8 lines, as it did before.
3317 * The doc/cmucl/ directory, containing old CMU CL documentation
3318 from the time of the fork, is no longer part of the base system.
3319 SourceForge has shut down its anonymous FTP service, and with it
3320 my original plan for distributing the old CMU CL documentation
3321 there. For now, if you need these files you can download an old
3322 SBCL source release and extract them from it.
3323 * The fasl file version number changed again, for dozens of reasons,
3324 some of which are apparent above.
3326 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3327 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3328 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3329 patch, so that DISASSEMBLE output is much nicer.
3330 * The code in the SB-PROFILE package now seems reasonably stable.
3331 I still haven't decided what the final interface should look like
3332 (I'd like PROFILE to interact cleanly with TRACE, since both
3333 facilities use function encapsulation) but if you have a need
3334 for profiling now, you can probably use it successfully with
3335 the current CMU-CL-style interface.
3336 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3337 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3338 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3339 link now returns the dangling link itself, and for similar
3340 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3341 link itself. (In these cases the old code signalled an error and
3342 looped endlessly, respectively.) Thus, DIRECTORY now works even
3343 in the presence of dangling and cyclic symbolic links.
3344 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3345 is now a supported extension again, since the consensus on
3346 sbcl-devel was that it can be useful for ordinary development
3347 work, not just for debugging SBCL itself.
3348 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3349 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3350 that currently-defined functions might be redefined later with
3351 different return types.
3352 * Hash tables can be printed readably, as inspired by CMU CL code
3353 of Eric Marsden and SBCL code of Martin Atzmueller.
3354 * better error handling in CLOS method combination, thanks to
3355 Martin Atzmueller porting Pierre Mai's CMU CL patches
3356 * more overflow fixes for >16Mbyte I/O buffers
3357 * A bug in READ has been fixed, so that now a single Ctrl-D
3358 character suffices to cause end-of-file on character streams.
3359 In particular, now you only need one Ctrl-D at the command
3360 line (not two) to exit SBCL.
3361 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3363 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3364 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3365 does the right thing.
3366 * The implementation of some type tests, especially for CONDITION
3367 types, is now tidier and maybe faster, due to CMU CL code
3368 originally by Douglas Crosher, ported by Martin Atzmueller.
3369 * Some math functions have been fixed, and there are new
3370 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3371 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3372 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3373 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3374 likely to be built into sbcl-0.7.0.)
3375 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3376 CL, ported to SBCL by Martin Atzmueller)
3377 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3378 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3379 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3380 * There's a new slam.sh hack to shorten the edit/compile/debug
3381 cycle for low-level changes to SBCL itself, and a new
3382 :SB-AFTER-XC-CORE target feature to control the generation of
3383 the after-xc.core file needed by slam.sh.
3384 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3385 COMPILE-FILE is no longer supported, so that now every function
3386 gets an entry point, so that block compilation looks a little
3387 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3388 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3389 returns the number of bytes consed since the system started,
3390 rather than the number consed since the first time the function
3391 was called. (The new definition parallels ANSI functions like
3392 CL:GET-INTERNAL-RUN-TIME.)
3393 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3394 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3395 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3396 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3397 * Fasl file version numbers are now independent of the target CPU,
3398 since historically most system changes which required version
3399 number changes have affected all CPUs equally. Similarly,
3400 the byte fasl file version is now equal to the ordinary
3403 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3404 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3405 is no longer recognized. I apologize for this, because it was
3406 listed in SB-EXT as a supported extension, but I found that
3407 its existing behavior was poorly specified, as well as incorrectly
3408 specified, and it looked like too much of a mess to straighten it
3409 out. I have enough on my hands trying to get ANSI stuff to work..
3410 * many patches ported from CMU CL by Martin Atzmueller, with
3411 half a dozen bug fixes in pretty-printing and the debugger, and
3412 half a dozen others elsewhere
3413 * fixed bug 13: Floating point infinities are now supported again.
3414 They might still be a little bit flaky, but thanks to bug reports
3415 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3416 as flaky as they were.
3417 * The --noprogrammer command line option is now supported. (Its
3418 behavior is slightly different in detail from what the old man
3419 page claimed it would do, but it's still appropriate under the
3420 same circumstances that the man page talks about.)
3421 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3422 are now supported, and enabled by default. Thus, the compiler can
3423 handle many floating point and complex operations much less
3424 inefficiently. (Thus e.g. you can implement a complex FFT
3426 * The compiler now detects type mismatches between DECLAIM FTYPE
3427 and DEFUN better, and implements CHECK-TYPE more correctly, and
3428 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3429 thanks to patches from Martin Atzmueller.
3430 * various fixes to make the cross-compiler more portable to
3431 ANSI-conforming-but-different cross-compilation hosts (notably
3432 Lispworks for Windows, following bug reports from Arthur Lemmens)
3433 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3434 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3435 fixed thanks to Pierre Mai's CMU CL patches.
3436 * fixes to keep the system from overflowing internal counters when
3437 it tries to use i/o buffers larger than 16M bytes
3438 * fixed bug 45a: Various internal functions required to support
3439 complex special functions have been merged from CMU CL sources.
3440 (When I was first setting up SBCL, I misunderstood a compile-time
3441 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3442 * improved support for type intersection and union, fixing bug 12
3443 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3444 more obscure bugs as well
3445 * some steps toward byte-compiling non-performance-critical
3446 parts of the system, courtesy of patches from Martin Atzmueller
3447 * Christophe Rhodes has made some debian packages of sbcl at
3448 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3449 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3450 stable, but are nonetheless usable. When he's ready, I'd be happy
3451 to add them to the SourceForge "File Releases" section. (And if
3452 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3453 * new fasl file format version number (because of changes in
3454 internal representation of (OR ..) types to accommodate the new
3455 support for (AND ..) types, among other things)
3457 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3458 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3460 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3462 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3463 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3464 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3465 * DESCRIBE now works on structure objects again.
3466 * Most function call argument type mismatches are now handled as
3467 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3468 know whether the function will be redefined before the call is
3469 executed. (The compiler could flag local calls with full WARNINGs,
3470 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3471 it doesn't keep track of enough information to know whether calls
3472 are local in this sense.)
3473 * Compiler output is now more verbose, with messages truncated
3474 later than before. (There should be some supported way for users
3475 to override the default verbosity, but I haven't decided how to
3476 provide it yet, so this behavior is still controlled by the internal
3477 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3478 src/compiler/ir1util.lisp.)
3479 * Fasl file format version numbers have increased again, because
3480 support for the Gray streams extension changes the layout of the
3481 system's STREAM objects.
3482 * The Gray subclassable streams extension now works, thanks to a
3483 patch from Martin Atzmueller.
3484 * The full LOAD-FOREIGN extension (not just the primitive
3485 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3486 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3487 but like most other programs, it defaults to copying the Unix
3488 environment from the original process instead of starting the
3489 new process in an empty environment.
3490 * Extensions which manipulate the Unix environment now support
3491 an :ENVIRONMENT keyword option which doesn't smash case or
3492 do other bad things. The CMU-CL-style :ENV option is retained
3493 for porting convenience.
3494 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3495 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3497 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3499 * A patch from Martin Atzmueller seems to have solved the SIGINT
3500 problem, and as far as we know, signal-handling now works cleanly.
3501 (If you find any new bugs, please report them!)
3502 * The system no longer defaults Lisp source file names to types
3503 ".l", ".cl", or ".lsp", but only to ".lisp".
3504 * The compiler no longer uses special default file extensions for
3505 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3506 seems to expect a single default extension for all compiled code,
3507 and there's no compelling reason to try to stretch the standard
3508 to allow two different extensions.) Instead, byte-compiled files
3509 default to the same extension as native-compiled files.
3510 * Fasl file format version numbers have increased again, because
3511 a rearrangement of internal implementation packages made some
3512 dumped symbols in old fasl files unreadable in new cores.
3513 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3514 many fewer weird special cases.
3515 * Bug #17 (differing COMPILE-FILE behavior between logical and
3516 physical pathnames) has been fixed, and some related misbehavior too,
3517 thanks to a patch from Martin Atzmueller.
3518 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3519 by Tim Moore, ported to SBCL by Martin Atzmueller.
3520 * Martin Atzmueller fixed several filesystem-related problems,
3521 including bug #36, in part by porting CMU CL patches, which were
3522 written in part by Paul Werkowski.
3523 * More compiler warnings in src/runtime/ are gone, thanks to
3524 more patches from Martin Atzmueller.
3525 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3528 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3530 * DESCRIBE now works on CONDITION objects.
3531 * The debugger now handles errors which arise when trying to print
3532 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3534 * The build system now uses an additional file, customize-target-features.lisp,
3535 to allow local modifications to the target *FEATURES* list. (The point of
3536 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3537 debugging features enabled, without having to worry about propagating it
3538 into everyone's system when I do a "cvs update".) When no
3539 customize-target-features.lisp file exists, the target *FEATURES* list
3540 should be constructed the same way as before.
3541 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3542 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3543 warn about behavior which is undefined under the ANSI spec. Note: This
3544 is specified by ANSI, but it's not very popular with programmers.
3545 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3546 macro in the SBCL sources for an example of a workaround which you
3547 might use to make portable ANSI-standard code which does what you want.
3548 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3549 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3550 when it's used as a non-toplevel form. (This is required in order
3551 to implement the DEFCONSTANT-EQX macro.)
3552 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3553 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3554 funny things, due to bugs in the implementation of EVAL-WHEN
3555 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3556 non-toplevel DEFCONSTANTs are uncommon.)
3557 * The core file version number and fasl file version number have been
3558 incremented, because the old noncompliant DEFCONSTANT behavior involved
3559 calling functions which no longer exist, and because I also took the
3560 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3561 * fixed bug 1 (error handling before read-eval-print loop starts), and
3562 redid debugger restarts and related debugger commands somewhat while
3564 ** The QUIT debugger command is gone, since it did something
3565 rather different than the SB-EXT:QUIT command, and since it never
3566 worked properly outside the main toplevel read/eval/print loop.
3567 Invoking the new TOPLEVEL restart provides the same functionality.
3568 ** The GO debugger command is also gone, since you can just invoke
3569 the CONTINUE restart directly instead.
3570 ** The TOP debugger command is also gone, since it's redundant with the
3571 FRAME 0 command, and since it interfered with abbreviations for the
3573 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3575 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3576 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3577 some porting work by Martin Atzmueller)
3578 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3579 with a boatload of patches which he ported to SBCL. Now that those
3581 ** The system tries to make sure that its low-priority messages
3582 are prefixed by semicolons, to help people who like to use
3583 syntax highlighting in their ILISP buffer. (This patch
3584 was originally due to Raymond Toy.)
3585 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3586 patches originally written by Raymond Toy.
3587 ** The compiler understands coercion between single-value and
3588 multiple-VALUES type expressions better, getting rid of some very
3589 weird behavior, thanks to patches originally by Robert MacLachlan
3590 and Douglas Crosher.
3591 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3592 lambda lists, thanks to a patch originally by Pierre Mai.
3593 ** The system no longer bogusly warns about "abbreviated type
3595 ** The compiler gets less confused by inlining and RETURN-FROM,
3596 thanks to some patches originally by Tim Moore.
3597 ** The system no longer hangs when dumping circular lists to fasl
3598 files, thanks to a patch originally from Douglas Crosher.
3599 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3600 undefined function error.
3601 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3602 in structure slot initforms) for now, documented workaround instead:-|
3603 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3604 * fixed bug 5: added stubs for various Gray stream functions called
3605 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3606 installed, at least appropriate type errors are generated
3607 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3608 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3609 consistently in DEFMETHOD forms.
3610 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3611 it doesn't seem to affect SBCL after all
3612 * The C runtime system now builds with better optimization and many
3613 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3615 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3617 * The system is now under CVS at SourceForge (instead of the
3618 CVS repository on my home machine).
3619 * The new signal handling code has been tweaked to treat register
3620 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3621 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3622 downstream, e.g. in debug-int.lisp, has implicit dependencies
3623 on the unsignedness of integer representation of machine words,
3624 and that caused the system to bomb out with infinite regress
3625 when trying to recover from type errors involving signed values,
3626 e.g. (BUTLAST '(1 2 3) -1).)
3627 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3628 (This was one of the bugs Peter Van Eynde reported back in July.)
3629 * The system now uses code inspired by Colin Walters' O(N)
3630 implementation of MAP (from the cmucl-imp@cons.org mailing
3631 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3632 inline the MAP operation, and there is more than one
3633 sequence argument to the MAP call (so that it can't just
3634 do ETYPECASE once and for all based on the type of the
3635 single sequence argument). (The old non-inline implementation
3636 of the general M-argument sequence-of-length-N case required
3637 O(M*N*N) time when any of the sequence arguments were LISTs.)
3638 * The QUIT :UNIX-CODE keyword argument has been renamed to
3639 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3640 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3642 * Raymond Wiker's patches to port dynamic loading from Linux to
3643 FreeBSD have been added.
3644 * The BUGS file is now more nearly up to date, thanks in large part
3645 to Martin Atzmueller's review of it.
3646 * The debugger now flushes standard output streams before it begins
3647 its output ("debugger invoked" and so forth).
3648 * The core version number and fasl file version number have both
3649 been incremented, because of incompatible changes in the layout
3651 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3652 used to not be called for a saved Lisp image.)
3653 * Martin Atzmueller's version of a patch to fix a compiler crash,
3654 as posted on sbcl-devel 13 September 2000, has been installed.
3655 * Instead of installing Martin Atzmueller's patch for the
3656 compiler transform for SUBSEQ, I deleted the compiler transform,
3657 and transforms for some similar consing operations.
3658 * A bug in signal handling which kept TRACE from working on OpenBSD
3660 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3661 away by constant folding
3662 * The system now defines its address space constants in one place
3663 (in the Lisp sources), and propagates them automatically elsewhere
3664 (through GENESIS and the sbcl.h file). Therefore, patching the
3665 address map is less unnecessarily tedious and error-prone. The
3666 Lisp names of address space constants have also been systematized.
3667 * CVS tags like dollar-Header-dollar have been removed from
3668 the sources, because they have never saved me trouble and
3669 they've been source of trouble working with patches and other
3670 diff-related operations.
3671 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3672 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3674 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3676 * The system has been ported to OpenBSD.
3677 * The system now compiles with a simple "sh make.sh" on the systems
3678 that it's supported on. I.e., now you no longer need to tweak
3679 text in the target-features.lisp-expr and symlinks in src/runtime/
3680 by hand, the make.sh takes care of it for you.
3681 * The system is no longer so grossly inefficient when compiling code
3682 involving vectors implemented as general (not simple) vectors (VECTOR T),
3683 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3684 which dares to use the various sequence functions on non-simple
3685 vectors, takes less of a performance hit.
3686 * There is now a primitive type predicate VECTOR-T-P
3687 to test for the (VECTOR T) type, so that e.g.
3688 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3689 can now be compiled with some semblance of efficiency. (The old code
3690 turned the type declaration into a full call to %TYPEP at runtime!)
3691 * AREF on (VECTOR T) is still not fast, since it's still compiled
3692 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3693 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3694 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3695 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3696 without making full calls to SUBTYPEP at runtime.
3697 (Some analogous efficiency issues for non-simple vectors specialized to
3698 element types other than T, or for non-simple multidimensional arrays,
3699 have not been addressed. They could almost certainly be handled the
3700 same way if anyone is motivated to do so.)
3701 * The changes in array handling break binary compatibility, so
3702 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3703 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3704 instead of (VALUES T T).
3705 * By following the instructions that Dan Barlow posted to sbcl-devel
3706 on 2 July 2000, I was able to enable primitive dynamic object
3707 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3708 functionality is not implemented (since it calls ld to resolve
3709 library references automatically, requiring RUN-PROGRAM for its
3710 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3711 try to resolve library references) is now supported.
3712 * The system now flushes the standard output streams when it terminates,
3713 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3714 them at several other probably-convenient times, e.g. in each pass of
3715 the toplevel read-eval-print loop, and after evaluating a form given
3716 as an "--eval" command-line option. (These changes were motivated by a
3717 discussion of stream flushing issues on cmucl-imp in August 2000.)
3718 * The source transform for TYPEP of array types no longer assumes
3719 that an array whose element type is a not-yet-defined type
3720 is implemented as an array of T, but instead punts, so that the
3721 type will be interpreted at runtime.
3722 * There is now some support for cross-compiling in make.sh: each of
3723 the phases of make.sh has its own script. (This should be transparent
3724 to people doing ordinary, non-cross-compile builds.)
3725 * Since my laptop doesn't have hundreds of megabytes of memory like
3726 my desktop machine, I became more motivated to do some items on
3727 my to-do list in order to reduce the size of the system a little:
3728 ** Arrange for various needed-only-at-cold-init things to be
3729 uninterned after cold init. To support this, those things have
3730 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3731 symbols with such names are now uninterned after cold init).
3732 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3733 fasl files for cold load.
3734 ** Remove the old compiler structure pooling code (which used to
3735 be conditional on the target feature :SB-ALLOC) completely.
3736 ** Redo the representation of some data in cold init to be more compact.
3737 (I also looked into supporting byte compiled code at bootstrap time,
3738 which would probably reduce the size of the system a lot, but that
3739 looked too complicated, so I punted for now.)
3740 * The maximum signal nesting depth in the src/runtime/ support code has
3741 been reduced from 4096 to 256. (I don't know any reason for the very
3742 large old value. If the new smaller value turns out to break something,
3743 I'll probably just bump it back up.)
3744 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3746 * Many, many bugs reported by Peter Van Eynde have been added to
3747 the BUGS list; some have even been fixed.
3748 * While enabling dynamic object file loading, I tried to make the
3749 code easier to understand, renaming various functions and variables
3750 with less ambiguous names, and changing some function calling
3751 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3752 * While trying to figure out how to do the OpenBSD port, I tried to
3753 clean up some of the code in src/runtime/. In particular, I dropped
3754 support for non-POSIX signal handling, added various comments,
3755 tweaked the code to reduce the number of compilation warnings, and
3756 renamed some files to increase consistency.
3757 * To support the new automatic configuration functionality in make.sh,
3758 the source file target-features.lisp-expr has been replaced with the
3759 source file base-target-features.lisp-expr and the machine-generated
3760 file local-target-features.lisp-expr.
3761 * fixed a stupid quoting error in make.sh so that using CMU CL
3762 "lisp -batch" as cross-compilation host works again
3764 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3766 * DESCRIBE no longer tries to call itself recursively to describe
3767 bound/fbound values, so that it no longer fails on symbols which are
3768 bound to themselves (like keywords, T, and NIL).
3769 * DESCRIBE now works on generic functions.
3770 * The printer now prints less-screwed-up representations of closures
3771 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3772 * A private symbol is used instead of the :EMPTY keyword previously
3773 used to mark empty slots in hash tables. Thus
3774 (DEFVAR *HT* (MAKE-HASH-TABLE))
3775 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3776 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3777 now does what ANSI says that it should. (You can still get
3778 similar noncompliant behavior if bang on the hash table
3779 implementation with all the symbols you get back from
3780 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3781 This breaks binary compatibility, since tests for equality to
3782 :EMPTY are wired into things like the macroexpansion of
3783 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3785 * There's now a minimal placeholder implementation for CL:STEP,
3786 as required by ANSI.
3787 * An obscure bug in the interaction of the normal compiler, the byte
3788 compiler, inlining, and structure predicates has been patched
3789 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3790 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3792 * Missing ordinary arguments in a macro call are now detected even
3793 when the macro lambda list contains &KEY or &REST.
3794 * The debugger no longer complains about encountering the top of the
3795 stack when you type "FRAME 0" to explicitly instruct it to go to
3796 the top of the stack. And it now prints the frame you request even
3797 if it's the current frame (instead of saying "You are here.").
3798 * As specified by ANSI, the system now always prints keywords
3799 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3801 * The default initial SIZE of HASH-TABLEs is now smaller.
3802 * Type information from CLOS class dispatch is now propagated
3803 into DEFMETHOD bodies, so that e.g.
3804 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3806 is now basically equivalent to
3807 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3808 (DECLARE (TYPE SINGLE-FLOAT X))
3810 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3811 operation, without having to do run-time type dispatch.
3812 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3813 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3814 * Since I don't seem to be making big file reorganizations very often
3815 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3816 is overflowing my ability to conveniently back them up), I've finally
3817 checked the system into CVS. (The CVS repository is on my home system,
3818 not at SourceForge -- putting it on SourceForge might come later.)
3819 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3820 high-level GC-NOTIFY-FOO functions send their output. (There's
3821 still very little control of where low-level verbose GC functions
3822 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3823 less than it used to -- the GC-NOTIFY-FOO functions are now under
3824 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3825 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3826 in only one place in the source code, and propagates it automatically
3827 everywhere that it's needed. Thus e.g. when I bump the version from
3828 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3829 * The C source files now include boilerplate legalese and documentation
3830 at the head of each file (just as the Lisp source files already did).
3831 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3832 to his page will be replaced with a link to his new CLiki service.
3834 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3836 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3837 * The build process now looks for GNU make under the default name "gmake",
3838 instead of "make" as it used to. If GNU make is not available as "gmake"
3839 on your system, you can change this default behavior by setting the
3840 GNUMAKE environment variable.
3841 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3842 can build without error under CMU CL.
3844 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3846 * There is now a partial SBCL user manual (with some new text and some
3847 text cribbed from the CMU CL manual).
3848 * The beginnings of a profiler have been added (starting with the
3849 CMU CL profiler and simplifying and cleaning up). Eventually the
3850 main interface should be through the TRACE macro, but for now,
3851 it's still accessed through vaguely CMU-CL-style functions and macros
3852 exported from the package SB-PROFILE.
3853 * Some problems left over from porting CMU CL to the new
3854 cross-compilation bootstrap process have been cleaned up:
3855 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3856 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3857 part when getting rid of the compiler *BACKEND* stuff.)
3858 ** The value of *NULL-TYPE* was screwed up, because it was
3859 being initialized before the type system knew the final
3860 definition of the 'NULL type. This screwed up several key
3861 optimizations in the compiler, causing inefficiency in all sorts
3862 of places. (I found it because I wanted to understand why
3863 GET-INTERNAL-RUN-TIME was consing.)
3864 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3865 PROCLAIM FTYPE information. Unfortunately this broke binary
3866 compatibility again, since now the forms output by DEFGENERIC
3867 to refer to functions which didn't exist in 0.6.3.
3868 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3869 can use the new (as of 0.6.3) transform for SOME into MAP into
3871 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3872 values to (INTEGER 0 1000000), so that the time code will no longer
3873 occasionally get blown up by Linux returning 1000000 microseconds
3874 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3875 its output conform to the ANSI spec. (Alas, this makes its output
3876 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3877 * A full call to MAP NIL with a single sequence argument no longer conses.
3878 * fixes to problems pointed out by Martin Atzmueller:
3879 * The manual page no longer talks about multiprocessing as though
3880 it were currently supported.
3881 * The ILISP support patches have been removed from the distribution,
3882 because as of version 5.10.1, ILISP now supports SBCL without us
3883 having to maintain patches.
3884 * added a modified version of Raymond Toy's recent CMU CL patch for
3885 EQUALP comparison of HASH-TABLE
3887 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3889 * The system still can't cross-compile itself with
3890 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3891 entails), but at least it can compile more of itself that way
3892 than it used to be able to, and various buglets which were uncovered
3893 by trying to cross-compile itself that way have now been fixed.
3894 * This release breaks binary compatibility again. This time
3895 at least I've incremented the FASL file format version to 2, so that the
3896 problem can be detected reliably instead of just causing weird errors.
3897 * various new style warnings:
3898 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3899 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3900 ** using the lexical binding of a variable named in the *FOO* style
3901 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3902 as specified by ANSI.
3903 * *RANDOM-STATE* is no longer automatically initialized from
3904 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3905 default behavior of the system is to repeat its behavior every time
3906 it's run. If you'd like to change this behavior, you can always
3907 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3908 old convention there was no comparably easy way to get the system to
3909 repeat its behavior every time it was run.
3910 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3911 FTYPE declarations has been removed, in favor of their ANSI
3912 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3913 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3914 the types of their sequence arguments can be determined at compile time.
3915 This is done through a new open code expansion for MAP which eliminates
3916 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3917 argument types can be determined at compile time.
3918 * The optimizer now transforms COERCE into an identity operation when it
3919 can prove that the coerced object is already of the correct type. (This
3920 can be a win for machine generated code, including the output of other
3921 optimization transforms, such as the MAP transform above.)
3922 * Credit information has been moved from source file headers into CREDITS.
3923 * Source file headers have been made more standard.
3924 * The CASE macro now compiles without complaining even when it has
3927 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3929 * (Note that the way that the PCL macroexpansions were rewritten
3930 to accommodate the change in DEFGENERIC below breaks binary
3931 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3932 not run under sbcl-0.6.2. Once we get out of alpha releases,
3933 i.e. hit release 1.0.0, we'll probably try to maintain binary
3934 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3935 and sbcl-1.4.4. Until then, however, it might be fairly common
3936 for maintenance releases to break binary compatibility.)
3937 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3939 * The reporting of unhandled signals has been changed to print some
3940 explanatory text as well as the report form. (Previously only
3941 the report form was printed.)
3942 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3943 it defines, so that the compiler no longer issues undefined function
3944 warnings for compiled-but-not-yet-loaded generic functions.
3945 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3946 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3947 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3948 section of the ANSI spec are supported.
3949 * The "" nickname for the "KEYWORD" package has been removed.
3950 The reader still handles symbol tokens which begin with a package marker
3951 as keywords, but it doesn't expose its mechanism for doing so in the
3952 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3953 * The system now issues STYLE-WARNINGs for contradictory TYPE
3954 proclamations. (Warnings for contradictory FTYPE proclamations would
3955 be nice too, but those can't be done usefully unless the type system
3956 is made smarter about FUNCTION types.)
3957 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3958 systematized, so that "*target-*.lisp is supposed to exist only on the
3959 target and imply that there's a related file which exists on the
3960 host, and *host-*.lisp is supposed to exist only on the host and imply
3961 that there's a related file which exists on the target. This involves a
3962 lot of renaming. Hopefully the acute confusion caused by the renaming
3963 will be justified by the reduction in chronic confusion..
3964 ** runtime-type.lisp -> early-target-type.lisp
3965 ** target-type.lisp -> late-target-type.lisp
3966 ** early-host-format.lisp -> early-format.lisp
3967 ** late-host-format.lisp -> late-format.lisp
3968 ** host-error.lisp -> misc-error.lisp
3969 ** early-error.lisp -> early-target-error.lisp
3970 ** late-error.lisp -> late-target-error.lisp
3971 ** host-defboot.lisp -> early-defboot.lisp
3972 ** code/misc.lisp -> code/target-misc.lisp
3973 ** code/host-misc.lisp -> code/misc.lisp
3974 ** code/numbers.lisp -> code/target-numbers.lisp
3975 ** code/early-numbers.lisp -> numbers.lisp
3976 ** early-host-type.lisp -> early-type.lisp
3977 ** late-host-type.lisp -> late-type.lisp
3978 ** host-typep.lisp -> typep.lisp
3979 ** load.lisp -> target-load.lisp
3980 ** host-load.lisp -> load.lisp
3981 ** host-disassem.lisp -> disassem.lisp
3982 ** host-insts.lisp -> insts.lisp
3983 ** byte-comp.lisp -> target-byte-comp.lisp
3984 ** host-byte-comp.lisp -> byte-comp.lisp
3985 ** host-signal.lisp -> signal.lisp
3986 ** host-defstruct.lisp -> defstruct.lisp
3987 ** late-target-type.lisp -> deftypes-for-target.lisp
3988 Furthermore, several other previously target-only files foo.lisp (e.g.
3989 hash-table.lisp and random.lisp) have been split into a target-and-host
3990 foo.lisp file and a target-only target-foo.lisp file, with their key type
3991 definitions in the target-and-host part, so that the cross-compiler will
3992 know more about target types.
3993 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3994 gone away. In their place are various *BACKEND-FOO* variables
3995 corresponding to the slots of the old structure.
3996 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3997 into the target SBCL has been fixed.
3998 * The chill.lisp system for loading cold code into a running SBCL
4000 * Support for the CMU CL "scavenger hook" extension has been removed.
4001 (It was undocumented and unused in the CMU CL sources that SBCL was
4002 derived from, and stale in sbcl-0.6.1.)
4003 * Various errors in the cross-compiler type system were detected
4004 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4005 (enabling various consistency checks). Many of them were fixed,
4006 but some hard problems remain, so the compiler is back to
4007 running without *TYPE-SYSTEM-INITIALIZED* for now.
4008 * As part of the cross-compiler type system cleanup, I implemented
4009 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4010 * I have started adding UNCROSS calls throughout the type system
4011 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4012 on cross-compiler input files will be able to go away, and various
4014 * CONSTANTP now returns true for quoted forms (as explicitly required
4017 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4019 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4020 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4021 (MAKE-LIST -1). In the longer term, it ought to become true
4022 that declarations are assertions even at SAFETY 1. For now, it's not
4023 quite true even at SAFETY 3, but it's at least more nearly true..
4024 (Note that this change seems to increases the size of the system by
4025 O(5%) and to decrease the speed of the compiler by 20% or more.)
4026 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4027 for the problem of printing dozens of lines of distracting information
4028 about low-level system machinery as part of the top stack frame
4029 on entry to the debugger when an undefined function was called.
4030 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4031 is not reset to COMMON-LISP-USER.
4032 * Compilation of stuff related to dyncount.lisp has been made conditional
4033 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4034 smaller. The various dyncount-related symbols have been moved into
4035 a new "SB-DYNCOUNT" package.
4036 * tty-inspect.lisp has been renamed to inspect.lisp.
4037 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4038 feature has gone away. (When we eventually port to other flavors of
4039 libc and/or Unix, we'll try to make the differences between flavors
4040 invisible at the user level.)
4041 * Various other *FEATURES* tags, and/or their associated conditionals,
4042 have been removed if obsolescent, or given better documentation, or
4043 sometimes given more-mnemonic names.
4045 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4047 * tidied up "make.sh" script
4048 * tidied up system directory structure
4049 * better "clean.sh" behavior
4050 * added doc/FOR-CMUCL-DEVELOPERS
4051 * many many small tweaks to output format, e.g. removing possibly-confusing
4052 trailing #\. character in DESCRIBE-INSTANCE
4053 * (EQUALP #\A 'A) no longer signals an error.
4054 * new hashing code, including EQUALP hashing
4055 * tidied up Lisp initialization and toplevel
4056 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4057 * command line argument processing
4058 * added POSIX-GETENV function to deal with Unix-ish environment variables
4059 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4060 terminating SBCL on EOF
4061 * non-verbose GC by default
4062 * There is no more "sbcl" shell script; the sbcl file is now the C
4063 runtime executable (just like CMU CL).
4064 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4066 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4067 symbol and package information as Lisp native symbol and package objects
4068 instead of strings naming symbols and strings naming packages. This way,
4069 whenever packages are renamed (as in warm init), debug information is
4070 transformed along with everything else.
4071 * tweaked the optimization policy declarations which control the building
4072 of SBCL itself. Now, among other things, the system no longer saves
4073 source location debugging information. (This helps two problems at once
4074 by reducing SBCL size and by keeping SBCL from trying to look for its
4075 sources -- which may not exist -- when reporting errors.)
4076 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4077 debugging and testing purposes
4078 * cleaned up printing, making the printer call PRINT-OBJECT for
4079 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4080 methods, giving nearly-ANSI behavior
4081 * converted almost all special variables to use *FOO* naming convention
4082 * deleted PARSE-TIME functionality, since it can be done portably
4083 * moved some files out of cold init into warm init
4084 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4086 * regularized formatting of source files
4087 * added an install.sh script
4088 * fixed ridiculous memory usage of cross-compiler by making
4089 compiler/alloc.lisp not try to do pooling unless it can hook
4090 itself into the GC of the cross-compilation host. Now the system
4091 builds nicely on my old laptop.
4092 * added :SB-ALLOC in target-features.lisp-expr
4093 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4094 was not implemented)
4095 * re-did condition handling and note reporting in the compiler. Notes
4096 are no longer handled by signalling conditions. Style warnings
4097 and warnings are handled more correctly and reported in such a way
4098 that it's easy to find one or the other in your output (so that you
4099 can e.g. figure out which of many problems caused COMPILE-FILE to
4101 * changed the severity of several compiler warnings from full WARNING
4102 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4103 compiler note reporting so that it doesn't use the condition system
4104 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4105 COMPILE-FILE command)
4106 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4107 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4108 no longer supported, since their implementation was deeply intertwingled
4109 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4110 * removed bogus "support" for compiler macros named (SETF FOO), and
4111 removed the compiler macro for SETF INFO (but only after making a fool
4112 of myself on the cmucl-imp mailing list by posting a bogus patch for
4113 DEFINE-COMPILER-MACRO..)
4114 * Compiled files containing forms which have side effects on the Lisp
4115 reader (such as DEFPACKAGE forms) are now handled more correctly.
4116 (Compiler queuing of top level lambdas has been suppressed by setting
4117 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4118 * deleted various currently-unused source files, e.g. gengc.lisp. They
4119 may be added back at some point e.g. when porting to other architectures,
4120 but until they are it's distracting to distribute them and to try to
4122 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4123 there were so many of them they're just distractions, and UNCROSS is
4124 known to be able to handle the current sources
4125 * moved PROFILE functionality into TRACE, so that it will be clear
4126 how the wrapping and unwrapping of functions when you profile them
4127 interacts with the wrapping and unwrapping of functions when you
4128 trace them. (Actually, the functionality isn't there yet, but at least
4129 the interface specification is there. Hopefully, the functionality will
4130 arrive with some future maintenance release.)
4131 * removed host-oops.lisp
4132 * changed signature of QUIT function to allow UNIX-CODE argument
4133 * fixed READ-SEQUENCE bug
4134 * tweaked verbose GC output so that it looks more like the progress
4135 output that ANSI specifies for functions like LOAD
4136 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4137 * added <http://sbcl.sourceforge.com> to the banner information printed by