1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
3 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
4 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
5 produce more efficient code on the x86 if the compiler can determine
6 that (+ INDEX <constant>) does not require a bounds check and FOO
7 has an element type at least 8 bits wide.
8 * bug fix: references to undefined variables are handled the same way
9 in toplevel forms as in the normal compiler
10 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
12 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
13 (thanks to Jon Buller)
14 * improvement: faster compilation times for complex functions
16 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
17 * improvement: experimental support for mach exception handling on
18 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
20 * improvement: support for GBK external format.
21 (thanks to Chun Tian (binghe))
22 * improvement: the debugger now displays variables that have been closed
23 over, in code compiled with (DEBUG 2) or higher
24 * improvement: support for executable cores on NetBSD (thanks to
26 * new feature: added a RESTART-FRAME debugger command
27 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
28 be used to provide an implementation for FILE-POSITION on Gray streams
29 (thanks to Eric Marsden)
30 * new feature: add syslog, openlog and closelog support to SB-POSIX
31 (thanks to Richard Kreuter)
32 * optimization: the function call overhead in code compiled with
33 a high DEBUG optimization setting is significantly
34 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
35 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
36 (thanks to Tony Martinez)
37 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
38 evaluated once (reported by Kevin Reid)
39 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
40 works again (thanks to Kilian Sprotte)
41 * bug fix: an error is signaled for tagbodies with duplicate tags
42 (thanks to Stephen Wilson)
43 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
44 * bug fix: Win32 port can now handle foreign code unwinding Lisp
45 stack frames from alien callbacks.
46 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
47 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
48 65k pages (thanks to David Woodhouse)
49 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
51 changes in sbcl-1.0.1 relative to sbcl-1.0:
52 * new platform: FreeBSD/x86-64, including support for threading.
53 * new feature: the compiler stores cross-referencing information
54 abount function calls (who-calls), macroexpansion (who-macroexpands)
55 and special variables (who-binds, who-sets, who-references) for code
56 compiled with (< SPACE 3). This information is available through the
57 sb-introspect contrib.
58 * new feature: users may subclass SEQUENCE, and have instances of
59 these classes interoperate with standard Common Lisp functions if
60 a number of methods are defined. (This feature is experimental
61 and the interface subject to change based on feedback from SBCL
62 users and the general community)
63 * improvement: sb-sprof traces call stacks to an arbitrary depth on
64 x86/x86-64, rather than the previous fixed depth of 8
65 * improvement: another pthread back-end of mutex "pthread-futex".
66 * bug fix: non-ascii command-line arguments are processed correctly
67 (thanks to Yaroslav Kavenchuk)
68 * bug fix: non-required arguments were not passed correctly when a method
69 defined using DEFMETHOD was called from a mop-generated method using
70 CALL-NEXT-METHOD (reported by Pascal Costanza)
71 * bug fix: recursion is now permitted in accessors through
72 SLOT-UNBOUND. (reported by Pascal Costanza)
73 * bug fix: an error was signaled at startup if the HOME environment
74 variable was defined, but had an empty value (reported by Peter Van Eynde)
75 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
76 signal a PROGRAM-ERROR, not a BUG.
77 * bug fix: SB-POSIX:READDIR works when built with large file support.
78 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
80 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
81 for better automatic indentation support. (thanks to Matt Pillsbury)
82 * optimization: loading generic functions no longer takes O(n^2) time,
83 proportional to the amount of methods in the generic function
84 (reported by Todd Sabin and Jeremy Brown)
85 * optimization: the FIND and POSITION family of sequence functions
86 are significantly faster on arrays whose element types have been
88 * improvements to the Windows port:
89 ** Intermittent heap corruption problems have been fixed. (thanks
90 to Alastair Bridgewater)
91 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
93 ** Lisp is able to unwind foreign exception frames from alien
94 callbacks. (thanks to Alastair Bridgewater)
96 changes in sbcl-1.0 relative to sbcl-0.9.18:
97 * improvement: experimental support for threading on FreeBSD/x86.
98 (thanks to NIIMI Satoshi)
99 * improvement: runtime option --dynamic-space-size can be used
100 to set the size of the dynamic space reserved on startup.
101 * improvement: floating point modes in effect are now saved in
102 core, and restored on startup.
103 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
104 startup, not time since first call to GET-INTERNAL-REAL-TIME.
105 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
106 threads are not running after *SAVE-HOOKS* have run.
107 * improvement: writes to CLOS instance slots are type-checked in code
108 compiled with (SAFETY 3)
109 * improvement: floating-point exception handling on FreeBSD (thanks to
111 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
112 (thanks to Zach Beane)
113 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
115 * improvement: added support for the Shift-JIS external format.
116 (contributed by NIIMI Satoshi)
117 * improvement: callbacks are supported on Linux/PPC. (thanks to
119 * bug fix: compiler bug triggered by a (non-standard) VALUES
120 declaration in a LET* was fixed. (reported by Kaersten Poeck)
121 * bug fix: file compiler no longer confuses validated and already
122 dumped structurres. (reported by Kaersten Poeck)
123 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
124 fill-pointer signals a type-error as required. (thanks to
126 * bug fix: disassemly of funcallable instances works.
127 * bug fix: single stepping on PPC.
128 * bug fix: fix thread-safety problems in the type system (generally
129 manifesting as nonsensical errors like "STRING is a bad type specifier
130 for sequences" or "The value 1 is not of type FIXNUM")
131 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
132 (reported by Josip Gracin)
133 * bug fix: an error is signaled for attempts to displace arrays with
134 incompatible element types (thanks to Mario Mommer)
135 * bug fix: more correct handling of wide characters in debug info
136 (bug reported by Attila Lendvai and fixed by Juho Snellman)
137 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
138 and don't cause extra consing
139 * optimization: MAP and MAP-INTO are significantly faster on vectors
140 whose elements types have been declared.
141 * Improvements to SB-SPROF:
142 ** Support for allocation profiling
143 ** Reduced profiling overhead, especially for long profiling runs
144 * Improvements to the Windows port:
145 ** floating point exceptions are now reported correctly.
146 ** stack exhaustion detection works partially.
147 ** more accurate GET-INTERNAL-REAL-TIME.
148 ** better RUN-PROGRAM behaviour in case of errors in the child
150 ** PROBE-FILE now simplifies pathnames correctly.
151 ** DIRECTORY now works correctly with :WILD-INFERIORS.
152 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
153 1st 1970 (although time timezone data will be unreliable).
154 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
156 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
157 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
158 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
160 * bug fix: two potential GC deadlocks affecting threaded builds.
161 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
162 returns T (reported by Anton Kazennikov)
163 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
164 with non-variable places
165 * bug fix: remove a race condition in the setting of
166 funcallable-instance functions, this should make threaded CLOS
167 code more stable against memory faults.
168 * bug fix: corruption of specials when unbinding is interrupted by an
169 asynchronous unwind (reported by Hannu Koivisto)
170 * improvement: the debugger will now also display local variables that
171 are only used once, for code compiled with a DEBUG optimization quality
174 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
175 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
176 * incompatible change: External-format support for FFI calls. The
177 SB-ALIEN:C-STRING no longer implies an ASCII
178 external-format. Instead the string is subject to external-format
179 conversion. Additionally return values of type SB-ALIEN:C-STRING
180 are by default of type (SIMPLE-ARRAY CHARACTER), not
181 SIMPLE-BASE-STRING. For an alien type that behaves like the old
182 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
183 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
184 * incompatible change: SB-EXT package no longer contains the
185 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
186 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
188 * incompatible change: the single-stepper is no longer available
189 on Alpha, Mips and Sparc platforms.
190 * minor incompatible change: the direct superclasses of
191 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
192 not (STANDARD-OBJECT FUNCTION). This makes the
193 class-precedence-lists of GENERIC-FUNCTION and
194 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
196 * minor incompatible change: the default stream external format on
197 non-unicode SBCL is no longer detected from the locale, but is always
199 * new feature: Add a version of evaluator that uses an interpreter instead
200 of the compiler. EVAL still uses the compiler by default, to switch it
201 to use the interpreter, set the value of the variable
202 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
203 * minor incompatible change: the single-stepper REPL has been merged
204 with the normal debugger (see the "Stepping" heading of the debugger help
205 for more details). The debugger command STEP will no longer switch
206 to the single-stepper REPL.
207 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
208 for a type now works.
209 * bug fix: loading of default sysinit file works. (thanks to Leonid
211 * bug fix: better detection of circularities in the file-compiler.
212 (reported by Marco Monteiro)
213 * bug fix: the CL pathname functions now work with files that have
214 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
215 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
216 systems with Unix98 pty semantics.
217 * bug fix: ASDF-INSTALL will now work with bsd tar.
218 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
220 * bug fix: timers expiring in dead threads no longer cause a
221 type-error (reported by Paul "Nonny Mouse").
222 * bug fix: thanks to more lightweight single-stepper instrumentation,
223 code compiled with (DEBUG 3) will compile and execute significantly faster,
224 and will have more accurate type-inferencing than before
225 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
226 whose bindings are modified
227 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
228 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
229 ** replace dummy implementations of CL:MACHINE-INSTANCE and
230 CL:SOFTWARE-VERSION with proper version
232 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
233 * feature: implemented the READER-METHOD-CLASS and
234 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
235 as specified by AMOP.
236 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
238 * optimization: faster LOGCOUNT implementation on x86 and x86-64
239 (thanks to Lutz Euler)
240 * optimization: hashing of general arrays and vectors has been
241 improved. (reported by Any Fingerhut)
242 * enhancement: SB-INTROSPECT is now able to find definitions of
243 profiled functions. (thanks to Troels Henriksen)
244 * enhancement: compiler-macro expansion applies now to FUNCALL forms
245 as well, allowing compiler-macros for SETF-functions to expand.
246 * enhancement: step-instrumentation no longer wraps known
247 single-value functions with multiple-value context, allowing
248 better type inference.
249 * fixed bug #337: use of MAKE-METHOD in method combination now works
250 even in the presence of user-defined method classes. (reported by
251 Bruno Haible and Pascal Costanza)
252 * fixed bug #339(c): if there are applicable methods not part of any
253 long-form method-combination group, call INVALID-METHOD-ERROR.
254 (reported by Bruno Haible)
255 * fixed bug #361: the :FUNCTION initarg in the protocol for
256 initialization of methods can now be used to override
257 internally-produced optimized functions. (reported by Bruno
259 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
260 system-provided lambda expression no longer cause warnings about
261 unbound #:|pv-table| symbols.
262 * bug fix: improved the handling of type declarations and the
263 detection of violations for keyword arguments with non-constant
265 * bug fix: potentially erronous calls to PATHNAME and
266 MERGE-PATHNAMES were being flushed in some cases.
267 (reported by Richard Kreuter)
268 * bug fix: compiled calls to TYPEP were mishandling obsolete
269 instances. (reported by James Bielman and Attila Lendvai)
270 * bug fix: format strings with ~> without matching ~< no longer
271 trigger an AVER, but signal an understandable error instead.
272 (reported by Antonio Martinez)
273 * bug fix: specifying an output-file without a pathname-type for
274 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
275 of input-file instead of "fasl". (reported by Robert Dodier)
276 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
277 (reported by James Y Knight).
278 * bug fix: compiler-macros-function did not consider the environment
279 argument for shadowing by local functions.
280 * bug fix: compiler-macros expansion was inhibited by local INLINE
282 * bug fix: inline expansions of known functions were subject to
283 step-instrumentation in high debug policies, leading to problems
285 * bug fix: compiler failed to differentiate between different CONS
287 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
288 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
289 * bug fix: #368: incorrect use of expressed vs. upgraded array
291 * bug fix: #306a: more precise unions of array types.
292 * thread-safety improvements:
293 ** CONDITION-WAIT could return early on Linux, if the thread was
294 interrupted and subsequently continued with SIGCONT.
295 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
297 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
298 * added support for the ucs-2 external format. (contributed by Ivan
300 * minor incompatible change: pretty printing of objects of type
301 (cons symbol) is, in the default pprint-dispatch-table, now
302 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
304 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
305 called later in a class's lifetime, possibly as late as when the
306 first instance of the class is created. Previously,
307 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
308 class became finalizeable.
309 * fixed bug: FILE-POSITION sometimes returned inconsistent results
310 for multibyte external-format streams. (thanks to "vbzoli")
311 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
312 with :ALLOCATION :CLASS inherited from superclasses of the
314 * fixed bug: anonymous classes can now be created using the :NAME
315 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
316 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
317 * fixed bug: core-files saved with :EXECUTABLE T can again be
318 executed when SBCL_HOME isn't set. (reported by James Knight)
319 * fixed bug: toplevel LOCALLY forms with declarations could
320 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
321 * fixed bug: printing from several different threads using different
322 values of *print-case* could cause invalid output, due to
323 some internal special variables of the printer not being bound
324 thread-locally (reported by Max Mikhanosha)
325 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
326 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
327 (reported by Pascal Costanza)
328 * fixed bug: FINALIZE-INHERITANCE is called from
329 REINITIALIZE-INSTANCE on classes when the class has previously
330 been finalized, as required by AMOP.
331 * minor code generation optimizations:
332 ** better register allocation in CLOS dispatching functions
333 ** overflow detection when coercing signed bytes to fixnums on x86-64
334 ** is now implemented with one IMUL instruction instead of three shifts
335 ** more efficient bit-vector access on x86 and x86-64
336 ** more efficient access to raw structure slots on x86 and x86-64
337 * fixed some bugs revealed by Paul Dietz' test suite:
338 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
341 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
342 * feature: thread support on Solaris/x86, and experimental thread support
344 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
346 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
347 default method for SLOT-UNBOUND.
348 * minor incompatible change: prevent the user from specializing the
349 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
350 somewhat counter to the intent of the protocol, I (CSR) think, and
351 additionally it just doesn't work in SBCL as currently
352 implemented, thanks to optimizations (that are always valid for
353 the other three SLOT-VALUEish functions, but not for the setter).
354 * minor incompatible change: the :SB-LDB feature is now enabled by
355 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
356 the low-level debugger.
357 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
358 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
359 down to an SB-SYS:FD-STREAM, allowing direct communication in
360 more cases. Temporary files and pipes are still used as a fallback
362 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
363 on the non-reentrant C functions localtime(3) and gmtime(3).
364 * bug fix: global optimization policy was not visible in LOCALLY and
366 * bug fix: class objects can be used as specializers in methods.
367 (reported by Pascal Costanza)
368 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
369 with a type error. (reported by blitz_ on #lisp)
370 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
371 failed with a type error. (reported by Pascal Bourguignon)
372 * bug fix: merging pathnames against defaults with :DIRECTORY
373 starting with '(:RELATIVE :BACK) should preserve the :BACK.
374 (reported by James Y Knight)
375 * bug fix: saving large (>2GB) cores on x86-64 now works
376 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
377 constant index and a (simple-array (signed-byte 32)) array
378 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
379 error (patch by Robert J. Macomber)
380 * feature: TIME now displays the amount of run-time spent in GC
381 * bug fix: The debugger now does a better job of respecting
382 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
383 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
384 workaround for bug 403.)
385 * bug fix: floating point exception handling now works on FreeBSD
386 (thanks to NIIMI Satoshi)
387 * fixed some bugs revealed by Paul Dietz' test suite:
388 ** MISC.641: LET-conversion were not supposed to work in late
390 * improvements to the Win32/x86 port:
391 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
392 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
394 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
395 * new feature: source path information is generated for macro-expansion
396 errors for use in IDE's like Slime (thanks to Helmut Eller)
397 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
398 * bug fix: compiler does not loop forever on an invalid type in
400 * improvement: compilation of most CLOS applications is significantly
402 * optimization: added a limited bytecode compiler for simple toplevel
403 forms, speeding up compilation and FASL loading
404 * bug fix: the statistical profiler now properly distinguishes anonymous
407 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
408 * minor incompatible change: in sbcl-0.9.11 (but not earlier
409 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
410 into subclasses of the finalized class. Now user calls to
411 FINALIZE-INHERITANCE finalize just the one class, and calls by the
412 system return before any subclasses are finalized.
413 * minor incompatible change: The reader no longer ignores errors
414 regarding non-existent packages in #+ and #- feature tests.
415 * new feature: command line options --no-sysinit, --no-userinit to
416 inhibit loading the corresponding init files
417 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
418 for sending data through UDP sockets (thanks to François-René Rideau)
419 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
420 error is signalled instead (thanks to François-René Rideau)
421 * minor incompatible change: Error signalling behaviour of lexical
422 operations violating package locks has changed slightly. Refer to
423 documentation on package locks for details.
424 * enhancement: EVAL can process IF-expressions without resorting to the
426 * bug fix: Some operations involving SIN, COS, or TAN weren't being
427 constant-folded properly due to missing out-of-line functions.
428 (reported by Mika Pihlajamäki)
429 * bug fix: LISTEN sometimes returned T even in cases where no data was
430 immediately available from the stream
431 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
432 were reversed. (reported by Levente Mészáros)
433 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
434 when given a vector displaced to an adjustable array. (reported
436 * bug fix: derivation of float boundaries from numbers outside the
437 appropriate float range (reported by John Wiseman)
438 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
440 * bug fix: precomputing cache entries for generic functions with
441 some subclasses of specializers as yet invalid does not attempt to
442 fill a cache line with a negative offset. (reported by Levente
444 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
445 allow condition class objects as arguments to DOCUMENTATION and
446 (SETF DOCUMENTATION); only find and set documentation for
447 structure names for the STRUCTURE doc-type. (suggested by Gary
449 * improvements to the Win32/x86 port:
450 ** (user-homedir-pathname) and default initialization file
451 locations now know about the user's "Documents and Settings"
452 directory (thanks to Yaroslav Kavenchuk)
453 ** run-program is implemented (thanks to Mike Thomas)
454 ** sockets support (thanks to Timothy Ritchey)
455 ** better backtrace support (thanks to Alastair Bridgewater)
456 ** sb-grovel supported
457 ** asdf-install and sb-posix work somewhat
458 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
459 * improvements to the Solaris/x86 port:
460 ** works on Solaris 11/Solaris Express
461 ** floating-point exception handling support
462 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
463 * fixed some bugs revealed by Paul Dietz' test suite:
464 ** REMOVE-METHOD returns its generic function argument even when
465 the method is not one of the generic functions' methods.
466 ** objects declared as MEMBER types can be admissible arguments to
468 ** printing characters should simply be printed by the FORMAT ~:C
470 ** compiler failure when compiling functions with hairy constant
471 defaults for optional parameters.
472 ** compiler produces wrong code when MAYBE-INLINE-expanding a
473 function, which is already optimized.
475 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
476 * new platform: experimental support for SBCL x86/Darwin, including
477 MacOS X 10.4.5 on Intel.
478 * new feature: Unicode character names are now known to the system
479 (through CHAR-NAME and NAME-CHAR).
480 * new feature: the filesystem location of SBCL's core file is
481 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
482 * minor incompatible change: the contrib modules SB-POSIX and
483 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
484 this change is to make it easier to distribute
485 location-independent binaries.
486 * bug fix: as implied by AMOP, standardized classes no longer have
487 slots named by external symbols of public packages. (reported by
489 * optimization: calling functions via constant symbols -- as in
490 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
491 via the function object as in (FUNCALL #'FOO).
492 * optimization: on x86 and x86-64, the calling convention for the
493 case of dealing with an unknown number of values has been altered
494 to be friendlier to the prediction heuristics implemented,
495 particularly with reference to CALL and RET pairing. (thanks to
496 Alastair Bridgewater)
497 * optimization: on x86, the code for access to thread-local symbol
498 values and binding/unbinding of thread-local symbols is smaller.
499 (thanks to Alastair Bridgewater)
500 * enhancement: CONSTANTP is now able to determine constantness of
501 more complex forms, including calls to constant-foldable standardized
502 functions and some special forms beyond QUOTE.
503 * fixed bug: occasional GC crashes on Solaris/x86
504 * optimization: x86-64 supports stack allocation of results of simple
505 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
506 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
507 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
508 (thanks to James Knight)
509 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
510 by the standard (thanks to Stephen Compall)
512 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
513 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
514 be used for bundling the runtime and the core file into one
515 executable binary. This feature is not currently supported on all SBCL
516 platforms. (thanks to James Bielman and NIIMI Satoshi)
517 * new feature: a generational or ephemeral garbage collector is now
518 the default on the PowerPC platform (both Linux and Darwin). The
519 old Cheney (stop and copy) collector is a build-time option.
520 (thanks to Cyrus Harmon, after Raymond Toy)
521 * minor incompatible change: the method by which SBCL finds its
522 contributed modules has changed; it no longer relies on symbolic
523 links from an $SBCL_HOME/systems directory, but searches directly
524 in the subdirectories of $SBCL_HOME.
525 * enhancement: the dynamic heap size on the Linux/PPC platform is
526 markedly larger, even using the older Cheney garbage collector.
527 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
528 certain complicated string types. (reported by Gary King)
529 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
530 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
531 * fixed bug: DOCUMENTATION on structures no longer signals an
532 error if no structure type of the right name exists. (reported by
534 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
535 applicable non-standard primary or :AROUND method on
536 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
537 no longer cause unbound variable errors. (reported by Kevin Reid)
538 * fixed bug: metacircle resolution in cases where methods have slots
539 added before the slots from STANDARD-METHOD. (reported by Jean
541 * fixed bug: the Power PC assembly code for calling into lisp
542 sometimes computed its return address wrongly (depending on the
543 15th bit of the address, fixed at link time). This used to
544 manifest itself in a segmentation violation while building PCL.
545 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
546 many others over the years)
547 * fixed bug: ROOM no longer reports silly numbers for stack usage if
548 the stack pointer (treated as a signed integer) is negative.
549 (thanks to Peter van Eynde)
551 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
552 * new platform: experimental support for the Windows operating
553 system has been added. (thanks to Alastair Bridgewater)
554 * fixed several bugs in and robustified the PPC FFI (including
555 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
556 * bug fix: allow non-simple string symbol names (reported by Paul
558 * bug fix: interrupt handling on NetBSD (thanks to Richard M
560 * bug fix: saving a core corrupted callbacks on x86/x86-64
561 * bug fix: closed a loophole in metacircularity detection and
562 grounding in the PCL implementation of CLOS.
563 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
564 (reported by tomppa on #lisp)
565 * bug fix: clear the direction flag on Lisp -> C transitions, as
566 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
567 (reported by Andrej Grozin and Hendrik Maryns)
568 * optimization: major improvements to GC efficiency on GENCGC platforms
569 * optimization: faster implementation of EQUAL
570 * optimization: emit more efficient opcodes for some common
571 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
572 * optimization: several other minor code-generation improvements on x86-64
573 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
575 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
576 * minor incompatible change: (SETF CLASS-NAME) and (SETF
577 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
578 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
579 by ANSI as a generic function, and for consistency (SETF
580 GENERIC-FUNCTION-NAME) is treated likewise.
581 * fixed bug #233.b: make constraint propagation notice when a variable
582 value is changed after it is referenced but before it is used
583 * fixed bug #296: no more arbitrary behaviour from filenames with
584 odd characters as --load arguments. (reported by Adam Warner)
585 * fixed bug #390: :CHARACTER-SET pathname components now work as
586 expected. (reported by Tim Daly Jr)
587 * fixed bug #391: complicated :TYPE intersections in slot
588 definitions no longer cause an error in PCL internals.
589 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
590 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
591 returning the number of octets which would be written to the
592 file-stream. (thanks to Robert J. Macomber)
593 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
594 arguments to the internal stream functions. (thanks to David
596 * bug fix: stack exhaustion is now handled on the x86/SunOS
597 platform. (thanks to Daisuke Homma)
598 * bug fix: more accurate ROOM results on GENCGC platforms
599 * optimization: improved type inference for arithmetic-for
600 index variables in LOOP
601 * optimization: faster floating-point SQRT on x86-64
602 * fixed some bugs revealed by Paul Dietz' test suite:
603 ** DOCUMENTATION returns NIL instead of "" for method combinations
604 that don't have a docstring
606 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
607 * minor incompatible change: (SETF CLASS-NAME) and (SETF
608 GENERIC-FUNCTION-NAME) are no longer generic functions, and
609 therefore are not customizeable by user code (as seems to be at
610 least permitted and maybe required by AMOP). As a consolation,
611 however, the SBCL implementation of these functions now calls
612 REINITIALIZE-INSTANCE as specified by AMOP.
613 * bug fix: slot-definition documentation is propagated as per ANSI
614 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
615 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
616 argument, as do eslotd initialization methods. (from Pascal
617 Costanza's "Closer" project)
618 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
619 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
621 * bug fix: it is now possible to have more than one subclass of
622 STANDARD-GENERIC-FUNCTION without causing stack overflow.
623 (reported by Bruno Haible, Pascal Costanza and others)
624 * bug fix: the dependent update protocol now works for generic
625 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
627 * bug fix: condition-class instances corresponding to
628 DEFINE-CONDITION forms are now created eagerly. (reported by
629 Kalle Olavi Niemitalo on comp.lang.lisp)
630 * bug fix: floating point printing is more accurate in some
631 circumstances. (thanks to Simon Alexander)
632 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
633 merged with *DEFAULT-PATHNAME-DEFAULTS*.
634 * bug fix: callbacks on OS X now preserve stack-alignment.
635 (thanks to Cyrus Harmon)
636 * enhancement: the x86-64 disassembler is much better at
637 disassembling SSE instructions. (thanks to Lutz Euler)
638 * enhancement: CHECK-TYPE on a variable now allows type-inference
639 (thanks to Vincent Arkesteijn)
640 * optimization: improved performance of EUC-JP external format.
641 (thanks to NIIMI Satoshi)
642 * optimization: performance improvements to IO on file streams of
643 :ELEMENT-TYPE CHARACTER
644 * optimization: much faster memory allocation on x86-64
645 * optimization: faster garbage collections (latency and throughput)
647 * optimization: faster bignum subtraction and division on x86 and x86-64
648 (thanks to Lutz Euler)
649 * optimization: more accurate type-inference for loops that use a
650 floating point index variable or a negative step.
652 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
653 * bug fix: add a workaround to SBCL looping infinitely at startup on
654 Linux kernels with apparently buggy implementations of personality().
655 (thanks to Svein Ove Aas)
656 * bug fix: Unicode symbols are correctly printed in LDB backtraces
657 (thanks to David Lichteblau)
658 * bug fix: local bindings shadow global symbol macros for
659 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
660 * optimization: non-open coded uses of numeric comparison operators
661 (e.g. >) no longer cons when called with more than one parameter
662 on platforms supporting dynamic-extent allocation.
663 * enhancement: saving cores with foreign code loaded is now supported
664 on MIPS/Linux in addition to the previously supported platforms.
665 * bug fix: FIND-RESTART now tests for activity, not applicability when given
666 a restart object as identifier. (reported by Helmut Eller for CMUCL)
667 * bug fix: division by zero in sb-sprof when no samples were collected
668 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
669 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
670 (thanks to Alexander Kjeldaas)
671 * new feature: ignoring the return values of destructive functions
672 like NREVERSE signals a compile-time style-warning.
673 (thanks to Kevin Reid)
674 * new feature: the alignment of alien structure fields can be
675 explicitly specified. (thanks to Cyrus Harmon)
676 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
677 and dump core on SIGQUIT
679 ** incompatible change: threads do not inherit values of specials
680 from their parents (see manual)
681 ** bug fix: threads stacks belonging to dead threads are freed by the
682 next exiting thread, no need to gc to collect thread stacks anymore
683 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
684 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
685 ** bug fix: dynamic variable and thread start related gc lossage
686 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
688 * fixed some bugs revealed by Paul Dietz' test suite:
689 ** SUBTYPEP is slightly more accurate on heinously complicated
690 CONS types where some of the members have uncertain (in the
691 NIL, NIL sense) type relationships to each other.
693 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
694 no longer purified unless :PURIFY T is explicitly specified.
695 ** Non-purified cores are significantly smaller than before
697 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
698 * new feature: timers based on Zach Beane's excellent timer package
699 * added support for the following external formats: koi8-u,
700 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
701 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
702 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
703 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
704 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
705 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
706 * incompatible change: a threaded SBCL will no longer revert to
707 non-threaded mode on non-NPTL systems, but refuse to start entirely.
708 * bug fix: interrupts are disabled until startup is complete; no
709 more sigsegvs when receiving a signal to soon
710 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
712 * bug fix: add a workaround for the memory randomization features in
713 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
714 workaround will only be in effect on systems with the proc filesystem
716 * bug fix: printing objects of type HASH-TABLE signals a
717 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
719 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
720 have been running for over 50 days. (reported by Gilbert Baumann)
721 * bug fix: the logic for getting names of functions gets less
722 confused when confronded with alternate-metaclass
723 funcallable-instances. (reported by Cyrus Harmon)
724 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
725 consistent, even on internal alternate-metaclass objects.
726 * bug fix: flush closure information collected by physical
727 environment analysis prepass before the main pass. (bug reported
729 * bug fix: compiler pack phase does not modify a hashtable, which is
730 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
731 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
732 may be expanded. (reported by Alexander <malishev> on c.l.l)
733 * bug fix: no more style-warnings when compiling code using LAST
734 under high-SPEED low-SPACE settings. (reported by David Wallin
736 * fixed some incompatibilities between SBCL's MOP and the MOP
738 ** the METAOBJECT class is now implemented;
739 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
740 STANDARD-OBJECT, as required;
741 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
742 now compatible (as required by VALIDATE-SUPERCLASS); there
743 remains a consistency requirement over the presence or absence
744 of the FUNCTION class in the superclass list of finalized
745 classes; see the manual for more details;
746 ** the :AROUND method for COMPUTE-SLOTS on
747 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
748 requested slot ordering.
750 ** bug fix: parent thread now can be gc'ed even with a live
752 ** bug fix: binding a special with PROGV to no value is not
753 the same as that symbol not having been bound (thanks to
755 ** bug fix: binding specials is thread safe (thanks to
757 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
758 and SB-SPROF do not die with 'no handler for signal XX in
759 interrupt_handle_now(..)' anymore
760 ** bug fix: WITH-TIMEOUT works with multiple running threads
761 * fixed some bugs revealed by Paul Dietz' test suite:
762 ** ENSURE-GENERIC-FUNCTION should take a method class object for
763 the :method-class keyword argument.
765 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
766 * new port: the Solaris operating system on x86 processors is now
767 mostly supported, though some rough edges in the environment
768 remain. (thanks to Daisuke Homma)
769 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
770 space for its heap. (thanks to Thiemo Seufer)
771 * optimization: SBCL on MIPS platforms now supports dynamic-extent
772 closures. (thanks to Thiemo Seufer)
773 * minor incompatible change: eof selects abort in the debugger.
774 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
775 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
777 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
778 deprecated, and will go away in a future revision of SBCL.
779 * minor incompatible change: GC-ON and GC-OFF are no longer
780 implemented with a counter, it does not matter how many times gc
781 is switched on or off
782 * bug fix: discriminating functions for generic function classes
783 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
784 make invalid assumptions about method precedence order. (reported
786 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
787 platforms. (thanks to Thiemo Seufer)
788 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
789 * bug fix: correct pathname printing with printer escaping is on.
790 (thanks to Kevin Reid)
791 * bug fix: complex VOP definitions in "user-space" no longer trigger
792 package locks. (reported by Zach Beane)
793 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
794 for funcallable-instance functions, and is no different from
796 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
797 characters in the current readtable. (reported by Nicholas Neuss)
798 * bug fix: SUBTYPEP on various CONS types returns more a more
799 accurate acknowledgment of its certainty.
800 * optimizations: REMOVE-DUPLICATES now runs in linear time on
801 lists in some cases. This partially fixes bug 384.
802 * flush all standard streams before prompting in the REPL and the
804 * bug fix: signal handling and triggering gc do not conflict
805 directly anymore, in particular a high frequency sb-sprof does
806 not prevent gc from running
807 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
808 approximation for timezone and DST information between the
809 universal time epoch and the smallest negative 32-bit time_t.
810 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
811 year 1899 so long as the encoded time is non-negative.
812 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
813 the correct answer for non-constant rotate amounts when performing
814 an inline 32-bit rotation.
816 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
817 there is only one thread in the session
818 ** bug fix: memory leak for streams created in one thread and
819 written to in another
820 ** bug fix: lockup when compiled with gcc4
821 ** bug fix: race that allows the gc to be triggered when gc is
823 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
824 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
826 ** bug fix: finished threads are gc'ed properly
827 * fixed some bugs revealed by Paul Dietz' test suite:
828 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
829 has arguments with a different set of applicable methods from
830 the orignal arguments.
831 ** The type error thrown by MAP now has a correct expected-type
833 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
834 name a compiled function.
835 ** Three MISC tests where a large bignum was improperly coerced to
836 a float that couldn't represent that bignum during type
837 derivation were fixed.
838 ** SUBTYPEP can now handle types involving the negation of a
839 list-form FUNCTION type.
840 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
841 as (COMPLEX (AND RATIO (NOT FIXNUM))).
842 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
844 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
845 * New feature: Experimental support for bivalent streams: streams
846 opened with :element-type :default now allow character and binary
847 (unsigned-byte 8) I/O
848 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
849 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
851 * bug fix: no more highly sporadic "couldn't check whether ~S is
852 readable" when reading a stream and an interrupt hits in the middle
853 of a select system call
854 * compiler better recognizes complex arrays (reported by Eduardo
856 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
857 for variables on SBCL built with linkage-tables. (reported by Luis
859 * various error reporting improvements.
860 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
861 (Thanks to Thiemo Seufer)
862 * bug fix: nested reader invokations work correctly
863 * bug fix: it is possible to have simultaneous references to foreign
864 code and foreign data with the same name.
866 ** added x86-64 support
867 ** incompatible change: the threading api now works with thread
868 objects instead of thread ids
869 ** bug fix: threads are protected from signals and interruption when
870 starting up or going down
871 ** bug fix: a race where an exiting thread could lose its stack to gc
872 ** bug fix: don't halt on infinite error in threads if possible
873 ** fixed numerous gc deadlocks introduced in the pthread merge
874 ** bug fix: fixed thread safety issues in read and print
875 ** bug fix: debugger doesn't hang on session lock if interrupted at
876 an inappropriate moment
877 ** bug fix: run-program is now thread safe(r)
878 ** bug fix: inner with-recursive-lock no longer releases the mutex
879 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
880 * fixed some bugs revealed by Paul Dietz' test suite:
881 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
882 EXPECTED-TYPE slots filled.
883 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
885 ** MAKE-CONDITION accepts classes as type-designators.
886 ** COMPILE may never return NIL.
887 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
888 range before calling Unix time functions
890 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
891 * numerous signal handling fixes to increase stability
892 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
893 * minor incompatible change: we now correctly canonize default
894 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
895 as per the MOP, rather than the historical (INITARG INITFUNCTION
896 INITFORM). (reported by Bruno Haible)
897 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
898 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
899 for more information.
900 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
901 pathname is a directory pathname.
902 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
903 * dynamic space size on PPC has been increased to 768Mb. (thanks to
905 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
906 :METACLASS argument in addition to a class name. (reported by
907 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
908 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
909 (reported by Bruno Haible; patch from Gerd Moellmann)
910 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
912 * bug fix: more cleanups to the floating point exception handling on
913 x86-64 (thanks to James Knight)
914 * bug fix: the compiler does not try to constant fold calls of
915 COERCE and BIT-* functions when they return freshly constructed
916 objects. (reported by Nikodemus Siivola and Paul Dietz)
917 * optimization: improved the MIPS versions of generic (in the
918 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
919 * optimization: direct conversion of (unsigned-byte 32) to floats on
920 the PowerPC platform.
921 * optimization: structure instances with raw slots now use less
922 memory, and probably show better memory locality. (thanks to
924 * optimization: DYNAMIC-EXTENT declarations for lists and closures
925 are treated as requests for stack allocation on the x86-64,
926 Alpha-32, PPC and SPARC platforms.
927 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
928 the wrong file descriptor; implementation of SOCKET-OPEN-P.
929 (thanks to Tony Martinez)
931 ** gcing a dead thread can no longer lead to lockups
932 ** threads block signals until they are set up properly
933 ** errno is no longer shared by threads
934 ** interrupt-thread restores the eflags register on x86
935 ** fixed some lockups due to gc/thread interaction
936 * fixed some bugs revealed by Paul Dietz' test suite:
937 ** invalid dotted lists no longer raise a read error when
939 ** signal an error if a symbol that names a declaration is used
940 as the name of a type, or vice versa
941 ** allow using the (declare (typespec var*)) abbreviation for
942 (declare (type typespec var*)) with all type specifiers
943 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
944 ** always return NIL from PPRINT-POP when OBJECT is NIL
945 ** don't signal errors when pretty-printing malformed LABELS,
946 FLET or MACROLET forms
947 ** declarations in a DOLIST body are also in scope for the
949 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
951 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
954 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
955 * fixed cross-compiler leakages that prevented building a 32-bit
956 target with a 64-bit host compiler.
957 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
958 opened with :IF-EXISTS :OVERWRITE.
959 * fixed bug 281: error for an invalid qualifier in a short-form method
960 combination method is not signalled until the faulty method is called.
961 * bug fix: iteration variable type inferrer failed to deal with open
962 intervals. (reported by Alan Shields)
963 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
964 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
965 or more runtime options were provided to the sbcl binary.
966 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
967 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
968 * optimization: there's now a fast-path for fixnum arguments in the
969 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
970 * optimization: the code generated on x86-64 is more compact thanks
971 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
972 * bug fix: floating-point exception handling now partly works on
973 x86-64. (Thanks to James Knight)
974 * improvement to the MIPS backend from Thiemo Seufer: C-style
975 64-bit long long arguments and return values to alien functions
977 * fixed some bugs revealed by Paul Dietz' test suite:
978 ** the type-error signalled from WARN has a filled-in DATUM slot.
979 ** the type-error required when a stream is not associated with
980 a file has the stream as its datum.
981 ** type-errors on single-floats on x86-64 no longer have
982 :INVALID-OBJECT as the datum
983 ** the type-errors signalled for invalid function names now have
984 a correct expected type
985 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
986 for structures defined with a :TYPE. Documentation strings for
987 typed structures are no longer immediately discarded
988 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
989 broadcast streams and synonym streams. FILE-LENGTH now also works
990 on broadcast streams.
992 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
993 * incompatible change: the --noprogrammer option, deprecated since
994 version 0.7.5, has been removed. Please use the equivalent
995 --disable-debugger option instead.
996 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
998 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
999 inoperational for a while now) has been completely removed.
1000 * null lexical environments are now printed as #<NULL-LEXENV>,
1001 significantly reducing the amount of clutter in typical backtraces.
1002 * documentation on weak pointers, finalization, and after GC hooks
1003 has been added to the manual.
1004 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1005 as other specialized array types, is much faster. SUBSEQ and
1006 COPY-SEQ on such arrays have also been sped up.
1007 * optimization: EQL is now more efficient when at least other argument
1008 is known to be of type (OR FIXNUM (NOT NUMBER)).
1009 * fixed bug: the runtime is now less vulnerable to changes in the
1010 size of the SBCL object on OS X, and virtual memory is reserved for
1011 all spaces that need to be at a fixed address.
1012 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1013 * fixed bug: finalizers and after GC hooks that cause consing are now
1015 * fixed bug: compiler error messages and summaries are now printed to
1016 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1017 * fixed inference of the upper bound of an iteration variable.
1018 (reported by Rajat Datta).
1019 * fixed bug 211e: calling local functions with duplicated constant
1020 keyword argument no longer causes a bogus style warning about an
1022 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1023 ftype declaration to be disregarded. (reported by Dave Roberts)
1024 * fixed bug 373: caused by erronous compilation of references to alien
1025 variables in the runtime on ppc/darwin.
1026 * fixed bug 376: CONJUGATE type deriver.
1027 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1028 (reported by Baughn on #lisp)
1029 * a cross-compiler bug on non-x86 platforms has been identified and
1030 fixed. (thanks to Bruno Haible)
1031 * improvements to the MIPS runtime code for increased stability.
1032 (thanks to Thiemo Seufer)
1033 * increased the maximimum compact environment size to allow
1034 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1035 * partial workaround for bug 135: don't name the function we're
1036 calling for hairy cases of EVAL, so as not to accumulate one environment
1037 entry per call to EVAL. (reported by Kevin Reid)
1038 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1039 * optimization: type testing for non-vector arrays should be faster.
1040 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1042 * bug fix: setting 31st element of a bit vector to zero did not work
1044 * bug fix: redefining a class definition which failed due to a
1045 previous accessor / function clash now works (but see BUGS entry
1046 #380 for more problems in this area). (thanks to Zach Beane)
1047 * the long form of DEFINE-METHOD-COMBINATION disables method group
1048 checking when given a single method group with pattern *. (thanks
1050 * on x86 compiler supports stack allocation of results of simple
1051 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1052 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1053 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1054 * fixed some bugs related to Unicode integration:
1055 ** the restarts for recovering from input and output encoding
1056 errors only appear when there is in fact such an error to
1058 * fixed some bugs revealed by Paul Dietz' test suite:
1059 ** MISC.549 and similar: late transformation of unsafe type
1060 assertions into derived types caused unexpected code
1062 ** SCALE-FLOAT type deriver is less wrong.
1063 ** type derivers for EXP, LOG and similar functions compute result
1064 types for complex arguments better.
1065 ** (MISC.563) CONJUGATE type deriver works for very restricted
1067 ** out-of-line type testers for character strings are available.
1068 ** EQUAL compiler transform understands specialness of objects
1070 ** accessing double-floats stored on the stack now works on x86-64.
1071 ** debugger internals could sometimes create invalid lispobjs,
1072 resulting in GC crashes.
1073 ** MISC.548: type check weakening can convert required type into
1075 ** initialization forms for bindings are not in scope of free special
1078 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1079 * incompatible change: thread support for non-NPTL systems has
1080 been removed - locking is buggy and unreliable. A threaded
1081 SBCL build will now warn at startup and refuse to create
1082 new threads, unless futex support is detected in the kernel
1083 * incompatible change: the top level REPL now has only an ABORT
1084 restart associated with it, not TOPLEVEL and ABORT as it used to.
1085 TOP and TOPLEVEL are now available as debugger commands for
1086 returning to the top level.
1087 * incompatible change: forms evaluated in the REPL now use the
1088 global optimization policy.
1089 * incompatible change: user- and system-initialization files are
1090 no longer processed with LOAD, but by READ and EVAL; hence the
1091 global optimization policy, startup package, readtable, etc,
1093 * The .fasl file format number has been incremented because of
1094 various incompatible changes.
1095 * internal entry point details and argument counts no longer appear
1096 in backtraces unless explicitly requested by setting
1097 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1098 * built-in and standard functions no longer have names like "top
1099 level local call to FOO".
1100 * fixed bug 32: functions defined in non-null lexical environments
1101 now have more legible printed representation
1102 * fixed bug 33: functions defined in non-null lexical environemnts
1103 are now more amenable to inspection by INSPECT.
1104 * workaround for bug 354: XEPs no longer appear in backtraces unless
1105 explicitly requested.
1106 * fixed bug: receiving the signal which results from attempting to
1107 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1108 SIGBUS on other *BSDs) on architectures where the C stack is also
1109 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1110 to debugger confusion or wild instability. (reported by Cheuksan
1112 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1113 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1114 (reported by Lutz Euler)
1115 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1116 compiled code. (reported by Paul Dietz)
1117 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1118 specialized parameter to a method to something that is not TYPEP
1119 the specializer is now possible.
1120 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1121 face of package deletion.
1122 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1123 pathname. (reported by David Tolpin on c.l.l)
1124 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1125 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1126 STRING-STREAM and FILE-STREAM, respectively.
1127 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1128 than previously; in addition, its test suite is now run on build.
1129 (thanks to Luke Gorrie)
1130 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1131 * optimization: added a immediate representation for single-floats
1133 * fixed some bugs revealed by Paul Dietz' test suite:
1134 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1135 ** attempting to create a package with a colliding nickname causes
1136 correctable errors to be signalled.
1137 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1138 ** several x86-64 backend bugs related to sign-extension of immediate
1141 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1142 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1143 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1145 * fixed loading of multiply forward-referenced layouts.
1146 (thanks to Cheuksan Wang)
1147 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1148 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1149 (see COMPILE-FILE documentation for details of :PRINT option)
1150 * fixed bugs 19 and 317: fixed-format floating point printing is
1151 more accurate. This also fixes a bug reported by Adam Warner
1152 related to the ~@F format directive.
1153 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1155 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1156 dispatch table if the from-char is a dispatch macro character.
1157 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1158 by the wordsize no longer ignore the last word. (reported by Lutz
1160 * fixed bug in type checking of dynamic-extent variables. (reported
1162 * optimization: sequence traversal functions use their freedom to
1163 coerce function designators to functions.
1164 * optimization: code with many calls to CLOS methods specialized on
1165 CLOS classes has had redundant type checks removed.
1166 * optimization: added declarations to speed up operations that access
1167 the internal character database (for example STRING-UPCASE)
1168 * optimization: comparison operations between floats and sufficiently small
1169 fixnums no longer create extra rationals
1170 * fixed some bugs related to Unicode integration:
1171 ** portions of multibyte characters at the end of buffers for
1172 character-based file input are correctly transferred to the
1173 start of the buffer at the next read.
1174 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1175 passing it through to OPEN.
1176 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1177 argument given to internal calls to OPEN.
1178 ** the built-in comment readers (introduced by character sequences
1179 ";" and "#|") are more forgiving to encoding errors; they will
1180 STYLE-WARN and then attempt to resync the stream at a character
1181 boundary. (thanks to Teemu Kalvas)
1182 * fixed some bugs in the x86-64 port:
1183 ** Negative short int return values from c-calls are sign-extended
1185 ** The stack is aligned to 16-bytes for c-calls, as required by
1186 the ABI. (thanks to Cheuksan Wang)
1187 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1188 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1189 for enums. (thanks to Vincent Arkesteijn)
1190 ** Multiple small optimizations and bugfixes for floating point
1192 * fixed some bugs revealed by Paul Dietz' test suite:
1193 ** Space, Tab, Linefeed, Return and Page have the invalid
1194 secondary constituent character trait.
1195 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1197 ** WITH-INPUT-FROM-STRING should only update the index place on
1199 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1200 works more reliably.
1201 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1202 with one-dimensional arrays with fill pointers.
1203 ** TYPE-OF failed on a complex with an integer realpart and a
1205 ** compiler failure during type inference for the code of form
1206 (IF (EQL X (THE ...)) ...) (MISC.535).
1208 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1209 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1210 platform. The port supports 61-bit fixnums, large memory spaces
1211 and reloading shared object files.
1212 * enhancement: saving cores with foreign code loaded is now
1213 supported on ppc/Darwin in addition to the previously supported
1215 * enhancement: the statistical profiler now walks deeper into the
1216 call stack for more meaningful call-graphs and accrued time
1217 reports (x86/x86-64 only). It also now reports time spent in
1219 * enhancement: it is now possible to trace most individual methods
1220 of a generic function in addition to tracing the generic function
1222 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1224 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1225 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1226 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1227 produces an error. (thanks to Vincent Arkesteijn)
1228 * bug fix: NAMESTRING on pathnames with :WILD components in their
1229 directories works correctly. (thanks to Artem V. Andreev)
1230 * fixed bug 125: compiler preserves identity of closures. (reported
1232 * bug fixed: functions with &REST arguments sometimes failed with
1233 "Undefined function" when compiled with (DEBUG 3). (reported by
1235 * bug fix: overflow during compiling of setting element of a bit
1236 vector with constant index and value. (reported by Timmy Douglas)
1237 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1238 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1239 * on x86 compiler partially supports stack allocation of dynamic-extent
1241 * GO and RETURN-FROM do not check the extent of their exit points
1242 when compiled with SAFETY 0.
1243 * fixed some bugs related to Unicode integration:
1244 ** encoding and decoding errors are now much more robustly
1245 handled; it should now be possible to recover even from invalid
1246 input or output to the terminal. (thanks to Teemu Kalvas)
1247 ** provided a first cut at implementing STRING-TO-OCTETS and
1248 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1249 ** altered the SB-MD5 contributed module slightly, changing the
1250 interface just enough for it to be supportable for builds where
1251 lisp characters are not eight bits.
1252 * fixed some bugs revealed by Paul Dietz' test suite:
1253 ** the FORMATTER-generated functions for ~V[ conditionals require
1254 the correct number of arguments.
1255 ** READ-FROM-STRING returns the mandated second value when applied
1256 to displaced strings.
1257 ** the #\Rubout and #\Backspace characters are treated as invalid
1258 constituent characters by the tokenizer.
1260 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1261 * new feature: reloading changed shared object files with
1262 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1263 * new feature: references to foreign variables and functions
1264 can now be compiled and loaded before the corresponding shared
1265 object file is loaded, as long as the foreign definitions are
1266 available at runtime.
1267 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1268 supported. (thanks to Dan Debertin)
1269 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1270 just keywords. (thanks to Vincent Arkesteijn)
1271 * fixed bug #331: structure-class instances corresponding to
1272 DEFSTRUCT forms are now created eagerly.
1273 * fixed bug #345: backtraces from calls to undefined functions work
1274 on x86 as well. Related bug #61 is now also partially fixed on x86
1275 (backtraces from throws to unknown catch tags.)
1276 * bug fix: lambda-list parsing is now stricter vrt. order and number
1277 of lambda-list keywords.
1278 * bug fix: as specified by AMOP, an error is signalled if a
1279 class-option appears multiple times in a DEFCLASS form. (reported
1281 * bug fix: RANDOM can be compiled when the compiler derives the type
1282 of its numeric argument as a disjoint set of small integers.
1283 (reported by Paul Dietz)
1284 * bug fix: starting a core saved with shared objects loaded when
1285 those objects are not available no longer causes threaded SBCL to
1286 hang. (reported by Sean Ross)
1287 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1289 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1290 argument, which must always be NIL. (reported by Kalle Niemitalo)
1291 * bug fix: printing 1.0d+23 no longer results in an error.
1292 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1293 * bug fix: structure slot setters preserve evaluation order. (thanks
1295 * fixed some bugs related to Unicode integration:
1296 ** RUN-PROGRAM can allow its child to take input from a Lisp
1297 stream. (reported by Stefan Scholl)
1298 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1299 locales. (reported by Ken Causey)
1300 * fixed some bugs revealed by Paul Dietz' test suite:
1301 ** INCF, DECF and REMF evaluate their place form as specified in
1303 ** FORMATTER expands ~{ ~} iteration directives with V or #
1304 parameters correctly.
1305 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1306 consequent uses no arguments correctly.
1307 ** the system has a partial understanding of the (COMPLEX RATIO)
1309 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1312 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1313 * new feature: a build-time option (controlled by the :SB-UNICODE
1314 keyword feature, enabled by default) for building the system with
1315 support for the entire 21-bit character space defined by the
1317 * new feature: the system now has rudimentary external-format
1318 support; the primary user-visible change at this time is that
1319 characters with the high bit set (such as lower-case-e-acute) will
1320 print correctly to a terminal in a UTF-8 environment.
1321 * minor incompatible change: BASE-CHAR no longer names a class;
1322 however, CHARACTER continues to do so, as required by ANSI.
1323 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1324 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1325 has been moved to the SB-EXT package (temporarily re-exported from
1327 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1328 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1330 * fixed bug #351: better error-handlind and reporting for malformed
1332 * fixed bug #350: bignum-printing is now more memory-efficient,
1333 allowing printing of very large bignums, eg. (expt 2 10000000).
1334 (reported by Bruno Haible)
1335 * fixed bug #302: better primitive-type selection for intersection
1337 * fixed bug #308: non-graphic characters now all have names, as
1338 required. (reported by Bruno Haible)
1339 * bug fix: redefining a class with different superclasses now correctly
1340 removes it from the direct-subclasses of its previous superclasses.
1341 (reported by David Morse)
1342 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1343 new value now works. (reported by Bruno Haible)
1344 * bug fix: correct canonicalization of multiple non-standard slot
1345 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1346 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1347 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1348 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1349 now exists, an signals an error.
1350 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1351 during macroexpansion for non-iteration variables as well. (reported
1352 by Bruno Haible for CMUCL)
1353 * bug fix: Cyclic structures and unprintable objects in compiler
1354 messages no longer cause errors. (reported by Bruno Haible)
1355 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1356 and READ-FROM-STRING all now return a primary value of NIL if
1357 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1358 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1359 (reported by Bruno Haible for CMUCL)
1360 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1361 do not designate a condition. (reported by Bruno Haible for
1363 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1364 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1365 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1366 permitting resolution in favour of any of the conflicting symbols.
1367 (reported by Bruno Haible for CMUCL)
1368 * bug fix: EQUAL compiler optimizations is less aggressive on
1369 strings which can potentially compare true despite having distinct
1370 specialized array element types.
1371 * bug fix: unit enumerations can be defined without dividing by
1372 zero. (thanks to Vincent Arkesteijn)
1373 * FORMAT compile-time argument count checking has been enhanced.
1374 (report from Bruno Haible for CMUCL)
1375 * a partial workaround for the bug 262: the compiler does not try to
1376 inline-expand a local function doing RETURN-FROM from a deleted
1377 BLOCK. (thanks to Peter Denno for the bug report and to David
1378 Wragg for the simple test case)
1379 * fixed some bugs revealed by Paul Dietz' test suite:
1380 ** RENAME-PACKAGE allows all package designators as new package
1382 ** constraint propagation and lambda variable substitution are
1383 more cautious in dealing with partially deleted code.
1384 ** compiler sometimes lost reoptimization passes.
1385 ** CERROR, given a condition as condition designator, passes its
1386 remaining arguments to the continue format control without
1388 ** Case-altering FORMAT directives work correctly on non-ASCII
1390 ** The REAL type specifier handles bounds outside the floating
1391 point ranges without signalling FLOATING-POINT-OVERFLOW.
1392 ** Functions with IR1-transformations can create intercomponent
1393 references to global functions.
1394 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1396 ** FORMAT ~R treats a nil value for its first parameter correctly.
1398 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1399 * enhancement: saving cores with foreign code loaded is now
1400 supported on x86/NetBSD and sparc/Linux in addition to the previously
1401 supported platforms.
1402 * bug fix: on some platforms repeated installations caused multiple
1403 copies of HTML documentation to be installed -- should not happen
1404 any more. (reported by Stefan Scholl)
1405 * bug fix: parsing self-recursive alien record types multiple times
1406 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1407 original patch by Helmut Eller for CMUCL)
1408 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1409 (thanks to Richard Kreuter)
1410 * bug fix: defining classes whose accessors are methods on existing
1411 generic functions in other (locked) packages no longer signals
1412 bogus package lock violations. (reported by François-René Rideau)
1413 * bug fix: special variables as DEFMETHOD parameters no longer have
1414 associated bogus type declarations. (reported by David Wragg and
1416 * bug fix: read-write consistency on streams of element-type
1417 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1418 * bug fix: redefiniton of the only method of a generic function with
1419 no DEFGENERIC no longer emits a full WARNING. In addition,
1420 redefinition of generic functions with no DEFGENERIC to an
1421 incompatible lambda list now signals an error. (thanks to Zach
1423 * bug fix: DEFGENERIC now works even when there's a function of the
1424 same name in an enclosing lexical environment. (thanks to Zach
1426 * fixed compiler failure, caused by instrumenting code during
1427 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1428 * optimization: added loop analysis and improved register allocation
1429 * fixed some bugs revealed by Paul Dietz' test suite:
1430 ** POSITION on displaced vectors with non-zero displacement
1431 returns the right answer.
1432 ** (SIMPLE-STRING) is a valid type specifier for sequence
1434 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1436 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1437 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1439 ** The justification version of the FORMAT ~< directive treats
1440 non-zero minpad parameter correctly.
1442 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1443 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1444 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1445 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1446 the supported interface.
1447 * new feature: Single-stepping of code compiled with DEBUG 2 or
1448 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1449 * new feature: saving cores with foreign code loaded is now
1450 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1451 Timothy Moore's work for CMUCL)
1452 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1453 parameters to * instead of NIL if no initform is supplied.
1454 (reported by Johan Bockgaard)
1455 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1456 correctly when FUNCALL appears as the car of the form. Note:
1457 despite this FUNCALL forms are not currently subject to
1458 compiler-macro expansion. (port of Raymond Toy's fix for the
1459 same from CMUCL, reported by Johan Bockgaard)
1460 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1461 (thanks for Teemu Kalvas)
1462 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1463 case instead of terminating the iteration (thanks for Julian
1464 Squires, Sean Champ and Raymond Toy)
1465 * bug fix: incorrect expansion of defgeneric that caused a style
1466 warning. (thanks for Zach Beane)
1467 * bug fix: slot accessor effective method computation works properly
1468 for classes with multiple non-standard applicable methods on
1469 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1470 * on x86 compiler supports stack allocation of results of LIST and
1471 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1472 CMUCL implementation by Gerd Moellmann)
1473 * fixed some bugs revealed by Paul Dietz' test suite:
1474 ** FORMAT strings with both the ~<~:;~> form of the justification
1475 directive and pretty-printing directives cause an error.
1477 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1478 * incompatible change: the internal functions
1479 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1480 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1481 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1482 instead of the old functions.
1483 * new feature: on platforms where "dladdr" is available foreign
1484 function names now appear in backtraces. (based on Helmut Eller's
1486 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1487 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1489 * documentation: ASDF manual is now distributed with SBCL.
1490 * bug fix: SBCL can now load its contributed modules with REQUIRE
1491 even if the system-provided entries have been removed from
1492 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1494 * bug fix: backtraces involving undefined functions or assembly
1495 routines are more informative. (thanks to Brian Downing)
1496 * bug fix: mutually referent alien structures now work correctly.
1497 (reported by Rick Taube)
1498 * bug fix: structures defined by WITH-ALIEN can be referred to
1499 within other definitions in the same WITH-ALIEN.
1500 * bug fix: division operators (MOD, TRUNCATE and the like) with
1501 constant zero divisors and integer dividends no longer generate
1503 * bug fix: provide default methods for INPUT-STREAM-P and
1504 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1505 * bug fix: improve the detection and resolution of MOP metacycles
1506 (where effective method computation depends on the generic function
1507 having its effective method computed). (reported by Bruno Haible)
1508 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1509 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1510 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1511 * bug fix: pathnames with relative directory components can be
1512 represented relative to default pathnames.
1513 * optimization: in taking the GCD of bignums, reduce the two bignums
1514 to approximately the same size (using Euclid's algorithm) before
1515 applying the more sophisticated binary GCD. (thanks to Juho
1517 * optimization: COUNT on bitvectors now operates word-at-a-time.
1518 * optimization: ASH with a positive, but not necessarily constant,
1519 (leftwards) shift, when in a modular context, is compiled to a
1521 * fixed some bugs revealed by Paul Dietz' test suite:
1522 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1523 if the corresponding argument is NIL.
1524 ** FORMAT directives accepting a minpad parameter treat negative
1526 ** literal commas as character parameters to FORMAT directives are
1528 ** literal spaces directly after ~<Newline> directives within a
1529 format-logical-block (~:< ~@:>) do not induce :FILL-style
1530 conditional newlines.
1531 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1532 as specified: it no longer includes conditional newlines).
1533 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1535 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1536 :PER-LINE-PREFIX argument does not evaluate to a string.
1538 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1539 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1540 the manual for details; for now, package locks can be disabled by
1541 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1542 if no major problems are found then it is likely that they will be
1543 compiled in unconditionally.
1544 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1545 unsupported operators on all platforms. To load a shared library
1546 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1547 object file, link it into a shared library outside of SBCL and
1548 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1550 * bug workaround: ROOM T and the GC were not working together
1551 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1552 reporting facility aren't preserved. That reporting has been
1553 disabled, so now until and unless someone figures out how to make it
1554 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1555 * minor incompatible change: as threatened around sbcl-0.8.0, the
1556 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1557 The symbols are also exported from SB-PCL for backwards
1558 compatibility, but more so than before SB-PCL should be treated as
1559 an implementation-internal package.
1560 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1562 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1563 It is known as of this release not to work on the Alpha, however.
1564 * fixed bug #167: errors signalled due to illegal syntax in method
1565 bodies are now more legible.
1566 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1567 designators and can hence be used with TYPEP.
1568 * fixed bug #333: CHECK-TYPE now ensures that the type error
1569 signalled, if any, has the right object to be accessed by
1570 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1571 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1572 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1574 * fixed bug #334: programmatic addition of slots using specialized
1575 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1576 and :ALLOCATION :CLASS slots.
1577 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1578 float exponents. (rereported by Peter Seibel)
1579 * fixed a bug: #\Space (and other whitespace characters) are no
1580 longer considered to be macro characters in standard syntax by
1581 GET-MACRO-CHARACTER.
1582 * fixed bug: initialization of condition class metaobjects no longer
1583 causes an instance of the condition to be created. (reported by
1585 * fixed bug: it is now possible to have slots such that
1586 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1587 description is neither :INSTANCE nor :CLASS.
1588 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1589 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1590 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1591 non-standard SLOT-DEFINITION-ALLOCATION values.
1592 * fixed bugs: various race conditions handling exiting threads.
1593 CL-PPCRE's thread torture test now passes.
1594 * fixed arguably-a-bug: GC time proportional to number of threads in
1595 system even when most of them are idle
1596 * optimization: improved performance of BIT and SBIT on bit-vectors.
1597 * .fasl file incompatibility: The fasl file version number has
1598 been incremented because of changes associated with package locks.
1600 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1601 * minor incompatible change: the system no longer provides
1602 optimization hints (conditions which are TYPEP
1603 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1604 mode; these hints are still emitted when higher SPEED optimization
1606 * new contrib module: a sampling profiler (profiling by statistical
1607 sampling, rather than by instrumenting functions) is available as
1608 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1609 * the behaviour of the standard function ED is now customizeable by
1610 third parties through a hook variable: see ED's documentation
1611 string for information on the protocol.
1612 * the compiler no longer emits efficiency notes for (FUNCALL X)
1613 when the type of X is uncertain under default optimization
1615 * fixed bug 276: mutating a binding of a specialized parameter to a
1616 method to something that is not TYPEP the specializer is now
1618 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1619 exhibit the required behaviour on the x86 platform. (thanks to
1620 Peter van Eynde, Eric Marsden and Bruno Haible)
1621 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1622 even for difficult arguments. (reported by Peter Graves)
1623 * fixed bug 141a: the backquote printer now descends quoted
1625 * fixed another bug in backquote printing: no more destructive
1626 modification of the form's list structure. (reported by Brian
1628 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1629 move between its address being taken and the call to
1630 interrupt_thread, fixing a crashing race condition.
1631 * the SB-POSIX contrib implementation has been adjusted so that it
1632 no longer exhibits ridiculously poor performance when constructing
1633 instances corresponding to C structs.
1635 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1636 * minor incompatible change: the sb-grovel contrib now treats C
1637 structures as alien (in the sense of SB-ALIEN) objects rather than
1638 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1639 has implications for memory management of client code
1640 (sb-grovel-returned objects must now be manually managed) and for
1641 type safety (alien objects now have full types).
1642 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1643 used to control emission of compiler diagnostics, rather than the
1644 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1645 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1646 quality should be considered deprecated.
1647 * install.sh now installs the user manual as well
1648 * (not quite a new documentable feature, but worth considering in
1649 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1650 beginnings of a semantically meaningful condition hierarchy is
1651 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1653 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1654 designator as the defaults argument.
1655 * fixed bug: Displaced arrays whose displaced-to array has become
1656 too small now cause ARRAY-DIMENSION to signal an error, providing
1657 for safer bounds-checking. (reported by Bruno Haible)
1658 * fixed bug: DEFCLASS slot definitions with identical :READER and
1659 :WRITER names now signal a reasonable error. (reported by Thomas
1661 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1663 * fixed bug: Invalid slot specification errors now print correctly.
1664 (thanks to Zach Beane)
1665 * fixed bug 320: Shared to local slot value transfers in class
1666 redefinitions now happen corrently from superclasses as
1667 well. (reported by Bruno Haible)
1668 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1670 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1671 improper lists correctly. (reported by Bruno Haible)
1672 * fixed bug 313: source-transform for <fun-name> was erroneously
1673 applied to a call of a value of a variable with name <fun-name>.
1674 (reported by Antonio Menezes Leitao)
1675 * fixed bug 307: The obsolete instance protocol ensures that
1676 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1677 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1678 remove dead unknown-values globs from the middle of the stack.
1679 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1680 easier to resume long computations after using *BREAK-ON-SIGNALS*
1681 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1682 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1683 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1684 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1685 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1686 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1688 * fixed a (fairly theoretical) bug in string printing: if
1689 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1690 does not have array-element-type equal to the most general string
1692 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1693 function name. (thanks to Nikodemus Siivola)
1694 * fixed bug: compile-time format string checker failed on
1695 non-closed ~{. (reported by Thomas F Burdick)
1696 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1697 don't warn on legal loop code involving more than one aggregate
1698 boolean. (thanks to Nikodemus Siivola)
1699 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1700 in keyword position to MAKE-PACKAGE was not being reported as
1702 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1703 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1705 * fixed bugs 280 and 312: the checking for multiple definitions in a
1706 file is less likely to become confused by uses of inline
1708 * fixed bug: the #S reader macro performs the keyword coercion
1709 specified for slot names. (reported by Kalle Niemitalo)
1710 * fixed bug: lambda lists may contain symbols whose names start with
1711 & but are not lambda-list-keywords; their occurrence triggers a
1713 * fixed bug 321: define-method-combination argument lambda lists do
1714 not cause invalid code to be generated when &optional or &aux
1715 variables have default values. (reported by Bruno Haible)
1716 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1717 have CLOS classes; this appears also to have fixed reported
1718 instability in stack exhaustion detection.
1719 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1720 MUFFLE-WARNING when no associated restart is present is now
1722 * optimization: rearranged the expansion of various defining macros
1723 so that each expands into only one top-level form in a
1724 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1726 * optimization: used a previously little-used slot in symbols to
1727 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1729 * fixed some bugs revealed by Paul Dietz' test suite:
1730 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1732 ** Fixed an optimization bug related to inheritance of initforms
1733 from local to shared slots.
1734 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1735 ** CAST optimizer forgot to flush argument derived type.
1736 ** print/read consistency on floats is now orders of magnitude
1737 more likely. (thanks also to Bruno Haible for a similar report
1739 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1740 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1741 for objects of type REAL. Make it so.
1742 ** SXHASH is specified (infelicitously) to respect similarity,
1743 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1744 it so. (thanks to Markus Ziegler)
1745 ** on the Alpha, the compiler succeeds in compiling functions
1746 returning a known number of arguments greater than 63.
1747 ** fixed handling of invalid NIL arguments in keyword position
1749 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1751 ** provided a readably-printable representation for RANDOM-STATE
1753 ** ensured that pathnames, where they have a namestring, always
1754 print using #P"..." syntax.
1756 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1757 * Support for the forthcoming 2.0 version of the NetBSD kernel
1758 running on x86 hardware has been added. (thanks to Perry
1759 E. Metzger most immediately, and others for their past work)
1760 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1761 current release version; SBCL's OpenBSD support had been broken
1762 since about the time of OpenBSD's switch to ELF binary format.)
1763 * [placeholder for DX summary]
1764 ** user code with &REST lists declared dynamic-extent, under high
1765 speed or space and low safety and debug optimization policy.
1766 * The manual has been converted to Texinfo format and the debugger
1767 chapter from the cmucl manual has been added.
1768 * A facility has been added to extract documentation strings from
1769 sbcl and store them as Texinfo-formatted snippets for inclusion in
1770 the manual (via Texinfo's @include directive)
1771 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1772 the test case to Dave Roberts)
1773 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1774 pointers. (reported by Sean Ross)
1775 * bug fix: PROFILE output is printed nicely even for large numerical
1776 values. (thanks to Zach Beane)
1777 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1778 greater than 32 handle EOF correctly.
1779 * bug fix: on X86 an immediate argument of the IMUL instruction is
1780 correctly printed in disassembly. (thanks to Lutz Euler)
1781 * bug fix: class slots in redefined classes preserve their old
1782 values. (thanks to Bruno Haible and Nikodemus Siivola)
1783 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1784 succeeds. (reported by Marco Baringer)
1785 * fixed some bugs revealed by Paul Dietz' test suite:
1786 ** READ-SEQUENCE now works on ECHO-STREAMs.
1787 ** RATIONALIZE works more according to its specification. (thanks
1790 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1791 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1792 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1793 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1794 should matter to you only if you rebind the printer control
1795 variables and then find you want different bindings in the
1796 debugger than in the ordinary execution of your program.)
1797 * The runtime build system has been tweaked to support building
1798 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1799 assembler and linker. (thanks to Nikodemus Siivola)
1800 * Unbound, undefined, undeclared variables now trigger full
1801 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1802 more often programmer error than deliberate exploitation of undefined
1804 * optimization: the hash algorithm for strings has changed to one
1805 that is less vulnerable to spurious collisions. (thanks to Juho
1807 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1808 needless bounds checking. (thanks to Juho Snellman)
1809 * optimization: implemented multiplication as a modular
1810 (UNSIGNED-BYTE 32) operation on the PPC backend.
1811 * fixed some bugs revealed by Paul Dietz' test suite:
1812 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1814 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1815 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1816 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1818 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1820 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1822 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1823 their output stream on EOF from read.
1824 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1825 have been read to end-of-file.
1826 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1828 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1829 description of determination of which consecutive characters
1831 ** Printing the "Space" character with escaping on now yields "#\\ ",
1832 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1833 ** Reading floating-point numbers with *READ-BASE* set to a number
1834 less than 10 works correctly.
1835 ** Reading floating-point numbers with *READ-BASE* set to a number
1836 more than 10 works correctly.
1837 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1838 the readtable currently in effect.
1840 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1841 * minor incompatible change: parsing of namestrings on a physical
1842 (Unix) host has changed; numbers after the final #\. in a
1843 namestring are no longer interpreted as a version field. This is
1844 intented to be largely invisible to the user, except that the
1845 meaning of the namestring "*.*.*" has changed: it now refers to a
1846 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1847 should usually be replaced by
1848 (make-pathname :name :wild :type :wild :version :wild)
1849 with the added benefit that this is more likely to be portable.
1850 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1851 option to OPEN now signals an error if the file being opened
1852 exists; this may have an impact on existing code.
1853 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1854 signals received, on the PowerPC platforms (both Linux and
1855 Darwin). (thanks to Pierre Mai for pointing out the location of
1857 * several fixes on OS X: The system now builds and runs cleanly on
1858 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1859 (thanks to Brian Mastenbrook)
1860 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1861 second-resolution: integer multiples of 1/3600 between -24 and 24.
1862 (thanks to Vincent Arkesteijn)
1863 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1864 type when called with 1 argument; PEEK-CHAR checked type of
1865 PEEK-TYPE only after having read first character from a
1866 stream. (reported by Peter Graves)
1867 * bug fix: the garbage collector now has much better locality
1868 behaviour, and in particular no longer treats all memory as being
1869 exhausted when many small objects point to each other in a deeply
1871 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1872 recognized as being TYPEP their class.
1873 * bug fix: the PUSHNEW documentation string has been corrected.
1874 (thanks to Vincent Arkesteijn)
1875 * bug fix: defaulting of the value for the last of an atypically
1876 large number of multiple values being bound was not being
1877 performed correctly on the Alpha or PPC platforms
1878 * optimization: implemented multiplication as a modular
1879 (UNSIGNED-BYTE 32) operation on the x86 backend.
1880 * optimization: SEARCH on simple-base-strings can now be open-coded.
1881 (see also contrib/compiler-extras.lisp for inspiration for
1882 teaching the compiler about the Boyer-Moore algorithm).
1883 * value, returned by MAX (and MIN) called with several EQUALP, but
1884 not EQL, arguments now does not depend on compiler settings.
1885 * fixed some bugs revealed by Paul Dietz' test suite:
1886 ** in stack analysis liveness information is propagated from
1887 non-local entry points.
1888 ** pathwise CAST removing failed when the CAST node did not start
1890 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1891 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1893 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1894 host is already defined.
1895 ** RENAME-FILE works on streams instead of signalling an internal
1897 ** PEEK-CHAR uses the current readtable when determining whether
1898 or not a character is whitespace.
1899 ** MERGE-PATHNAMES handles the case when the pathname does not
1900 specify a name while the default-pathname specifies a version.
1901 ** Pathnames now stand a better chance of respecting print/read
1903 ** Attempting to use standardized file system operators with a
1904 pathname with invalid :DIRECTORY components signals a
1906 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1908 ** DIRECTORY on logical pathnames is more correct.
1909 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1910 signal a TYPE-ERROR if their argument is not a stream.
1911 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1912 designator argument does not designate a stream.
1913 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1914 examining the synonym.
1915 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1917 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1918 element-type larger than ([UN]SIGNED-BYTE 32).
1920 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1921 * When built with the :SB-FUTEX feature, threaded builds now take
1922 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1923 for faster/more reliable mutex and condition variable support.
1924 * Incompatible change (but one you probably shouldn't have been using
1925 anyway): the interface and code for arbitrating between multiple
1926 threads in the same user session has been redesigned.
1927 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1928 error when called without an explicit environment argument.
1929 (thanks to Nikodemus Siivola)
1930 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1931 confused about their position. (thanks to Adam Warner and Gerd
1933 * bug fix: Pretty printing backquoted forms with unquotations in the
1934 argument list position of various code constructs such as LAMBDA
1935 now works correctly. (reported by Paul Dietz)
1936 * bug fix: Pretty printing unquotations no longer loses all
1937 stream position information.
1938 * optimization: performance of string output streams is now less
1939 poor for multiple small sequence writes.
1940 * optimization: performance of CSUBTYPEP in the presence of complex
1941 expressions involving CONS and NOT many times has been improved.
1942 (reported by Paul Dietz)
1943 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1945 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1946 streams for the REPL, for the debugger, and so on. (thanks to
1948 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1949 * fixed some bugs revealed by Paul Dietz' test suite:
1950 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1951 is ignored if it is too small, rather than propagating through
1952 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1953 ** extremely complex negations of CONS types were not being
1954 sufficiently canonicalized, leading to inconsistencies in
1956 ** VALUES tranformer lost derived type.
1958 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1959 * fixed a bootstrapping bug: the build process no longer assumes
1960 that the various BOOLE-related constants have the same value in
1961 host and target lisps. (noted by Paul Dietz' test suite on an
1962 SBCL binary built from CLISP)
1963 * The system can now be dynamically linked on the MIPS platform,
1964 which enables dynamic loading of foreign code from Lisp. (thanks
1965 to Ralf Baechle for discussions on the MIPS ABI)
1966 * The system now records debugging information for its own source
1967 files in a filesystem-position-independent manner, relative to
1968 the "SYS" logical host.
1969 * fixed a compiler bug: MV-LET convertion did not check references
1970 to the "max args" entry point. (reported by Brian Downing)
1971 * tweaked disassembly notes to be less confident about proclaiming
1972 some instruction as an LRA. (thanks to Brian Downing)
1973 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1974 now each have their own history, command character, and other
1975 characteristics. (thanks to David Lichteblau)
1976 * fixed some bugs revealed by Paul Dietz' test suite:
1977 ** compiler failure in compiling LOGAND expressions including a
1979 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1980 shift greater than 32.
1981 ** FLUSH-DEST did not mark blocks for type check regeneration.
1982 ** HANDLER-CASE failed to accept declarations in handler clauses
1983 in some circumstances.
1985 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1986 * New code in contrib/sb-introspect (still probably not entirely
1987 stable yet) provides some support for smart Lisp development
1988 environments like SLIME.
1989 * The conditions signalled for errors occurring when loading .fasl
1990 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1991 in a way which should help ASDF recover gracefully.
1992 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1993 stuff has been cleaned up. If you code contrib/ stuff, this might
1994 affect you, and you can look at contrib/README, contrib/STANDARDS,
1995 and/or the 0.8.4.27 diff to check.
1996 * In full calls the compiler now does not generate checks for declared
1997 argument types for all arguments.
1998 * various threading fixes
1999 ** and some experimental patches which didn't make it into
2000 the main tree for this release, but which are shipped in
2001 contrib/experimental-thread.patch as a possible fix for some
2002 failures (deadlock, spinning...) in GC-intensive multithreaded
2004 * fixed PPC build problem (source code incompatibility of different
2005 library versions): added offsetof() hackery which attempts to divine
2006 where glibc maintainers put uc_mcontext today
2007 * fixed bug 282: compiler does not trust type assertions while passing
2008 arguments to a full call.
2009 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2010 &OPTIONAL VALUES type parameter.
2011 * fix bug 214: algorithm for noting rejected templates is now more
2012 similar to that of template seletion. (also reported by rydis on
2014 * fixed bug 141b: printing backquoted information readably and prettily
2015 inserts a space where necessary.
2016 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2017 (CHARACTER 10) are now reported as errors, rather than propagated
2018 as unknown types. (reported by piso on #lisp)
2019 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2020 with values NIL and :ERROR. (thanks to Milan Zamazal)
2021 * fixed bug 191c: CLOS now does proper keyword argument checking as
2022 described in CLHS 7.6.5 and 7.6.5.1.
2023 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2024 longer raise an error; further, using a list as a for-as-arithmetic
2025 counter now raises a meaningful error.
2026 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2027 now understood by sequence creation functions such as MAKE-SEQUENCE
2029 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2030 a STRING-INPUT-STREAM.
2031 * compiler enhancement: SIGNUM is now better able to derive the type
2033 * type declarations inside WITH-SLOTS are checked. (reported by
2035 * fixed some bugs revealed by Paul Dietz' test suite:
2036 ** incorrect optimization of TRUNCATE for a positive first
2037 argument and negative second.
2038 ** compiler failure in let-convertion during flushing dead code.
2039 ** compiler failure while deriving type of TRUNCATE on an
2040 interval, containing 0.
2041 ** ASH of a negative bignum by a negative bignum count now returns
2043 ** intersection of CONS types now canonicalizes properly, fixing
2044 inconsistencies in SUBTYPEP.
2046 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2047 * incompatible change: The --disable-debugger command line
2048 option now clobbers the debugger at a more fundamental
2049 level, by redefining #'INVOKE-DEBUGGER instead of by
2050 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2051 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2052 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2053 enter the debugger REPL and then suffer recursive errors
2054 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2055 while under the new system, BREAK in --disable-debugger mode
2056 terminates the system just as an unhandled error would.
2057 * fixed compiler performance when processing loops with a step >1;
2058 * bug fix: DOCUMENTATION now retrieves generic function
2059 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2060 support has been systematized, and now supports the methods
2061 specified by ANSI, along with a default method and a method for
2062 slot documentation. (reported by Nathan Froyd)
2063 * bug fix: effective methods associated with a generic function are
2064 no longer cached over a change of that generic function's method
2065 combination. (reported by Andreas Fuchs)
2066 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2067 names a non-existent directory. (thanks to Andreas Fuchs)
2068 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2069 the PPC platform, lead to stack corruption; this has been fixed.
2070 (reported by Rainer Joswig)
2071 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2072 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2073 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2074 arithmetic in general, is now much more reliable on the PPC
2076 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2077 platform now returns the right answer.
2078 * optimization: restored some effective method precomputation in
2079 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2080 precomputation is now tunable.
2081 * optimization: compiler-internal data structure use has been
2082 reviewed, and changes have been made that should improve the
2083 performance of the compiler by about 20%.
2084 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2085 simple-base-strings and simple-bit-vectors is improved.
2086 * optimization: the optimization of 32-bit logical and arithmetic
2087 functions introduced in version 0.8.3 on the x86 has been
2088 implemented on the mips, ppc and sparc platforms; an
2089 implementation of the same facility, but for 64-bit arithmetic,
2090 has been added for the alpha.
2091 * microoptimization: the compiler is better able to make use of the
2092 x86 LEA instruction for multiplication by constants.
2093 * bug fix: in some situations compiler did not report usage of
2094 generic arithmetic in (SPEED 3) policy.
2095 * bug 145b fix: compiler used wrong type specifier while converting
2096 MEMBER-types to numeric.
2097 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2099 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2100 index argument now works properly on the Alpha platform.
2101 * bug fix: floating point exception treatment on the Alpha platform
2103 * bug fix: FILE-POSITION works much better on string input and
2104 output streams. (thanks to Nikodemus Siivola)
2105 * bug fix: many threading/garbage collection symptoms sorted.
2106 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2107 might be pseudo-atomic.
2108 * internal change: Stopping for GC is now done with signals not
2109 ptrace. GC is now done in whichever thread wanted it, instead of
2111 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2113 * bug fix: result form in DO is not contained in the implicit
2115 * incompatible change: ICR structure is changed; the value part of
2116 CONTINUATION is now called LVAR; corresponding functions are
2117 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2118 * added type deriver for ISQRT (thanks to Robert E. Brown).
2119 * bug fix: better support for loading from the command line when an
2120 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2122 * fixed some bugs revealed by Paul Dietz' test suite:
2123 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2124 ** ROUND and FROUND now give the right answer when given very
2125 small float arguments.
2126 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2128 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2129 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2130 ** LCM should return a non-negative integer.
2131 ** PARSE-INTEGER returned the index of a terminator instead of the
2132 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2133 ** PARSE-INTEGER returned an incorrect index being applied to a
2135 ** LCM with two arguments of 0 returns 0 rather than signalling
2137 ** unsigned addition of a 32-bit constant with the high bit set no
2138 longer causes an internal compiler error on the x86.
2139 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2140 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2141 with negative last argument.
2142 ** byte specifiers with zero size and position no longer cause
2143 an error during type derivation.
2144 ** bignum multiplication on the Alpha platform now returns the
2146 * porting: The system now builds on SuSE AMD64, although it still
2147 generates a 32-bit binary.
2148 * .fasl file incompatibility: The fasl file version number has
2149 been incremented (because of the changes to internal compiler
2150 data structures referred to above).
2152 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2153 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2154 more accurately, on the Darwin kernel running on PowerPC hardware.
2155 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2156 * Compiler code deletion notes now signal a condition of type
2157 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2158 an associated MUFFLE-WARNING restart.
2159 * The compiler now performs limited argument count validation of
2160 constant format strings in FORMAT, and where appropriate in ERROR,
2161 CERROR and WARN. (thanks to Gerd Moellmann)
2162 * New ASDF-INSTALL contrib can be used for automatic download and
2163 installation of third-party Lisp code from CCLAN or other sites
2165 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2166 which forces another thread to execute a function supplied by the
2168 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2169 now accept and act upon their :ELEMENT-TYPE keyword argument.
2170 (reported by Martin Atzmueller, Edi Weitz)
2171 * bug fix: FILE-POSITION now accepts position designators up to
2172 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2173 is the greater. (thanks to Patrik Nordebo)
2174 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2175 Kalle Olavi Niemitalo)
2176 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2177 printable. (reported by Eric Marsden)
2178 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2179 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2180 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2181 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2182 (thanks to Stig Sandoe)
2183 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2184 UNREAD-CHAR now correctly includes the unread character in the
2185 target sequence. (thanks to Gerd Moellmann)
2186 * bug fix in threaded builds: the system can now be suspended and
2187 resumed by shell job control with minimal disruption.
2188 * bug fixes in times and timezones >2038AD
2189 * better handling of "where is GNU make?" problem in build scripts
2190 (thanks to Nikodemus Siivola)
2191 * new optimization: inside a named function any reference to a
2192 function with the same name is considered to be a self-reference;
2193 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2194 optimization quality.
2195 * new optimization on x86: logical functions and + now have
2196 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2197 used when the result is truncated to 32 bits.
2198 * VALUES declaration is partially enabled.
2199 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2200 * fixed some bugs revealed by Paul Dietz' test suite:
2201 ** The system now obeys the constraint imposed by
2202 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2203 types form a lattice under type intersection.
2204 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2205 ** ASSOC now ignores NIL elements in an alist.
2206 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2207 and (1+ MOST-POSITIVE-FIXNUM) answers.
2208 ** The addition of a method with invalid qualifiers to a generic
2209 function does not cause an error to be signalled immediately;
2210 a warning is signalled, and the error is generated only on
2211 calling the generic function.
2212 * changed .fasl file version number, in part to add type codes for
2213 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2214 obscure ANSI requirements
2216 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2217 * fixed bug 148: failure to inline-expand a local function left
2218 garbage, confusing the compiler.
2219 * fixed bugs 3cd: structure slot readers perform type check if the
2220 slot can have an invalid value (i.e. it is either not initialized
2221 or can be written with a less specific slot writer).
2222 * bug fix: the compiler now traps array references to elements off
2223 the end of an array; previously, the bounds checking in some
2224 circumstances could go off-by-one.
2225 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2227 * type declarations for array element types now obey the description
2228 on the CLHS page "Declaration TYPE", as per discussions on
2229 sbcl-help around 2003-05-08. This means that a declaration
2230 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2231 declaration, all references to BAR will be asserted or assumed
2232 (with THE, so dependent on compiler policy) to involve objects of
2233 type FOO. Note that no such declaration is implied in
2234 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2235 * declared types of functions from the "Conditions"
2236 chapter. (reported by Paul Dietz)
2237 * bug fix: CERROR accepts a function as its first argument.
2238 * bug fix: NTH an NTHCDR accept a bignum as index
2239 arguments. (reported by Adam Warner)
2240 * optimization: character compare routines now optimize comparing
2241 against a constant character. (reported by Gilbert Baumann)
2242 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2243 argument now works properly on the MIPS platform.
2244 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2245 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2246 declared to have a functional type, does not cause an error
2248 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2249 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2250 * fixed bug: if last continuation of a deleted block has a
2251 destination, this destination should be deleted too. (reported by
2253 * fixed a bug in the bootstrap process: the host compiler's values
2254 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2255 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2256 test case from Patrik Nordebo)
2257 * improved the ability of the disassembler on the PPC platform to
2258 provide helpful disassembly notes.
2259 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2260 the class in more cases than previously.
2261 * bug fix: FILE-POSITION now understands :START and :END for
2262 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2263 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2264 8) data. (thanks to David Lichteblau)
2265 * bug fix: it is possible to add a method to a generic function
2266 without lambda list.
2267 * bug fix: reader failed to signal END-OF-FILE inside an
2268 object representation. (reported by Nikodemus Siivola)
2269 * fixed some bugs revealed by Paul Dietz' test suite:
2270 ** LAST and [N]BUTLAST should accept a bignum.
2271 ** condition slot accessors are methods.
2272 ** (VECTOR NIL) is a subtype of STRING.
2274 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2275 * minor incompatible change: some nonsensical specialized lambda
2276 lists (used in DEFMETHOD) which were previously ignored now signal
2278 * minor incompatible change: the system is now aware of the types of
2279 variables in the COMMON-LISP package, and will signal errors for
2280 most violations of these type constraints (where previously they
2281 were silently accepted).
2282 * minor incompatible change: COMPILE-FILE now uses the freedom
2283 afforded (ANSI 3.2.2.3) to use derived function types for
2284 functions defined in the same file. This also permits the system
2285 to warn on static type mismatches and function
2286 redefinition. (Currently it does not work with high DEBUG level.)
2287 * minor incompatible change: VALUES declaration is disabled.
2288 * When issuing notes, the compiler now signals a condition of type
2289 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2290 restart for use in user handlers. It is expected that the
2291 COMPILER-NOTE condition will eventually become a condition
2292 supertype to a hierarchy of note types, which will then be
2293 handleable in a similar fashion. However, at the moment, no such
2294 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2295 but it's an implementation detail, not a classification for the
2297 * Changes in type checking closed the following bugs:
2298 ** type checking of unused values (192b, 194d, 203);
2299 ** template selection based on unsafe type assertions (192c, 236);
2300 ** type checking in branches (194bc).
2301 * A short form of VALUES type specifier has ANSI meaning (it has
2302 increased the number of situations when SBCL cannot perform type
2304 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2305 keywords or constants is permissible.
2306 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2307 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2308 classes. (thanks to Antonio Martinez)
2309 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2310 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2311 argument to a pretty-print stream. Instead, it leaves any such
2312 operations to DESCRIBE-OBJECT methods.
2313 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2315 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2316 lambda lists. (thanks to David Lichteblau)
2317 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2318 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2319 (thanks to Gerd Moellmann)
2320 * fixed bug 246: increased compilation speed of long
2321 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2323 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2325 * DEFCONSTANT now throws a condition of type
2326 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2327 constant to a non-EQL value; CONTINUE and ABORT restarts
2328 respectively change and preserve the value.
2329 * fixed bug 63: The code walker, part of the implementation of CLOS,
2330 is now better at handling symbol macros.
2331 * bug fix: There is no longer an internal implementation type named
2332 CL:LENGTH. (reported by Raymond Toy)
2333 * bug fix: In macro-like defining macros/special operators the
2334 implicit block does not enclose the lambda list.
2335 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2336 enclosed in parentheses) are not suitable as type specifiers, and
2337 their use properly signals an error now.
2338 * bug fix: An argument count mismatch for a type specifier in code
2339 being compiled no longer causes an unhandled error at compile
2340 time, but signals a compile-time warning.
2341 * fixed simple vector readable printing
2342 * bug fix: DESCRIBE takes more care over whether the class
2343 precedence list slot of a class is bound before accessing it.
2344 (reported by Markus Krummenacker)
2345 * bug fix: FORMATTER can successfully compile pretty-printer format
2346 strings which use variants of the ~* directive inside.
2347 * bug fix: SEARCH now applies its TEST predicate to the elements of
2348 the arguments in the correct order. (thanks to Wolfhard Buss)
2349 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2350 in type inference. (thanks to Robert E. Brown)
2351 * bug fix: Reading in symbols with an explicit package name of ""
2352 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2353 * fixed some bugs revealed by Paul Dietz' test suite:
2354 ** NIL is now allowed as a structure slot name.
2355 ** Arbitrary numbers, not just REALs, are allowed in certain
2356 circumstances in LOOP for-as-arithmetic clauses.
2357 ** Multiple class redefinitions before slot access no longer
2358 causes a type error.
2359 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2360 association between the name and a class.
2361 ** Generic functions with non-standard method-combination and over
2362 five methods all of which return constants no longer return NIL
2363 after the first few invocations. (thanks to Gerd Moellmann)
2364 ** CALL-NEXT-METHOD with no arguments now passes the original
2365 values of the arguments, even in the presence of assignment.
2366 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2368 ** DISASSEMBLE works with closures and funcallable instances.
2369 ** ADD-METHOD now returns the generic function, not the new method.
2370 ** FIND-METHOD signals an error if the lengths of the specializers
2371 is incompatible with the generic function, even if the ERRORP
2373 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2374 which its argument is a member.
2375 ** DEFCLASS only redefines the class named by its class-name
2376 argument if that name is the proper name of the class;
2377 otherwise, it creates a new class.
2378 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2379 of the UNBOUND-SLOT condition to the name of the slot.
2380 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2381 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2382 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2383 SLOT-MAKUNBOUND in the specified fashion.
2385 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2386 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2387 cross-compilation host. As a consequence, we can now bootstrap our
2388 way up to SBCL starting with a bare gcc toolchain and human-readable
2389 source code (first the source to CLISP, then the source to SBCL).
2390 * A contributed module containing a partial implementation of the
2391 simple-streams interface has been included. (thanks to Rudi
2393 * A contributed module implementing the RFC1321 Message Digest
2394 Algorithm, known as MD5, has been included.
2395 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2396 no longer has any effect, as the code controlled by this feature
2397 has been deleted. (As far as we know, no-one has ever built using
2398 this feature, and its semantics were confused in any case).
2399 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2400 work on conditions (as required by the ANSI specification),
2401 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2402 expected behaviour on conditions. Users should note, however,
2403 that such behaviour is not required by the ANSI specification,
2404 and so use of this behaviour may render their code unportable.
2405 * fixed some bugs revealed by Paul Dietz' test suite:
2406 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2408 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2409 ** SXHASH on condition objects no longer returns NIL.
2410 ** :ALLOCATION :CLASS slots are better treated; their values are
2411 updated on class redefinition, and initforms inherited from
2412 superclasses are applied.
2413 ** REMOVE-METHOD returns its generic function argument even when
2414 no method was removed.
2415 ** SHARED-INITIALIZE now initializes the values of the requested
2416 slots, including those with :ALLOCATION :CLASS.
2417 ** ALLOCATE-INSTANCE now works on structure classes defined via
2418 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2420 ** SLOT-EXISTS-P now works on conditions, as well as structures
2422 ** MAKE-LOAD-FORM now has the required methods on
2423 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2424 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2425 keyword, and now implements the SLOT-NAMES argument.
2426 ** methods with &OPTIONAL arguments no longer allow too many
2427 arguments to be passed in the call without error.
2428 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2429 option is consistent with the required arguments of the generic
2430 function lambda list.
2431 * bug fix: REQUIRE accepts a string designator. (Thanks to
2433 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2434 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2435 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2436 * bug fix: The compiler now checks for duplicated variables in macro
2438 * bug fix: SETQ on globals returns the correct value.
2439 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2440 not a COMPILER-ERROR (followed by some other strange error on
2441 choosing the CONTINUE restart).
2442 * bug fix: make.sh and friends are now more consistent in the way that
2443 they look for GNU "make".
2445 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2446 * experimental native threads support (on x86 Linux >=2.4 only).
2447 This is not compiled in by default: you need to add :SB-THREAD to
2448 the target features. See the "Beyond ANSI" chapter of the manual
2450 * fix for longstanding nonANSIism: The old distinction between
2451 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2452 The return value from CL:FIND-CLASS is now a CLOS class, and
2453 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2454 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2455 * An interface to the MetaObject Protocol, as described in Kiczales,
2456 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2457 MIT Press, 1991, is available from the SB-MOP package.
2458 * incompatible change: the SB-PCL package should now be considered
2459 a private implementation detail, and no longer a semi-private MOP
2461 * minor incompatible change: due to rearrangement for threads, the
2462 control stack and binding stack are now allocated at arbitrary
2463 addresses instead of being hardcoded per-port. Users affected by
2464 this probably have to be doing advanced things with shared
2465 libraries, and will know who they are.
2466 * minor incompatible change: Previously, all --eval forms used were
2467 processed with READ before any of them were processed with EVAL.
2468 Now each --eval form is processed with both READ and EVAL before
2469 the next --eval form is processed. (Thus package operations like
2470 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2471 work as the user might reasonably expect.)
2472 * minor incompatible change: *STANDARD-INPUT* is now only an
2473 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2475 * minor incompatible change: Y-OR-N-P is now character-oriented,
2476 not line oriented. Also, YES-OR-NO-P now works without errors.
2477 (thanks to Antonio Martinez)
2478 * sb-aclrepl module improvements: an integrated inspector, added
2479 repl features, and a bug fix to :trace command.
2480 * Known functions, which cannot be open coded by the backend, are
2481 considered to be able to check types of their arguments. (fixing
2482 a bug report by Nathan J. Froyd)
2483 * fixed a bug in computing method discriminating functions: It is
2484 now possible to define methods specialized on classes which have
2485 forward-referenced superclasses. (thanks to Gerd Moellmann)
2486 * fixed evaluation order in optional entries (reported by Gilbert
2488 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2489 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2490 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2491 required-by-AMOP NAME argument, as well as CLASS and
2492 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2493 * fixed bug 20: DEFMETHOD can define methods using names that are
2494 not the proper names of classes to designate class specializers.
2495 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2496 Unix file descriptors, instead of blowing up. (thanks to Antonio
2498 * Garbage collection refactoring: user-visible change is that a
2499 call to the GC function during WITHOUT-GCING will not do garbage
2500 collection until the end of the WITHOUT-GCING. If you were doing
2501 this you were probably losing anyway.
2502 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2503 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2504 TYPEP the latter but not the former.
2505 * The compiler issues a full WARNING for calls to undefined functions
2506 with names from the CL package.
2507 * MAP-INTO for a vector destination is open coded. (reported by
2508 Brian Downing on c.l.l)
2509 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2510 documentation string.
2511 * fixed some bugs revealed by Paul Dietz' test suite:
2512 ** COPY-ALIST now signals an error if its argument is a dotted
2514 ** Condition slots are now accessed more correctly in the presence
2515 of multiple initargs for a given slot.
2516 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2517 exclude restarts of the same name associated with a different
2519 ** DEFCLASS of forward-referenced classes with another
2520 forward-referenced class in the superclasses list no longer
2522 ** Condition slots are now initialized once each, not multiple
2523 times. (thanks to Gerd Moellmann)
2524 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2526 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2527 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2528 arguments contain duplicated elements.
2529 ** RESTART-CASE understands local macros.
2530 ** RESTART-CASE associates exactly its own restarts with a condition.
2531 ** ENDP in safe mode checks its argument to be of type LIST.
2532 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2533 in question is unbound.
2534 ** Optimized MAKE-INSTANCE functions no longer cause internal
2535 assertion failures in the presence of duplicate initargs.
2536 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2537 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2538 propagated even in the ctor optimized implementation of
2540 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2542 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2543 * a better implementation of SXHASH on (simple) bit vectors,
2544 measured both in execution speed and in distribution of results
2545 over the positive fixnums, has been installed. Likewise, a better
2546 implementation of EQUAL for simple bit vectors is now available.
2547 * fixed CEILING optimization for a divisor of form 2^k.
2548 * fixed bug 240 (emitting extra style warnings "using the lexical
2549 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2550 by Antonio Martinez)
2551 * fixed SXHASH, giving different results for NIL depending on type
2552 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2553 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2554 variable. (found by Rolf Wester)
2555 * SBCL does not ignore type declarations for special
2556 variables. (reported by rif on c.l.l 2003-03-05)
2557 * some bug fixes in contrib/sb-aclrepl/
2558 * fixed some bugs revealed by Paul Dietz' test suite:
2559 ** a bug in the CONS type specifier, whereby the CAR and CDR
2560 types got intertwined, has been fixed;
2561 ** the type system is now able to reason about the interaction
2562 between INTEGER and RATIO types more completely;
2563 ** APPEND, [N]REVERSE and NRECONC check that those their
2564 arguments, which must be proper lists, are really so;
2565 ** An array specialized to be unable to hold elements has been
2566 implemented, as required -- yes, really -- by ANSI;
2567 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2568 on malformed property lists;
2570 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2571 * incompatible packaging change: in line with Unix convention,
2572 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2573 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2574 * REQUIRE and PROVIDE are now optionally capable of doing something
2575 useful. See the documentation string for REQUIRE.
2576 * infrastructure for a managed SBCL contrib system: contributed
2577 modules in this release include:
2578 ** the ASDF system definition facility;
2579 ** an interface to the BSD Sockets API;
2580 ** an ACL-like convenience interface to the repl;
2581 (thanks to Kevin Rosenberg)
2582 ** an implementation of ROTATE-BYTE, with efficient implementation
2584 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2585 gives a closer approximation to the right answer than previously.
2586 (thanks to Raymond Toy)
2587 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2588 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2589 optional environment arguments, as required by ANSI.
2590 * fixed bugs in other functions taking environment objects, allowing
2591 calls with an explicit NIL environment argument to be compiled
2593 * fixed bug 228: primary return values from
2594 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2595 COMPILE or FUNCTION.
2596 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2597 :TYPE will no longer signal errors on innocuous objects.
2598 * fixed bug 231b: SETQ is better at respecting type declarations in
2599 the lexical environment.
2600 * fixed a bug in DEFCLASS: classes named by symbols with no or
2601 unprintable packages can now be defined.
2602 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2603 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2604 * fixed compiler failure related to checking types of functions.
2605 (reported by Robert E. Brown)
2606 * the compiler is now much more consistent in its error-checking
2607 treatment of bounding index arguments to sequence functions: in
2608 (SAFETY 3) code, errors will be signalled in almost all cases if
2609 invalid sequence bounding indices are passed to functions defined
2610 by ANSI to operate on sequences.
2611 * fixed a bug in the build procedure: documentation of SBCL-specific
2612 packages is now preserved and available in the final Lisp image.
2613 * lifted FDEFINITION lookup out of loops in the implementation of
2614 many list operations. (thanks to Robert E. Brown)
2615 * fixed a bug in the reader: the #n# reader macro now works for
2616 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2617 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2618 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2619 E. Brown and Tony Martinez respectively)
2620 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2621 SLOT-BOUNDP: in particular, it is now able to optimize them much
2622 better, and is now not vulnerable to having packages renamed.
2623 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2624 always called when a slot is not present in an instance. (thanks
2626 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2627 longer signal an error on unreading a character following EOF on
2628 the previous constituent stream. (thanks to Tony Martinez)
2629 * fixed some bugs revealed by Paul Dietz' test suite:
2630 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2631 not just nonnegative fixnums;
2632 ** the logical bit-array operators such as BIT-AND now accept an
2633 explicit NIL for their "opt-arg" argument (to indicate a
2634 freshly-consed result bit-array);
2635 ** ELT now signals an error on an invalid sequence index in safe
2637 ** the type system is now cleverer about negations of numeric
2638 types, and consequently understands the BIGNUM and RATIO types
2640 ** the type system is now cleverer about the interaction between
2641 INTEGER and RATIO types: while bugs still remain, many more
2642 cases are accurately computed;
2643 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2644 if it is in the last clause;
2645 ** CONSTANTLY now correctly returns a side-effect-free function in
2647 ** DECLARE is no longer treated as a special-operator; in
2648 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2649 * incremented fasl file version number due to the change in the
2650 DEFSTRUCT-SLOT-DESCRIPTION structure.
2652 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2653 * minor incompatible change: code processed by the "interpreter" or
2654 EVAL now has a compilation optimization policy of (DEBUG 2)
2655 (changed from (DEBUG 1)) to improve debuggability of interactive
2656 development, and to allow the use of the debug RETURN command in
2658 * an experimental implementation of the RETURN command for the
2659 debugger has been included. (thanks to Frederik Kuivinen)
2660 * fixed bug 62: constraints were not propagated into a loop.
2661 * fixed bug in embedded calls of SORT (reported and investigated by
2663 * fixed some bugs revealed by Paul Dietz' test suite:
2664 ** printing and reading of arrays with some dimensions having
2665 length 0 (thanks to Gerd Moellmann);
2666 ** BOA constructor with &AUX argument without a default value does
2667 not cause a type error;
2668 ** CONSTANTP now returns true for all self-evaluating objects.
2670 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2671 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2672 accessors that are related by inheritance, as specified in the
2673 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2674 Valtteri Vuorikoski)
2675 * The compiler is now able to inline functions that were defined in
2676 a complex lexical environment (e.g. inside a MACROLET).
2677 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2679 * Support for the upcoming FreeBSD-5.0 release has been included.
2680 (thanks to Dag-Erling Smorgrav)
2681 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2682 effect when it is not in a toplevel context.
2683 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2684 stand a better chance of being correct. (thanks to Gerd
2686 * fixed bug in COERCE, which now signals an error on coercing a
2687 rational to a bounded real type which excludes the expected
2689 * The compiler is now able to derive types more accurately from the
2690 COERCE and COMPILE functions.
2691 * fixed bug 223: functional binding is considered to be constant
2692 only for symbols in the CL package.
2693 * fixed bug 231: SETQ did not check the type of a variable being set
2694 (reported by Robert E. Brown)
2695 * A new optimization for MAKE-INSTANCE has been included, fixing
2696 various bugs (including relating to :ALLOCATION :CLASS slots and
2697 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2699 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2700 ** As required by ANSI, LOOP now disallows anonymous collection
2701 clauses such as COLLECT I in conjunction with aggregate boolean
2702 clauses such as THEREIS (= I 1);
2703 ** LOOP now signals an error when any variable is reused in the
2704 same loop (including the potentially useful construct analogous
2705 to WITH A = 1 WITH A = (1+ A);
2706 ** IT is only a special loop symbol within the first clause of a
2707 conditional loop clause;
2708 ** LOOP with a typed iteration variable over a hashtable now
2709 signals a type error iff it should.
2710 * fixed some other bugs revealed by Paul Dietz' test suite:
2711 ** FILE-STREAM now names the class previously known as FD-STREAM;
2712 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2713 argument) no longer signals an error;
2714 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2715 of the slot symbol, rather than using the current package
2716 ((:CONC-NAME "") continues to intern the slot's name in the
2718 * incremented fasl file version number, because of the incompatible
2719 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2720 of the new implementation of DEFINE-COMPILER-MACRO.
2722 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2723 * Support for building SBCL for MIPS platforms running in
2724 little-endian mode has now been checked in, and basic
2725 functionality on said platforms verified.
2726 * minor incompatible change: PCL now records the pathname of a file
2727 in which methods and the like are defined, rather than its
2729 * minor incompatible change: TRUENAME now considers the truename of
2730 a file naming a directory to be the pathname with :DIRECTORY
2731 component indicating that directory.
2732 * minor incompatible change: a NAMED clause in the extended form of
2733 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2734 reason for the previous behaviour is unclear.
2735 * more systematization and improvement of CLOS and MOP conformance
2736 in PCL (thanks to Gerd Moellman and Pierre Mai):
2737 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2739 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2740 primary methods with no specializers;
2741 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2743 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2744 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2745 FORWARD-REFERENCED-CLASSes; error reporting on
2746 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2748 ** SXHASH on CLOS instances now uses a slot internal to the
2749 instance to return different numbers on distinct instances,
2750 while preserving the same return value through invocations of
2752 ** DEFMETHOD signals errors when methods with longer incongruent
2753 lambda lists are added to generic functions;
2754 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2755 CLASS, as specified in AMOP;
2756 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2757 on the order returned by the primary method for classes of
2758 class STANDARD-CLASS;
2759 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2760 * fixed some bugs shown by Paul Dietz' test suite:
2761 ** DOLIST puts its body in TAGBODY;
2762 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2764 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2765 value producing form;
2766 ** if more variables are given to PROGV than values, extra
2767 variables are bound and made to have no value;
2768 ** NSUBSTITUTE on list arguments gets the right answer with
2770 ** ELT signals an error of type TYPE-ERROR when the index argument
2771 is not a valid sequence index;
2772 ** LOOP signals (at macroexpansion time) an error of type
2773 PROGRAM-ERROR when duplicate variable names are found;
2774 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2775 ** FUNCALL of special-operators now cause an error of type
2777 ** PSETQ now works as required in the presence of side-effecting
2778 symbol-macro places;
2779 ** NCONC accepts any object as its last argument;
2780 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2782 ** loop-for-as-package does not require a package to be explicitely
2784 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2786 * fixed bug 166: compiler preserves "there is a way to go"
2787 invariant when deleting code.
2788 * fixed bug 172: macro lambda lists with required arguments after
2789 &REST arguments now cause an error to be signalled. (thanks to
2791 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2792 bug. (thanks to Matthew Danish)
2793 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2795 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2796 arguments are lexically rebound. (thanks to Gerd Moellmann and
2798 * fixed bug 194: error messages are now more informative when there
2799 is no primary method applicable in a call to a generic
2800 function. (thanks to Gerd Moellmann)
2801 * fixed bug in command line argument checking (thanks to Julian
2803 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2804 (thanks to Matthew Danish)
2805 * incremented fasl file version number, because of the
2806 SXHASH-related changes in the layout of CLOS data structures
2808 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2809 * minor incompatible change: The runtime (the Unix executable named
2810 "sbcl") is now much pickier about the .core files it will load.
2811 Essentially it now requires .core files to descend from the same
2812 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2813 as the runtime does. (The intent is to prevent the crashes which
2814 can occur, and which can even be reported as mysterious failures,
2815 when people patch the sources or change the build parameters
2816 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2817 sbcl and .core files.)
2818 * fixed bug: VALUES-LIST is no longer optimized away.
2819 * fixed bug 142: The FFI conversion of C string values to Lisp
2820 string values no longer conses excessively. (thanks to Nathan
2821 Froyd porting Raymond Toy's fix to CMU CL)
2822 * began to systematize and improve MOP conformance in PCL (thanks to
2823 Nathan Froyd, Gerd Moellman and Pierre Mai):
2824 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2826 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2827 ** FINALIZE-INHERITANCE is now called on class finalization;
2828 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2829 argument precedence order.
2830 * fixed bug 202: The compiler no longer fails on functions whose
2831 derived types contradict their declared type.
2832 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2833 so it can be non-toplevel.
2834 * The fasl file version number has changed (because of the new
2835 implementation of DEFMACRO).
2836 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2837 safe code, that any length requirement by their type-specifier
2838 argument is valid. The exceptions to this are described in bug
2840 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2841 functions now check the types of their inputs as required by ANSI.
2842 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2843 introduced symbol is DECLAREd to be SPECIAL.
2844 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2845 * fixed bug: :COUNT argument to sequence functions may be negative
2846 * fixed bug: body of DO-SYMBOLS may contain declarations
2847 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2848 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2849 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2850 symbol macro only once
2851 * fixed printing of call frame when argument list is unavailable
2852 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2853 * compiler no longer signals WARNING on unknown keyword
2856 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2857 * A beta-quality port to the mips architecture running Linux,
2858 based on the old CMUCL backend, has been made. It has been tested
2859 on a big-endian kernel, and works sufficiently well to be able to
2860 rebuild itself; it has not been tested in little-endian mode.
2861 * fixed an inconsistency between gencgc.c and purify.c which made
2862 dumping/loading .core files unreliable
2863 * fixed bug 120a: The compiler now deals correctly with IFs where
2864 the consequent is the same as the alternative, instead of
2865 misderiving the return type. (thanks to Alexey Dejneka)
2866 * fixed bug 113: Logical pathnames are now dumpable (the logical
2867 host is resolved at load-time, throwing an error if it is not
2869 * fixed bug 174: FORMAT's error message is slightly clearer when a
2870 non-printing character is used in a format directive.
2871 * fixed several bugs in compiler checking of type declarations, i.e.
2872 violations of the Python "declarations are assertions" principle
2873 (thanks to Alexey Dejneka)
2874 * fixed several bugs in PCL's error checking (thanks to Gerd
2876 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2878 * fixed bug in compilation of functions as first class values
2879 (thanks to Antonio Martinez-Shotton)
2880 * The compiler's handling TYPE-ERRORs which it can prove will
2881 inevitably happen at runtime has been cleaned up and corrected
2882 in several ways. (thanks to Alexey Dejneka)
2883 * improved argument type checking for various basic arithmetic
2884 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2885 lavished on them in the past that they can be compiled in many
2886 ways in different special cases
2887 * fixed bug 181: compiler checks validity of user supplied type
2889 * cleaned up code flushing in optimization: Function calls which
2890 should signal errors for safety purposes (e.g. which ANSI says
2891 should signal errors when their arguments are of incorrect type)
2892 are no longer optimized away.
2893 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2894 * incremented fasl file version number, because changes in the
2895 implementation of sequence functions like COERCE caused
2896 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2897 in old inline expansions) to become undefined. (Actually these
2898 changes were later undone, so we might very well be binary
2899 compatible with 0.7.7 after all, but leaving the version number
2900 incremented seemed like the simplest and most conservative
2903 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2904 * An alpha-quality port to the parisc architecture running Linux,
2905 based on the old CMUCL backend, has been made. This, even more so
2906 than the other backends, should be considered still a work in
2907 progress; known problems include that the Linux kernel in 64-bit
2908 mode does not propagate the correct sigcontext structure to
2909 userspace, and consequently SBCL on a parisc64 kernel will not
2911 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2912 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2913 suppressed.) Also now that the compiler is looking at declarations
2914 in the environment, it checks optimization declarations as well,
2915 and suppresses inlining when (> DEBUG SPEED).
2916 * More fixes have been made to treatment of floating point exception
2917 treatment and other Unix signals. In particular, floating point
2918 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2919 * The detection and handling of control stack exhaustion (infinite
2920 or very deeply nested recursion) has changed. Stack exhaustion
2921 detection is now done by write-protecting pages at the OS level
2922 and applies at all optimization settings; when found, a
2923 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2924 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2925 be caught using IGNORE-ERRORS.
2926 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2927 correctly when transforming recursive LABELS functions to LETs.
2928 (thanks to Alexey Dejneka)
2929 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2930 that are names of constants or global variables.
2931 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2932 alien routines with docstrings.
2933 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2934 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2936 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2937 (thanks to Eric Marsden)
2938 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2939 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2940 * Bug fix: The SPARC backend can now compile functions involving
2941 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2942 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2943 to a functions expecting a PATHNAME-DESIGNATOR.
2944 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2945 lambda lists. (thanks to Alexey Dejneka)
2946 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2947 functions so that EXPORTing the name of the class doesn't cause
2948 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2949 to Antonio Martinez for reporting this)
2950 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2951 constant-folded without failing an assertion. (thanks to Einar
2952 Floystad Dorum for reporting this)
2953 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2954 types are now dealt with more correctly by the compiler.
2955 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2956 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2957 behaviour analogous to RENAME-FILE. This puts its behaviour more
2958 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2960 * The fasl file version number has changed again. (because of the
2961 bug fix involving the names of PCL MAKE-INSTANCE functions)
2963 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2964 * bug fix: Floating point exceptions are treated much more
2965 consistently on the x86/Linux and PPC/Linux platforms.
2966 * Array initialization with :INITIAL-ELEMENT is now much faster for
2967 cases when the compiler cannot open code the array creation, but
2968 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2969 array accesses have also seen a speed increase.
2970 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2971 specified. (This was at the root of some bad interactions between
2972 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2974 * bug fix: Internal error arguments for undefined functions are now
2975 computed correctly on the PPC/Linux platform.
2976 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2977 Raymond Toy's patch for CMU CL)
2978 * Support for the Solaris 9 operating environment has been included
2979 (thanks to Daniel Merritt)
2980 * A very ugly but hopefully complete draft of the missing FFI chapter
2981 of the manual has been created by reformatting the corresponding
2982 CMU CL manual chapter into (currently very ugly and incoherent)
2983 DocBook and bringing it up to date for SBCL behavior. Thus, the
2984 manual is now essentially complete, at least by my extreme
2985 once-and-only-once standards, whereby it's acceptable to refer to
2986 the doc strings of SB-EXT functions as the primary documentation.
2987 * The fasl file version number has changed again, due to cleanup of
2988 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2989 is no longer a static symbol.)
2991 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2992 * SBCL now builds with OpenMCL (version 0.12) as the
2993 cross-compilation host; also, more progress has been made toward
2994 bootstrapping under CLISP.
2995 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2997 * bug 158 fixed: The compiler can now deal with integer loop
2998 increments different from 1; fixing this turned out also to fix
3000 * bug 169 fixed: no more bogus warnings about using lexical bindings
3001 despite the presence of perfectly good SPECIAL declarations (thanks
3002 to David Lichteblau)
3003 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3004 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3005 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3007 * bug 180 fixed: Method combination specifications no longer ignore
3008 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3009 * bug fix: Structure type predicate functions now check their argument
3010 count as they should.
3011 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3012 correctly. (thanks to Pierre Mai)
3013 * minor incompatible change: The --noprogrammer option is deprecated
3014 in favor of the new --disable-debugger option, which is very similar.
3015 (The major difference is that it takes effect at a slightly different
3016 time at startup, causing handling of errors in --sysinit and
3017 --userinit files will be affected differently.) The
3018 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3019 been added to allow this functionality to be controlled from ordinary
3020 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3021 maintainers, who might want to run non-interactive scripts to
3022 build SBCL cores which will later be used interactively.)
3023 * minor incompatible change: The LOAD function no longer, when given
3024 a wild pathname to load, loads all files matching that pathname.
3025 Instead, an error of type FILE-ERROR is signalled.
3027 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3028 * bug 147 fixed: The compiler preserves its block link/count
3029 invariants more correctly now so that it doesn't crash. (thanks
3031 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3033 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3034 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3035 no-value-for-this-character as NIL (as specified by ANSI).
3036 * HOST-NAMESTRING on physical pathnames now returns a string that is
3037 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3038 (thanks to Christophe Rhodes)
3039 * The Alpha port handles icache flushing more correctly. (thanks to
3041 * More progress has been made toward bootstrapping under CLISP. (thanks
3042 to Christophe Rhodes)
3043 * The fasl file format has changed again, because dynamic loading
3044 on OpenBSD (which has non-ELF object files) motivated some cleanups
3045 in the way that foreign symbols are transformed and passed around.
3046 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3047 is no longer treated as whitespace by the reader, but instead as
3048 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3049 |A
\7fB|, instead of A as it used to.
3051 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3052 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3053 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3054 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3055 this as of 0.7.1.45, but was left out of the previous news section
3056 (thanks to Dan Barlow)
3057 * SBCL now runs on the Solaris operating system on SPARC architectures
3058 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3059 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3060 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3061 * SPARC backend cleanups, allowing builds of cores optimized for V8
3062 and V9 SPARCS, and also emission of code targeted to a particular
3063 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3065 * SBCL is closer to bootstrapping under CLISP, thanks to various
3066 fixes by Christophe Rhodes.
3067 * The fasl file format has changed again, to allow the compiler's
3068 INFO database to support symbol macros.
3069 * The user manual (in doc/) is formatted into HTML more nicely.
3070 (thanks to coreythomas)
3071 * The system is smarter about SUBTYPEP relationships, especially
3072 those involving NOT types (including types such as ATOM which are
3073 represented internally using NOT types). Thus SUBTYPEP is less
3074 likely to return (VALUES NIL NIL) in general, and in particular
3075 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3077 * The fasl file format has changed again, because the internal
3078 representation of types now includes a new slot to support the new
3079 SUBTYPEP-of-NOT-types logic.
3080 * (not a change in the main branch of SBCL, but a related prototype
3081 which can hopefully be merged into the main branch of SBCL in the
3082 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3083 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3084 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3085 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3086 dispatch macro character combinations. (thanks to Alexey Dejneka)
3087 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3088 that superficially look like logical namestrings correctly.
3089 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3091 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3092 default supplied-p) for &optional and &key arguments. (thanks to
3095 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3096 * incompatible change: The compiler is now less aggressive about
3097 tail call optimization, doing it only when (> SPACE DEBUG) or
3098 (> SPEED DEBUG). (This is an incompatible change because there are
3099 programs which relied on the old CMU-CL-style behavior to optimize
3100 away their unbounded recursion which will now die of stack overflow.)
3101 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3102 for non-GENCGC systems has been increased to 20M (since that
3103 seems much closer to the likely performance optimum for modern
3104 systems than the old 4M value was)
3105 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3106 and *DEBUG-PRINT-LEVEL*
3107 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3108 of CMU CL's support for SPARC, and various endianness and other
3109 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3110 * new syntactic sugar for the Unix command line: --load foo.bar is now
3111 an alternate notation for --eval '(load "foo.bar")'.
3113 ** The system now detects stack overflow and handles it gracefully,
3114 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3115 optimization settings. (This is a good thing in general, and
3116 its introduction in this version should be particularly timely
3117 for anyone whose code fails because of suppression of tail
3119 ** The system now hunts for the C variable "environ" in a more
3120 devious way, to avoid segfaults when the C library version differs
3121 between compile time and run time. (thanks to Christophe Rhodes)
3122 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3123 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3124 ** The compiler no longer issues bogus style warnings for undefined
3125 classes in the same source file as the DEFCLASSes which defined
3126 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3128 ** fixes in CONDITION class precedence list for undefined function
3129 errors (thanks to Alexei Dejneka)
3130 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3131 correctly. (thanks to Dan Barlow)
3132 ** portability fixes aiming at bootstrapping under CLISP (thanks
3133 to Dave McDonald and Christophe Rhodes)
3134 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3135 ** fixes in type translation and and type inference (thanks to
3137 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3138 ** various fixes in the new ports (thanks to Dan Barlow)
3139 * several changes related to debugging:
3140 ** suppression of tail recursion, as noted above
3141 ** stack overflow detection, as noted above
3142 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3143 is now the default. (For some time encapsulation has been more
3144 reliable than the breakpoint-based :ENCAPSULATE NIL
3145 implementation, at least on X86 systems; and I just noticed that
3146 encapsulation also seems closer to the spirit of the ANSI
3149 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3151 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3152 up properly again. (There was a packaging bug in 0.7.0 which
3153 left their definitions in SB-SYS::LOAD-FOREIGN and
3154 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3155 vital for most things which interface to C-level interfaces,
3156 like extensions working with sockets or databases or
3157 Perl-compatible regexes or whatever, and the need to fix
3158 this bug is the main reason that 0.7.1 was released so
3160 ** DEFGENERIC is now choosier about the methods it redefines, so that
3161 reLOADing a previously-LOADed file containing DEFGENERICs does
3162 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3163 cycle now works as it should. (thanks to Alexey Dejneka)
3164 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3165 fixed by Christophe Rhodes. (He actually submitted this patch
3166 months ago, and I delayed until after 0.7.0.)
3167 ** Bug 111 (internal compiler confusion about runtime checks on
3168 FUNCTION types) was fixed by Alexey Dejneka.
3169 * Some internal cleanups (getting rid of variables which aren't
3170 needed now that the byte interpreter is gone) caused the fasl
3171 file format number to change again.
3173 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3174 * major incompatible change: The default fasl file extension, i.e. the
3175 default extension for files produced by COMPILE-FILE, has changed
3176 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3178 ** There are many changes in the implementation of the compiler.
3179 SBCL is now essentially a compiler-only implementation of ANSI
3180 Common Lisp. EVAL still "interprets" a few special cases, but
3181 almost all the interesting cases are handled by creating
3182 a LAMBDA expression, calling COMPILE on it, then calling
3183 FUNCALL on the result.
3184 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3185 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3186 Since the code is newer, there might still be some new bugs
3187 (though not as many as before Martin Atzmueller's fixes:-). But
3188 the new code is substantially simpler and clearer, and hopefully
3189 any remaining bugs will be simpler, less fundamental, and more
3190 fixable then the bugs in the old code.
3191 ** The revised compiler is still a little unsteady on its feet.
3193 *** The debugging information it produces (particularly the names
3194 of FUNCTION objects) is sometimes much less useful than what
3195 the old compiler produced.
3196 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3197 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3198 within a MACROLET) has been temporarily weakened.
3199 ** There are new compiler optimizations for various functions:
3200 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3201 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3202 *** the math functions TRUNCATE, FLOOR, and CEILING
3203 *** the function-of-all-trades COERCE
3204 Mostly these should be transparent, but there's one
3205 potentially-annoying problem (bug 117): when the compiler
3206 inline-expands a function and does type analysis on the result,
3207 it can create control paths which have type mismatches, and
3208 when it can't prove that those control paths aren't taken,
3209 it will issue WARNINGs about the type mismatches. This is
3210 a particular problem in practice for the new sequence functions.
3211 It's not clear how this should be fixed, and for now, a
3212 workaround is given in the entry for 117 in the BUGS file.
3213 ** (Because of the interaction between the two previous items --
3214 occasional inlining problems and new inline expansions -- some
3215 of the new sequence function optimizations won't really kick in
3216 completely until debugging information, and then inlining, are
3217 straightened out in some future version.)
3218 * minor incompatible changes:
3219 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3220 DIRECTORY no longer implicitly promotes NIL slots of its
3221 pathname argument to :WILD. In particular, when you ask for the
3222 contents of a directory (which you used to be able to do without
3223 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3224 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3225 ** changes in behavior that ANSI explicitly defines to be
3226 implementation dependent:
3227 *** The new compiler-only implementation still conforms with ANSI,
3228 but acts a little different than before. Besides the obvious
3229 changes in performance tradeoffs (that the cost per form passed
3230 to EVAL has gone up, and the cost per form executed by EVAL
3231 has gone down), the behavior of the system changes a little
3232 because there are no longer any interpreted function objects.
3233 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3234 e.g. doing COMPILE on the output of interactive DEFUN is
3236 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3238 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3239 has changed from (:CL) to NIL.
3240 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3241 e.g. #\Newline and #\Tab, don't have names specified in the
3242 ANSI Common Lisp standard, is now based on their ASCII symbolic
3243 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3244 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3245 are no longer used for output.
3246 ** changes in internal implementation constants:
3247 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3248 4 million. (If your application spends a lot of time GCing and
3249 you have a lot of RAM, you might want to experiment with
3250 increasing it even more.)
3251 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3252 However, almost all old code should still continue to work without
3253 immediate update, as SB-C-CALL is now a (deprecated) nickname
3255 ** Old operator names in the style DEF-FOO are now deprecated in
3256 favor of new corresponding names DEFINE-FOO, for consistency with
3257 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3258 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3259 internal symbols, but a few supported extensions like
3260 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3261 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3262 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3263 etc. as you get deeper into recursive calls to the debugger
3264 command loop, instead of the old "5]", "5]]", "5]]]"
3265 sequence. (I was motivated to do this when squabbles between
3266 ILISP and SBCL left me very deeply nested in the debugger. In the
3267 short term, this change will probably provoke more ILISP/SBCL
3268 squabbles, but hopefully it will be an improvement in the long run.)
3269 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3270 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3271 function, since declaiming return types involving aliens is
3272 (1) annoyingly messy to do by hand and (2) vital to efficient
3273 compilation of code which calls such functions.
3274 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3275 longer reexported by the SB-EXT package. They're solely useful
3276 for alien code, so it seems more logical that you should get
3277 them from the SB-ALIEN package, not in SB-EXT.
3278 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3279 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3280 features. Instead, the code that they used to control is always
3281 built into the system.
3282 * many other bug fixes
3283 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3284 advantage of the new EVAL-WHEN stuff and to clean them up in
3285 general, and they are now more ANSI-compliant in a number of
3286 ways. Martin Atzmueller is responsible for a lot of this.
3287 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3289 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3290 *** correct ERROR type for various file operations
3291 *** some fixes for Lisp streams
3292 *** DEFMETHOD syntax checking
3293 *** changing old weird representation of debug information as
3294 strings (which, among their other deficiencies, don't transform
3295 correctly when you rename packages, and don't change their
3296 print representation when you change things like *PACKAGE*
3297 and *PRINT-LENGTH*) to symbols and lists of symbols
3298 He also made several improvements and fixed several bugs in DESCRIBE.
3299 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3301 *** misbehavior of WRITE-STRING/WRITE-LINE
3302 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3303 and several other LOOP problems as well
3304 *** DIRECTORY when similar filenames are present
3305 *** DEFGENERIC with :METHOD options
3306 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3307 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3308 *** argument ordering in FIND with :TEST option
3309 *** mishandled package designator argument in APROPOS-LIST
3310 *** various problems in the backquote readmacro
3311 *** a bug in APROPOS
3312 *** probably some others that I'm not describing very well here,
3313 since the CVS log documents them by reference to sbcl-devel
3314 messages, and the SourceForge archives aren't working well.:-(
3315 ** Dan Barlow improved the Alpha port (and is making progress on the
3316 PPC port, for those of you who think different).
3317 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3318 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3319 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3320 continues to work on the SPARC port (for those of you in a position
3321 to look down upon our little PC-compatible boxes from a great height).
3322 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3323 and uses that copy, rather than the current dynamic value, when
3324 it's trying to decide whether to truncate output. Thus e.g.
3325 (let ((*print-lines* 50))
3326 (pprint-logical-block (stream nil)
3328 (let ((*print-lines* 8))
3329 (print (aref possiblybigthings i) stream)))))
3330 should now truncate the logical block only at 50 lines, instead of
3331 often truncating it at 8 lines, as it did before.
3332 * The doc/cmucl/ directory, containing old CMU CL documentation
3333 from the time of the fork, is no longer part of the base system.
3334 SourceForge has shut down its anonymous FTP service, and with it
3335 my original plan for distributing the old CMU CL documentation
3336 there. For now, if you need these files you can download an old
3337 SBCL source release and extract them from it.
3338 * The fasl file version number changed again, for dozens of reasons,
3339 some of which are apparent above.
3341 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3342 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3343 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3344 patch, so that DISASSEMBLE output is much nicer.
3345 * The code in the SB-PROFILE package now seems reasonably stable.
3346 I still haven't decided what the final interface should look like
3347 (I'd like PROFILE to interact cleanly with TRACE, since both
3348 facilities use function encapsulation) but if you have a need
3349 for profiling now, you can probably use it successfully with
3350 the current CMU-CL-style interface.
3351 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3352 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3353 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3354 link now returns the dangling link itself, and for similar
3355 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3356 link itself. (In these cases the old code signalled an error and
3357 looped endlessly, respectively.) Thus, DIRECTORY now works even
3358 in the presence of dangling and cyclic symbolic links.
3359 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3360 is now a supported extension again, since the consensus on
3361 sbcl-devel was that it can be useful for ordinary development
3362 work, not just for debugging SBCL itself.
3363 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3364 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3365 that currently-defined functions might be redefined later with
3366 different return types.
3367 * Hash tables can be printed readably, as inspired by CMU CL code
3368 of Eric Marsden and SBCL code of Martin Atzmueller.
3369 * better error handling in CLOS method combination, thanks to
3370 Martin Atzmueller porting Pierre Mai's CMU CL patches
3371 * more overflow fixes for >16Mbyte I/O buffers
3372 * A bug in READ has been fixed, so that now a single Ctrl-D
3373 character suffices to cause end-of-file on character streams.
3374 In particular, now you only need one Ctrl-D at the command
3375 line (not two) to exit SBCL.
3376 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3378 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3379 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3380 does the right thing.
3381 * The implementation of some type tests, especially for CONDITION
3382 types, is now tidier and maybe faster, due to CMU CL code
3383 originally by Douglas Crosher, ported by Martin Atzmueller.
3384 * Some math functions have been fixed, and there are new
3385 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3386 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3387 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3388 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3389 likely to be built into sbcl-0.7.0.)
3390 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3391 CL, ported to SBCL by Martin Atzmueller)
3392 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3393 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3394 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3395 * There's a new slam.sh hack to shorten the edit/compile/debug
3396 cycle for low-level changes to SBCL itself, and a new
3397 :SB-AFTER-XC-CORE target feature to control the generation of
3398 the after-xc.core file needed by slam.sh.
3399 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3400 COMPILE-FILE is no longer supported, so that now every function
3401 gets an entry point, so that block compilation looks a little
3402 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3403 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3404 returns the number of bytes consed since the system started,
3405 rather than the number consed since the first time the function
3406 was called. (The new definition parallels ANSI functions like
3407 CL:GET-INTERNAL-RUN-TIME.)
3408 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3409 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3410 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3411 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3412 * Fasl file version numbers are now independent of the target CPU,
3413 since historically most system changes which required version
3414 number changes have affected all CPUs equally. Similarly,
3415 the byte fasl file version is now equal to the ordinary
3418 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3419 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3420 is no longer recognized. I apologize for this, because it was
3421 listed in SB-EXT as a supported extension, but I found that
3422 its existing behavior was poorly specified, as well as incorrectly
3423 specified, and it looked like too much of a mess to straighten it
3424 out. I have enough on my hands trying to get ANSI stuff to work..
3425 * many patches ported from CMU CL by Martin Atzmueller, with
3426 half a dozen bug fixes in pretty-printing and the debugger, and
3427 half a dozen others elsewhere
3428 * fixed bug 13: Floating point infinities are now supported again.
3429 They might still be a little bit flaky, but thanks to bug reports
3430 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3431 as flaky as they were.
3432 * The --noprogrammer command line option is now supported. (Its
3433 behavior is slightly different in detail from what the old man
3434 page claimed it would do, but it's still appropriate under the
3435 same circumstances that the man page talks about.)
3436 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3437 are now supported, and enabled by default. Thus, the compiler can
3438 handle many floating point and complex operations much less
3439 inefficiently. (Thus e.g. you can implement a complex FFT
3441 * The compiler now detects type mismatches between DECLAIM FTYPE
3442 and DEFUN better, and implements CHECK-TYPE more correctly, and
3443 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3444 thanks to patches from Martin Atzmueller.
3445 * various fixes to make the cross-compiler more portable to
3446 ANSI-conforming-but-different cross-compilation hosts (notably
3447 Lispworks for Windows, following bug reports from Arthur Lemmens)
3448 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3449 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3450 fixed thanks to Pierre Mai's CMU CL patches.
3451 * fixes to keep the system from overflowing internal counters when
3452 it tries to use i/o buffers larger than 16M bytes
3453 * fixed bug 45a: Various internal functions required to support
3454 complex special functions have been merged from CMU CL sources.
3455 (When I was first setting up SBCL, I misunderstood a compile-time
3456 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3457 * improved support for type intersection and union, fixing bug 12
3458 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3459 more obscure bugs as well
3460 * some steps toward byte-compiling non-performance-critical
3461 parts of the system, courtesy of patches from Martin Atzmueller
3462 * Christophe Rhodes has made some debian packages of sbcl at
3463 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3464 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3465 stable, but are nonetheless usable. When he's ready, I'd be happy
3466 to add them to the SourceForge "File Releases" section. (And if
3467 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3468 * new fasl file format version number (because of changes in
3469 internal representation of (OR ..) types to accommodate the new
3470 support for (AND ..) types, among other things)
3472 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3473 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3475 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3477 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3478 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3479 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3480 * DESCRIBE now works on structure objects again.
3481 * Most function call argument type mismatches are now handled as
3482 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3483 know whether the function will be redefined before the call is
3484 executed. (The compiler could flag local calls with full WARNINGs,
3485 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3486 it doesn't keep track of enough information to know whether calls
3487 are local in this sense.)
3488 * Compiler output is now more verbose, with messages truncated
3489 later than before. (There should be some supported way for users
3490 to override the default verbosity, but I haven't decided how to
3491 provide it yet, so this behavior is still controlled by the internal
3492 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3493 src/compiler/ir1util.lisp.)
3494 * Fasl file format version numbers have increased again, because
3495 support for the Gray streams extension changes the layout of the
3496 system's STREAM objects.
3497 * The Gray subclassable streams extension now works, thanks to a
3498 patch from Martin Atzmueller.
3499 * The full LOAD-FOREIGN extension (not just the primitive
3500 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3501 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3502 but like most other programs, it defaults to copying the Unix
3503 environment from the original process instead of starting the
3504 new process in an empty environment.
3505 * Extensions which manipulate the Unix environment now support
3506 an :ENVIRONMENT keyword option which doesn't smash case or
3507 do other bad things. The CMU-CL-style :ENV option is retained
3508 for porting convenience.
3509 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3510 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3512 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3514 * A patch from Martin Atzmueller seems to have solved the SIGINT
3515 problem, and as far as we know, signal-handling now works cleanly.
3516 (If you find any new bugs, please report them!)
3517 * The system no longer defaults Lisp source file names to types
3518 ".l", ".cl", or ".lsp", but only to ".lisp".
3519 * The compiler no longer uses special default file extensions for
3520 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3521 seems to expect a single default extension for all compiled code,
3522 and there's no compelling reason to try to stretch the standard
3523 to allow two different extensions.) Instead, byte-compiled files
3524 default to the same extension as native-compiled files.
3525 * Fasl file format version numbers have increased again, because
3526 a rearrangement of internal implementation packages made some
3527 dumped symbols in old fasl files unreadable in new cores.
3528 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3529 many fewer weird special cases.
3530 * Bug #17 (differing COMPILE-FILE behavior between logical and
3531 physical pathnames) has been fixed, and some related misbehavior too,
3532 thanks to a patch from Martin Atzmueller.
3533 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3534 by Tim Moore, ported to SBCL by Martin Atzmueller.
3535 * Martin Atzmueller fixed several filesystem-related problems,
3536 including bug #36, in part by porting CMU CL patches, which were
3537 written in part by Paul Werkowski.
3538 * More compiler warnings in src/runtime/ are gone, thanks to
3539 more patches from Martin Atzmueller.
3540 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3543 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3545 * DESCRIBE now works on CONDITION objects.
3546 * The debugger now handles errors which arise when trying to print
3547 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3549 * The build system now uses an additional file, customize-target-features.lisp,
3550 to allow local modifications to the target *FEATURES* list. (The point of
3551 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3552 debugging features enabled, without having to worry about propagating it
3553 into everyone's system when I do a "cvs update".) When no
3554 customize-target-features.lisp file exists, the target *FEATURES* list
3555 should be constructed the same way as before.
3556 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3557 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3558 warn about behavior which is undefined under the ANSI spec. Note: This
3559 is specified by ANSI, but it's not very popular with programmers.
3560 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3561 macro in the SBCL sources for an example of a workaround which you
3562 might use to make portable ANSI-standard code which does what you want.
3563 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3564 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3565 when it's used as a non-toplevel form. (This is required in order
3566 to implement the DEFCONSTANT-EQX macro.)
3567 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3568 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3569 funny things, due to bugs in the implementation of EVAL-WHEN
3570 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3571 non-toplevel DEFCONSTANTs are uncommon.)
3572 * The core file version number and fasl file version number have been
3573 incremented, because the old noncompliant DEFCONSTANT behavior involved
3574 calling functions which no longer exist, and because I also took the
3575 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3576 * fixed bug 1 (error handling before read-eval-print loop starts), and
3577 redid debugger restarts and related debugger commands somewhat while
3579 ** The QUIT debugger command is gone, since it did something
3580 rather different than the SB-EXT:QUIT command, and since it never
3581 worked properly outside the main toplevel read/eval/print loop.
3582 Invoking the new TOPLEVEL restart provides the same functionality.
3583 ** The GO debugger command is also gone, since you can just invoke
3584 the CONTINUE restart directly instead.
3585 ** The TOP debugger command is also gone, since it's redundant with the
3586 FRAME 0 command, and since it interfered with abbreviations for the
3588 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3590 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3591 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3592 some porting work by Martin Atzmueller)
3593 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3594 with a boatload of patches which he ported to SBCL. Now that those
3596 ** The system tries to make sure that its low-priority messages
3597 are prefixed by semicolons, to help people who like to use
3598 syntax highlighting in their ILISP buffer. (This patch
3599 was originally due to Raymond Toy.)
3600 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3601 patches originally written by Raymond Toy.
3602 ** The compiler understands coercion between single-value and
3603 multiple-VALUES type expressions better, getting rid of some very
3604 weird behavior, thanks to patches originally by Robert MacLachlan
3605 and Douglas Crosher.
3606 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3607 lambda lists, thanks to a patch originally by Pierre Mai.
3608 ** The system no longer bogusly warns about "abbreviated type
3610 ** The compiler gets less confused by inlining and RETURN-FROM,
3611 thanks to some patches originally by Tim Moore.
3612 ** The system no longer hangs when dumping circular lists to fasl
3613 files, thanks to a patch originally from Douglas Crosher.
3614 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3615 undefined function error.
3616 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3617 in structure slot initforms) for now, documented workaround instead:-|
3618 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3619 * fixed bug 5: added stubs for various Gray stream functions called
3620 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3621 installed, at least appropriate type errors are generated
3622 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3623 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3624 consistently in DEFMETHOD forms.
3625 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3626 it doesn't seem to affect SBCL after all
3627 * The C runtime system now builds with better optimization and many
3628 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3630 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3632 * The system is now under CVS at SourceForge (instead of the
3633 CVS repository on my home machine).
3634 * The new signal handling code has been tweaked to treat register
3635 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3636 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3637 downstream, e.g. in debug-int.lisp, has implicit dependencies
3638 on the unsignedness of integer representation of machine words,
3639 and that caused the system to bomb out with infinite regress
3640 when trying to recover from type errors involving signed values,
3641 e.g. (BUTLAST '(1 2 3) -1).)
3642 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3643 (This was one of the bugs Peter Van Eynde reported back in July.)
3644 * The system now uses code inspired by Colin Walters' O(N)
3645 implementation of MAP (from the cmucl-imp@cons.org mailing
3646 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3647 inline the MAP operation, and there is more than one
3648 sequence argument to the MAP call (so that it can't just
3649 do ETYPECASE once and for all based on the type of the
3650 single sequence argument). (The old non-inline implementation
3651 of the general M-argument sequence-of-length-N case required
3652 O(M*N*N) time when any of the sequence arguments were LISTs.)
3653 * The QUIT :UNIX-CODE keyword argument has been renamed to
3654 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3655 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3657 * Raymond Wiker's patches to port dynamic loading from Linux to
3658 FreeBSD have been added.
3659 * The BUGS file is now more nearly up to date, thanks in large part
3660 to Martin Atzmueller's review of it.
3661 * The debugger now flushes standard output streams before it begins
3662 its output ("debugger invoked" and so forth).
3663 * The core version number and fasl file version number have both
3664 been incremented, because of incompatible changes in the layout
3666 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3667 used to not be called for a saved Lisp image.)
3668 * Martin Atzmueller's version of a patch to fix a compiler crash,
3669 as posted on sbcl-devel 13 September 2000, has been installed.
3670 * Instead of installing Martin Atzmueller's patch for the
3671 compiler transform for SUBSEQ, I deleted the compiler transform,
3672 and transforms for some similar consing operations.
3673 * A bug in signal handling which kept TRACE from working on OpenBSD
3675 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3676 away by constant folding
3677 * The system now defines its address space constants in one place
3678 (in the Lisp sources), and propagates them automatically elsewhere
3679 (through GENESIS and the sbcl.h file). Therefore, patching the
3680 address map is less unnecessarily tedious and error-prone. The
3681 Lisp names of address space constants have also been systematized.
3682 * CVS tags like dollar-Header-dollar have been removed from
3683 the sources, because they have never saved me trouble and
3684 they've been source of trouble working with patches and other
3685 diff-related operations.
3686 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3687 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3689 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3691 * The system has been ported to OpenBSD.
3692 * The system now compiles with a simple "sh make.sh" on the systems
3693 that it's supported on. I.e., now you no longer need to tweak
3694 text in the target-features.lisp-expr and symlinks in src/runtime/
3695 by hand, the make.sh takes care of it for you.
3696 * The system is no longer so grossly inefficient when compiling code
3697 involving vectors implemented as general (not simple) vectors (VECTOR T),
3698 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3699 which dares to use the various sequence functions on non-simple
3700 vectors, takes less of a performance hit.
3701 * There is now a primitive type predicate VECTOR-T-P
3702 to test for the (VECTOR T) type, so that e.g.
3703 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3704 can now be compiled with some semblance of efficiency. (The old code
3705 turned the type declaration into a full call to %TYPEP at runtime!)
3706 * AREF on (VECTOR T) is still not fast, since it's still compiled
3707 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3708 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3709 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3710 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3711 without making full calls to SUBTYPEP at runtime.
3712 (Some analogous efficiency issues for non-simple vectors specialized to
3713 element types other than T, or for non-simple multidimensional arrays,
3714 have not been addressed. They could almost certainly be handled the
3715 same way if anyone is motivated to do so.)
3716 * The changes in array handling break binary compatibility, so
3717 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3718 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3719 instead of (VALUES T T).
3720 * By following the instructions that Dan Barlow posted to sbcl-devel
3721 on 2 July 2000, I was able to enable primitive dynamic object
3722 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3723 functionality is not implemented (since it calls ld to resolve
3724 library references automatically, requiring RUN-PROGRAM for its
3725 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3726 try to resolve library references) is now supported.
3727 * The system now flushes the standard output streams when it terminates,
3728 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3729 them at several other probably-convenient times, e.g. in each pass of
3730 the toplevel read-eval-print loop, and after evaluating a form given
3731 as an "--eval" command-line option. (These changes were motivated by a
3732 discussion of stream flushing issues on cmucl-imp in August 2000.)
3733 * The source transform for TYPEP of array types no longer assumes
3734 that an array whose element type is a not-yet-defined type
3735 is implemented as an array of T, but instead punts, so that the
3736 type will be interpreted at runtime.
3737 * There is now some support for cross-compiling in make.sh: each of
3738 the phases of make.sh has its own script. (This should be transparent
3739 to people doing ordinary, non-cross-compile builds.)
3740 * Since my laptop doesn't have hundreds of megabytes of memory like
3741 my desktop machine, I became more motivated to do some items on
3742 my to-do list in order to reduce the size of the system a little:
3743 ** Arrange for various needed-only-at-cold-init things to be
3744 uninterned after cold init. To support this, those things have
3745 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3746 symbols with such names are now uninterned after cold init).
3747 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3748 fasl files for cold load.
3749 ** Remove the old compiler structure pooling code (which used to
3750 be conditional on the target feature :SB-ALLOC) completely.
3751 ** Redo the representation of some data in cold init to be more compact.
3752 (I also looked into supporting byte compiled code at bootstrap time,
3753 which would probably reduce the size of the system a lot, but that
3754 looked too complicated, so I punted for now.)
3755 * The maximum signal nesting depth in the src/runtime/ support code has
3756 been reduced from 4096 to 256. (I don't know any reason for the very
3757 large old value. If the new smaller value turns out to break something,
3758 I'll probably just bump it back up.)
3759 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3761 * Many, many bugs reported by Peter Van Eynde have been added to
3762 the BUGS list; some have even been fixed.
3763 * While enabling dynamic object file loading, I tried to make the
3764 code easier to understand, renaming various functions and variables
3765 with less ambiguous names, and changing some function calling
3766 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3767 * While trying to figure out how to do the OpenBSD port, I tried to
3768 clean up some of the code in src/runtime/. In particular, I dropped
3769 support for non-POSIX signal handling, added various comments,
3770 tweaked the code to reduce the number of compilation warnings, and
3771 renamed some files to increase consistency.
3772 * To support the new automatic configuration functionality in make.sh,
3773 the source file target-features.lisp-expr has been replaced with the
3774 source file base-target-features.lisp-expr and the machine-generated
3775 file local-target-features.lisp-expr.
3776 * fixed a stupid quoting error in make.sh so that using CMU CL
3777 "lisp -batch" as cross-compilation host works again
3779 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3781 * DESCRIBE no longer tries to call itself recursively to describe
3782 bound/fbound values, so that it no longer fails on symbols which are
3783 bound to themselves (like keywords, T, and NIL).
3784 * DESCRIBE now works on generic functions.
3785 * The printer now prints less-screwed-up representations of closures
3786 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3787 * A private symbol is used instead of the :EMPTY keyword previously
3788 used to mark empty slots in hash tables. Thus
3789 (DEFVAR *HT* (MAKE-HASH-TABLE))
3790 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3791 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3792 now does what ANSI says that it should. (You can still get
3793 similar noncompliant behavior if bang on the hash table
3794 implementation with all the symbols you get back from
3795 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3796 This breaks binary compatibility, since tests for equality to
3797 :EMPTY are wired into things like the macroexpansion of
3798 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3800 * There's now a minimal placeholder implementation for CL:STEP,
3801 as required by ANSI.
3802 * An obscure bug in the interaction of the normal compiler, the byte
3803 compiler, inlining, and structure predicates has been patched
3804 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3805 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3807 * Missing ordinary arguments in a macro call are now detected even
3808 when the macro lambda list contains &KEY or &REST.
3809 * The debugger no longer complains about encountering the top of the
3810 stack when you type "FRAME 0" to explicitly instruct it to go to
3811 the top of the stack. And it now prints the frame you request even
3812 if it's the current frame (instead of saying "You are here.").
3813 * As specified by ANSI, the system now always prints keywords
3814 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3816 * The default initial SIZE of HASH-TABLEs is now smaller.
3817 * Type information from CLOS class dispatch is now propagated
3818 into DEFMETHOD bodies, so that e.g.
3819 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3821 is now basically equivalent to
3822 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3823 (DECLARE (TYPE SINGLE-FLOAT X))
3825 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3826 operation, without having to do run-time type dispatch.
3827 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3828 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3829 * Since I don't seem to be making big file reorganizations very often
3830 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3831 is overflowing my ability to conveniently back them up), I've finally
3832 checked the system into CVS. (The CVS repository is on my home system,
3833 not at SourceForge -- putting it on SourceForge might come later.)
3834 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3835 high-level GC-NOTIFY-FOO functions send their output. (There's
3836 still very little control of where low-level verbose GC functions
3837 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3838 less than it used to -- the GC-NOTIFY-FOO functions are now under
3839 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3840 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3841 in only one place in the source code, and propagates it automatically
3842 everywhere that it's needed. Thus e.g. when I bump the version from
3843 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3844 * The C source files now include boilerplate legalese and documentation
3845 at the head of each file (just as the Lisp source files already did).
3846 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3847 to his page will be replaced with a link to his new CLiki service.
3849 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3851 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3852 * The build process now looks for GNU make under the default name "gmake",
3853 instead of "make" as it used to. If GNU make is not available as "gmake"
3854 on your system, you can change this default behavior by setting the
3855 GNUMAKE environment variable.
3856 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3857 can build without error under CMU CL.
3859 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3861 * There is now a partial SBCL user manual (with some new text and some
3862 text cribbed from the CMU CL manual).
3863 * The beginnings of a profiler have been added (starting with the
3864 CMU CL profiler and simplifying and cleaning up). Eventually the
3865 main interface should be through the TRACE macro, but for now,
3866 it's still accessed through vaguely CMU-CL-style functions and macros
3867 exported from the package SB-PROFILE.
3868 * Some problems left over from porting CMU CL to the new
3869 cross-compilation bootstrap process have been cleaned up:
3870 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3871 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3872 part when getting rid of the compiler *BACKEND* stuff.)
3873 ** The value of *NULL-TYPE* was screwed up, because it was
3874 being initialized before the type system knew the final
3875 definition of the 'NULL type. This screwed up several key
3876 optimizations in the compiler, causing inefficiency in all sorts
3877 of places. (I found it because I wanted to understand why
3878 GET-INTERNAL-RUN-TIME was consing.)
3879 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3880 PROCLAIM FTYPE information. Unfortunately this broke binary
3881 compatibility again, since now the forms output by DEFGENERIC
3882 to refer to functions which didn't exist in 0.6.3.
3883 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3884 can use the new (as of 0.6.3) transform for SOME into MAP into
3886 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3887 values to (INTEGER 0 1000000), so that the time code will no longer
3888 occasionally get blown up by Linux returning 1000000 microseconds
3889 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3890 its output conform to the ANSI spec. (Alas, this makes its output
3891 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3892 * A full call to MAP NIL with a single sequence argument no longer conses.
3893 * fixes to problems pointed out by Martin Atzmueller:
3894 * The manual page no longer talks about multiprocessing as though
3895 it were currently supported.
3896 * The ILISP support patches have been removed from the distribution,
3897 because as of version 5.10.1, ILISP now supports SBCL without us
3898 having to maintain patches.
3899 * added a modified version of Raymond Toy's recent CMU CL patch for
3900 EQUALP comparison of HASH-TABLE
3902 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3904 * The system still can't cross-compile itself with
3905 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3906 entails), but at least it can compile more of itself that way
3907 than it used to be able to, and various buglets which were uncovered
3908 by trying to cross-compile itself that way have now been fixed.
3909 * This release breaks binary compatibility again. This time
3910 at least I've incremented the FASL file format version to 2, so that the
3911 problem can be detected reliably instead of just causing weird errors.
3912 * various new style warnings:
3913 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3914 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3915 ** using the lexical binding of a variable named in the *FOO* style
3916 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3917 as specified by ANSI.
3918 * *RANDOM-STATE* is no longer automatically initialized from
3919 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3920 default behavior of the system is to repeat its behavior every time
3921 it's run. If you'd like to change this behavior, you can always
3922 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3923 old convention there was no comparably easy way to get the system to
3924 repeat its behavior every time it was run.
3925 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3926 FTYPE declarations has been removed, in favor of their ANSI
3927 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3928 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3929 the types of their sequence arguments can be determined at compile time.
3930 This is done through a new open code expansion for MAP which eliminates
3931 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3932 argument types can be determined at compile time.
3933 * The optimizer now transforms COERCE into an identity operation when it
3934 can prove that the coerced object is already of the correct type. (This
3935 can be a win for machine generated code, including the output of other
3936 optimization transforms, such as the MAP transform above.)
3937 * Credit information has been moved from source file headers into CREDITS.
3938 * Source file headers have been made more standard.
3939 * The CASE macro now compiles without complaining even when it has
3942 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3944 * (Note that the way that the PCL macroexpansions were rewritten
3945 to accommodate the change in DEFGENERIC below breaks binary
3946 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3947 not run under sbcl-0.6.2. Once we get out of alpha releases,
3948 i.e. hit release 1.0.0, we'll probably try to maintain binary
3949 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3950 and sbcl-1.4.4. Until then, however, it might be fairly common
3951 for maintenance releases to break binary compatibility.)
3952 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3954 * The reporting of unhandled signals has been changed to print some
3955 explanatory text as well as the report form. (Previously only
3956 the report form was printed.)
3957 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3958 it defines, so that the compiler no longer issues undefined function
3959 warnings for compiled-but-not-yet-loaded generic functions.
3960 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3961 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3962 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3963 section of the ANSI spec are supported.
3964 * The "" nickname for the "KEYWORD" package has been removed.
3965 The reader still handles symbol tokens which begin with a package marker
3966 as keywords, but it doesn't expose its mechanism for doing so in the
3967 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3968 * The system now issues STYLE-WARNINGs for contradictory TYPE
3969 proclamations. (Warnings for contradictory FTYPE proclamations would
3970 be nice too, but those can't be done usefully unless the type system
3971 is made smarter about FUNCTION types.)
3972 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3973 systematized, so that "*target-*.lisp is supposed to exist only on the
3974 target and imply that there's a related file which exists on the
3975 host, and *host-*.lisp is supposed to exist only on the host and imply
3976 that there's a related file which exists on the target. This involves a
3977 lot of renaming. Hopefully the acute confusion caused by the renaming
3978 will be justified by the reduction in chronic confusion..
3979 ** runtime-type.lisp -> early-target-type.lisp
3980 ** target-type.lisp -> late-target-type.lisp
3981 ** early-host-format.lisp -> early-format.lisp
3982 ** late-host-format.lisp -> late-format.lisp
3983 ** host-error.lisp -> misc-error.lisp
3984 ** early-error.lisp -> early-target-error.lisp
3985 ** late-error.lisp -> late-target-error.lisp
3986 ** host-defboot.lisp -> early-defboot.lisp
3987 ** code/misc.lisp -> code/target-misc.lisp
3988 ** code/host-misc.lisp -> code/misc.lisp
3989 ** code/numbers.lisp -> code/target-numbers.lisp
3990 ** code/early-numbers.lisp -> numbers.lisp
3991 ** early-host-type.lisp -> early-type.lisp
3992 ** late-host-type.lisp -> late-type.lisp
3993 ** host-typep.lisp -> typep.lisp
3994 ** load.lisp -> target-load.lisp
3995 ** host-load.lisp -> load.lisp
3996 ** host-disassem.lisp -> disassem.lisp
3997 ** host-insts.lisp -> insts.lisp
3998 ** byte-comp.lisp -> target-byte-comp.lisp
3999 ** host-byte-comp.lisp -> byte-comp.lisp
4000 ** host-signal.lisp -> signal.lisp
4001 ** host-defstruct.lisp -> defstruct.lisp
4002 ** late-target-type.lisp -> deftypes-for-target.lisp
4003 Furthermore, several other previously target-only files foo.lisp (e.g.
4004 hash-table.lisp and random.lisp) have been split into a target-and-host
4005 foo.lisp file and a target-only target-foo.lisp file, with their key type
4006 definitions in the target-and-host part, so that the cross-compiler will
4007 know more about target types.
4008 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4009 gone away. In their place are various *BACKEND-FOO* variables
4010 corresponding to the slots of the old structure.
4011 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4012 into the target SBCL has been fixed.
4013 * The chill.lisp system for loading cold code into a running SBCL
4015 * Support for the CMU CL "scavenger hook" extension has been removed.
4016 (It was undocumented and unused in the CMU CL sources that SBCL was
4017 derived from, and stale in sbcl-0.6.1.)
4018 * Various errors in the cross-compiler type system were detected
4019 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4020 (enabling various consistency checks). Many of them were fixed,
4021 but some hard problems remain, so the compiler is back to
4022 running without *TYPE-SYSTEM-INITIALIZED* for now.
4023 * As part of the cross-compiler type system cleanup, I implemented
4024 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4025 * I have started adding UNCROSS calls throughout the type system
4026 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4027 on cross-compiler input files will be able to go away, and various
4029 * CONSTANTP now returns true for quoted forms (as explicitly required
4032 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4034 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4035 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4036 (MAKE-LIST -1). In the longer term, it ought to become true
4037 that declarations are assertions even at SAFETY 1. For now, it's not
4038 quite true even at SAFETY 3, but it's at least more nearly true..
4039 (Note that this change seems to increases the size of the system by
4040 O(5%) and to decrease the speed of the compiler by 20% or more.)
4041 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4042 for the problem of printing dozens of lines of distracting information
4043 about low-level system machinery as part of the top stack frame
4044 on entry to the debugger when an undefined function was called.
4045 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4046 is not reset to COMMON-LISP-USER.
4047 * Compilation of stuff related to dyncount.lisp has been made conditional
4048 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4049 smaller. The various dyncount-related symbols have been moved into
4050 a new "SB-DYNCOUNT" package.
4051 * tty-inspect.lisp has been renamed to inspect.lisp.
4052 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4053 feature has gone away. (When we eventually port to other flavors of
4054 libc and/or Unix, we'll try to make the differences between flavors
4055 invisible at the user level.)
4056 * Various other *FEATURES* tags, and/or their associated conditionals,
4057 have been removed if obsolescent, or given better documentation, or
4058 sometimes given more-mnemonic names.
4060 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4062 * tidied up "make.sh" script
4063 * tidied up system directory structure
4064 * better "clean.sh" behavior
4065 * added doc/FOR-CMUCL-DEVELOPERS
4066 * many many small tweaks to output format, e.g. removing possibly-confusing
4067 trailing #\. character in DESCRIBE-INSTANCE
4068 * (EQUALP #\A 'A) no longer signals an error.
4069 * new hashing code, including EQUALP hashing
4070 * tidied up Lisp initialization and toplevel
4071 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4072 * command line argument processing
4073 * added POSIX-GETENV function to deal with Unix-ish environment variables
4074 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4075 terminating SBCL on EOF
4076 * non-verbose GC by default
4077 * There is no more "sbcl" shell script; the sbcl file is now the C
4078 runtime executable (just like CMU CL).
4079 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4081 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4082 symbol and package information as Lisp native symbol and package objects
4083 instead of strings naming symbols and strings naming packages. This way,
4084 whenever packages are renamed (as in warm init), debug information is
4085 transformed along with everything else.
4086 * tweaked the optimization policy declarations which control the building
4087 of SBCL itself. Now, among other things, the system no longer saves
4088 source location debugging information. (This helps two problems at once
4089 by reducing SBCL size and by keeping SBCL from trying to look for its
4090 sources -- which may not exist -- when reporting errors.)
4091 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4092 debugging and testing purposes
4093 * cleaned up printing, making the printer call PRINT-OBJECT for
4094 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4095 methods, giving nearly-ANSI behavior
4096 * converted almost all special variables to use *FOO* naming convention
4097 * deleted PARSE-TIME functionality, since it can be done portably
4098 * moved some files out of cold init into warm init
4099 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4101 * regularized formatting of source files
4102 * added an install.sh script
4103 * fixed ridiculous memory usage of cross-compiler by making
4104 compiler/alloc.lisp not try to do pooling unless it can hook
4105 itself into the GC of the cross-compilation host. Now the system
4106 builds nicely on my old laptop.
4107 * added :SB-ALLOC in target-features.lisp-expr
4108 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4109 was not implemented)
4110 * re-did condition handling and note reporting in the compiler. Notes
4111 are no longer handled by signalling conditions. Style warnings
4112 and warnings are handled more correctly and reported in such a way
4113 that it's easy to find one or the other in your output (so that you
4114 can e.g. figure out which of many problems caused COMPILE-FILE to
4116 * changed the severity of several compiler warnings from full WARNING
4117 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4118 compiler note reporting so that it doesn't use the condition system
4119 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4120 COMPILE-FILE command)
4121 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4122 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4123 no longer supported, since their implementation was deeply intertwingled
4124 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4125 * removed bogus "support" for compiler macros named (SETF FOO), and
4126 removed the compiler macro for SETF INFO (but only after making a fool
4127 of myself on the cmucl-imp mailing list by posting a bogus patch for
4128 DEFINE-COMPILER-MACRO..)
4129 * Compiled files containing forms which have side effects on the Lisp
4130 reader (such as DEFPACKAGE forms) are now handled more correctly.
4131 (Compiler queuing of top level lambdas has been suppressed by setting
4132 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4133 * deleted various currently-unused source files, e.g. gengc.lisp. They
4134 may be added back at some point e.g. when porting to other architectures,
4135 but until they are it's distracting to distribute them and to try to
4137 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4138 there were so many of them they're just distractions, and UNCROSS is
4139 known to be able to handle the current sources
4140 * moved PROFILE functionality into TRACE, so that it will be clear
4141 how the wrapping and unwrapping of functions when you profile them
4142 interacts with the wrapping and unwrapping of functions when you
4143 trace them. (Actually, the functionality isn't there yet, but at least
4144 the interface specification is there. Hopefully, the functionality will
4145 arrive with some future maintenance release.)
4146 * removed host-oops.lisp
4147 * changed signature of QUIT function to allow UNIX-CODE argument
4148 * fixed READ-SEQUENCE bug
4149 * tweaked verbose GC output so that it looks more like the progress
4150 output that ANSI specifies for functions like LOAD
4151 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4152 * added <http://sbcl.sourceforge.com> to the banner information printed by