1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
3 * improvement: experimental support for mach exception handling on
4 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
6 * improvement: support for GBK external format.
7 (thanks to Chun Tian (binghe))
8 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
9 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
10 (thanks to Tony Martinez)
11 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
12 evaluated once (reported by Kevin Reid)
14 changes in sbcl-1.0.1 relative to sbcl-1.0:
15 * new platform: FreeBSD/x86-64, including support for threading.
16 * new feature: the compiler stores cross-referencing information
17 abount function calls (who-calls), macroexpansion (who-macroexpands)
18 and special variables (who-binds, who-sets, who-references) for code
19 compiled with (< SPACE 3). This information is available through the
20 sb-introspect contrib.
21 * new feature: users may subclass SEQUENCE, and have instances of
22 these classes interoperate with standard Common Lisp functions if
23 a number of methods are defined. (This feature is experimental
24 and the interface subject to change based on feedback from SBCL
25 users and the general community)
26 * improvement: sb-sprof traces call stacks to an arbitrary depth on
27 x86/x86-64, rather than the previous fixed depth of 8
28 * improvement: another pthread back-end of mutex "pthread-futex".
29 * bug fix: non-ascii command-line arguments are processed correctly
30 (thanks to Yaroslav Kavenchuk)
31 * bug fix: non-required arguments were not passed correctly when a method
32 defined using DEFMETHOD was called from a mop-generated method using
33 CALL-NEXT-METHOD (reported by Pascal Costanza)
34 * bug fix: recursion is now permitted in accessors through
35 SLOT-UNBOUND. (reported by Pascal Costanza)
36 * bug fix: an error was signaled at startup if the HOME environment
37 variable was defined, but had an empty value (reported by Peter Van Eynde)
38 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
39 signal a PROGRAM-ERROR, not a BUG.
40 * bug fix: SB-POSIX:READDIR works when built with large file support.
41 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
43 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
44 for better automatic indentation support. (thanks to Matt Pillsbury)
45 * optimization: loading generic functions no longer takes O(n^2) time,
46 proportional to the amount of methods in the generic function
47 (reported by Todd Sabin and Jeremy Brown)
48 * optimization: the FIND and POSITION family of sequence functions
49 are significantly faster on arrays whose element types have been
51 * improvements to the Windows port:
52 ** Intermittent heap corruption problems have been fixed. (thanks
53 to Alastair Bridgewater)
54 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
56 ** Lisp is able to unwind foreign exception frames from alien
57 callbacks. (thanks to Alastair Bridgewater)
59 changes in sbcl-1.0 relative to sbcl-0.9.18:
60 * improvement: experimental support for threading on FreeBSD/x86.
61 (thanks to NIIMI Satoshi)
62 * improvement: runtime option --dynamic-space-size can be used
63 to set the size of the dynamic space reserved on startup.
64 * improvement: floating point modes in effect are now saved in
65 core, and restored on startup.
66 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
67 startup, not time since first call to GET-INTERNAL-REAL-TIME.
68 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
69 threads are not running after *SAVE-HOOKS* have run.
70 * improvement: writes to CLOS instance slots are type-checked in code
71 compiled with (SAFETY 3)
72 * improvement: floating-point exception handling on FreeBSD (thanks to
74 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
75 (thanks to Zach Beane)
76 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
78 * improvement: added support for the Shift-JIS external format.
79 (contributed by NIIMI Satoshi)
80 * improvement: callbacks are supported on Linux/PPC. (thanks to
82 * bug fix: compiler bug triggered by a (non-standard) VALUES
83 declaration in a LET* was fixed. (reported by Kaersten Poeck)
84 * bug fix: file compiler no longer confuses validated and already
85 dumped structurres. (reported by Kaersten Poeck)
86 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
87 fill-pointer signals a type-error as required. (thanks to
89 * bug fix: disassemly of funcallable instances works.
90 * bug fix: single stepping on PPC.
91 * bug fix: fix thread-safety problems in the type system (generally
92 manifesting as nonsensical errors like "STRING is a bad type specifier
93 for sequences" or "The value 1 is not of type FIXNUM")
94 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
95 (reported by Josip Gracin)
96 * bug fix: an error is signaled for attempts to displace arrays with
97 incompatible element types (thanks to Mario Mommer)
98 * bug fix: more correct handling of wide characters in debug info
99 (bug reported by Attila Lendvai and fixed by Juho Snellman)
100 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
101 and don't cause extra consing
102 * optimization: MAP and MAP-INTO are significantly faster on vectors
103 whose elements types have been declared.
104 * Improvements to SB-SPROF:
105 ** Support for allocation profiling
106 ** Reduced profiling overhead, especially for long profiling runs
107 * Improvements to the Windows port:
108 ** floating point exceptions are now reported correctly.
109 ** stack exhaustion detection works partially.
110 ** more accurate GET-INTERNAL-REAL-TIME.
111 ** better RUN-PROGRAM behaviour in case of errors in the child
113 ** PROBE-FILE now simplifies pathnames correctly.
114 ** DIRECTORY now works correctly with :WILD-INFERIORS.
115 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
116 1st 1970 (although time timezone data will be unreliable).
117 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
119 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
120 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
121 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
123 * bug fix: two potential GC deadlocks affecting threaded builds.
124 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
125 returns T (reported by Anton Kazennikov)
126 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
127 with non-variable places
128 * bug fix: remove a race condition in the setting of
129 funcallable-instance functions, this should make threaded CLOS
130 code more stable against memory faults.
131 * bug fix: corruption of specials when unbinding is interrupted by an
132 asynchronous unwind (reported by Hannu Koivisto)
133 * improvement: the debugger will now also display local variables that
134 are only used once, for code compiled with a DEBUG optimization quality
137 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
138 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
139 * incompatible change: External-format support for FFI calls. The
140 SB-ALIEN:C-STRING no longer implies an ASCII
141 external-format. Instead the string is subject to external-format
142 conversion. Additionally return values of type SB-ALIEN:C-STRING
143 are by default of type (SIMPLE-ARRAY CHARACTER), not
144 SIMPLE-BASE-STRING. For an alien type that behaves like the old
145 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
146 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
147 * incompatible change: SB-EXT package no longer contains the
148 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
149 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
151 * incompatible change: the single-stepper is no longer available
152 on Alpha, Mips and Sparc platforms.
153 * minor incompatible change: the direct superclasses of
154 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
155 not (STANDARD-OBJECT FUNCTION). This makes the
156 class-precedence-lists of GENERIC-FUNCTION and
157 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
159 * minor incompatible change: the default stream external format on
160 non-unicode SBCL is no longer detected from the locale, but is always
162 * new feature: Add a version of evaluator that uses an interpreter instead
163 of the compiler. EVAL still uses the compiler by default, to switch it
164 to use the interpreter, set the value of the variable
165 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
166 * minor incompatible change: the single-stepper REPL has been merged
167 with the normal debugger (see the "Stepping" heading of the debugger help
168 for more details). The debugger command STEP will no longer switch
169 to the single-stepper REPL.
170 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
171 for a type now works.
172 * bug fix: loading of default sysinit file works. (thanks to Leonid
174 * bug fix: better detection of circularities in the file-compiler.
175 (reported by Marco Monteiro)
176 * bug fix: the CL pathname functions now work with files that have
177 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
178 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
179 systems with Unix98 pty semantics.
180 * bug fix: ASDF-INSTALL will now work with bsd tar.
181 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
183 * bug fix: timers expiring in dead threads no longer cause a
184 type-error (reported by Paul "Nonny Mouse").
185 * bug fix: thanks to more lightweight single-stepper instrumentation,
186 code compiled with (DEBUG 3) will compile and execute significantly faster,
187 and will have more accurate type-inferencing than before
188 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
189 whose bindings are modified
190 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
191 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
192 ** replace dummy implementations of CL:MACHINE-INSTANCE and
193 CL:SOFTWARE-VERSION with proper version
195 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
196 * feature: implemented the READER-METHOD-CLASS and
197 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
198 as specified by AMOP.
199 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
201 * optimization: faster LOGCOUNT implementation on x86 and x86-64
202 (thanks to Lutz Euler)
203 * optimization: hashing of general arrays and vectors has been
204 improved. (reported by Any Fingerhut)
205 * enhancement: SB-INTROSPECT is now able to find definitions of
206 profiled functions. (thanks to Troels Henriksen)
207 * enhancement: compiler-macro expansion applies now to FUNCALL forms
208 as well, allowing compiler-macros for SETF-functions to expand.
209 * enhancement: step-instrumentation no longer wraps known
210 single-value functions with multiple-value context, allowing
211 better type inference.
212 * fixed bug #337: use of MAKE-METHOD in method combination now works
213 even in the presence of user-defined method classes. (reported by
214 Bruno Haible and Pascal Costanza)
215 * fixed bug #339(c): if there are applicable methods not part of any
216 long-form method-combination group, call INVALID-METHOD-ERROR.
217 (reported by Bruno Haible)
218 * fixed bug #361: the :FUNCTION initarg in the protocol for
219 initialization of methods can now be used to override
220 internally-produced optimized functions. (reported by Bruno
222 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
223 system-provided lambda expression no longer cause warnings about
224 unbound #:|pv-table| symbols.
225 * bug fix: improved the handling of type declarations and the
226 detection of violations for keyword arguments with non-constant
228 * bug fix: potentially erronous calls to PATHNAME and
229 MERGE-PATHNAMES were being flushed in some cases.
230 (reported by Richard Kreuter)
231 * bug fix: compiled calls to TYPEP were mishandling obsolete
232 instances. (reported by James Bielman and Attila Lendvai)
233 * bug fix: format strings with ~> without matching ~< no longer
234 trigger an AVER, but signal an understandable error instead.
235 (reported by Antonio Martinez)
236 * bug fix: specifying an output-file without a pathname-type for
237 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
238 of input-file instead of "fasl". (reported by Robert Dodier)
239 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
240 (reported by James Y Knight).
241 * bug fix: compiler-macros-function did not consider the environment
242 argument for shadowing by local functions.
243 * bug fix: compiler-macros expansion was inhibited by local INLINE
245 * bug fix: inline expansions of known functions were subject to
246 step-instrumentation in high debug policies, leading to problems
248 * bug fix: compiler failed to differentiate between different CONS
250 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
251 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
252 * bug fix: #368: incorrect use of expressed vs. upgraded array
254 * bug fix: #306a: more precise unions of array types.
255 * thread-safety improvements:
256 ** CONDITION-WAIT could return early on Linux, if the thread was
257 interrupted and subsequently continued with SIGCONT.
258 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
260 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
261 * added support for the ucs-2 external format. (contributed by Ivan
263 * minor incompatible change: pretty printing of objects of type
264 (cons symbol) is, in the default pprint-dispatch-table, now
265 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
267 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
268 called later in a class's lifetime, possibly as late as when the
269 first instance of the class is created. Previously,
270 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
271 class became finalizeable.
272 * fixed bug: FILE-POSITION sometimes returned inconsistent results
273 for multibyte external-format streams. (thanks to "vbzoli")
274 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
275 with :ALLOCATION :CLASS inherited from superclasses of the
277 * fixed bug: anonymous classes can now be created using the :NAME
278 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
279 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
280 * fixed bug: core-files saved with :EXECUTABLE T can again be
281 executed when SBCL_HOME isn't set. (reported by James Knight)
282 * fixed bug: toplevel LOCALLY forms with declarations could
283 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
284 * fixed bug: printing from several different threads using different
285 values of *print-case* could cause invalid output, due to
286 some internal special variables of the printer not being bound
287 thread-locally (reported by Max Mikhanosha)
288 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
289 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
290 (reported by Pascal Costanza)
291 * fixed bug: FINALIZE-INHERITANCE is called from
292 REINITIALIZE-INSTANCE on classes when the class has previously
293 been finalized, as required by AMOP.
294 * minor code generation optimizations:
295 ** better register allocation in CLOS dispatching functions
296 ** overflow detection when coercing signed bytes to fixnums on x86-64
297 ** is now implemented with one IMUL instruction instead of three shifts
298 ** more efficient bit-vector access on x86 and x86-64
299 ** more efficient access to raw structure slots on x86 and x86-64
300 * fixed some bugs revealed by Paul Dietz' test suite:
301 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
304 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
305 * feature: thread support on Solaris/x86, and experimental thread support
307 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
309 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
310 default method for SLOT-UNBOUND.
311 * minor incompatible change: prevent the user from specializing the
312 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
313 somewhat counter to the intent of the protocol, I (CSR) think, and
314 additionally it just doesn't work in SBCL as currently
315 implemented, thanks to optimizations (that are always valid for
316 the other three SLOT-VALUEish functions, but not for the setter).
317 * minor incompatible change: the :SB-LDB feature is now enabled by
318 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
319 the low-level debugger.
320 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
321 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
322 down to an SB-SYS:FD-STREAM, allowing direct communication in
323 more cases. Temporary files and pipes are still used as a fallback
325 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
326 on the non-reentrant C functions localtime(3) and gmtime(3).
327 * bug fix: global optimization policy was not visible in LOCALLY and
329 * bug fix: class objects can be used as specializers in methods.
330 (reported by Pascal Costanza)
331 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
332 with a type error. (reported by blitz_ on #lisp)
333 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
334 failed with a type error. (reported by Pascal Bourguignon)
335 * bug fix: merging pathnames against defaults with :DIRECTORY
336 starting with '(:RELATIVE :BACK) should preserve the :BACK.
337 (reported by James Y Knight)
338 * bug fix: saving large (>2GB) cores on x86-64 now works
339 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
340 constant index and a (simple-array (signed-byte 32)) array
341 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
342 error (patch by Robert J. Macomber)
343 * feature: TIME now displays the amount of run-time spent in GC
344 * bug fix: The debugger now does a better job of respecting
345 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
346 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
347 workaround for bug 403.)
348 * bug fix: floating point exception handling now works on FreeBSD
349 (thanks to NIIMI Satoshi)
350 * fixed some bugs revealed by Paul Dietz' test suite:
351 ** MISC.641: LET-conversion were not supposed to work in late
353 * improvements to the Win32/x86 port:
354 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
355 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
357 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
358 * new feature: source path information is generated for macro-expansion
359 errors for use in IDE's like Slime (thanks to Helmut Eller)
360 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
361 * bug fix: compiler does not loop forever on an invalid type in
363 * improvement: compilation of most CLOS applications is significantly
365 * optimization: added a limited bytecode compiler for simple toplevel
366 forms, speeding up compilation and FASL loading
367 * bug fix: the statistical profiler now properly distinguishes anonymous
370 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
371 * minor incompatible change: in sbcl-0.9.11 (but not earlier
372 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
373 into subclasses of the finalized class. Now user calls to
374 FINALIZE-INHERITANCE finalize just the one class, and calls by the
375 system return before any subclasses are finalized.
376 * minor incompatible change: The reader no longer ignores errors
377 regarding non-existent packages in #+ and #- feature tests.
378 * new feature: command line options --no-sysinit, --no-userinit to
379 inhibit loading the corresponding init files
380 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
381 for sending data through UDP sockets (thanks to François-René Rideau)
382 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
383 error is signalled instead (thanks to François-René Rideau)
384 * minor incompatible change: Error signalling behaviour of lexical
385 operations violating package locks has changed slightly. Refer to
386 documentation on package locks for details.
387 * enhancement: EVAL can process IF-expressions without resorting to the
389 * bug fix: Some operations involving SIN, COS, or TAN weren't being
390 constant-folded properly due to missing out-of-line functions.
391 (reported by Mika Pihlajamäki)
392 * bug fix: LISTEN sometimes returned T even in cases where no data was
393 immediately available from the stream
394 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
395 were reversed. (reported by Levente Mészáros)
396 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
397 when given a vector displaced to an adjustable array. (reported
399 * bug fix: derivation of float boundaries from numbers outside the
400 appropriate float range (reported by John Wiseman)
401 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
403 * bug fix: precomputing cache entries for generic functions with
404 some subclasses of specializers as yet invalid does not attempt to
405 fill a cache line with a negative offset. (reported by Levente
407 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
408 allow condition class objects as arguments to DOCUMENTATION and
409 (SETF DOCUMENTATION); only find and set documentation for
410 structure names for the STRUCTURE doc-type. (suggested by Gary
412 * improvements to the Win32/x86 port:
413 ** (user-homedir-pathname) and default initialization file
414 locations now know about the user's "Documents and Settings"
415 directory (thanks to Yaroslav Kavenchuk)
416 ** run-program is implemented (thanks to Mike Thomas)
417 ** sockets support (thanks to Timothy Ritchey)
418 ** better backtrace support (thanks to Alastair Bridgewater)
419 ** sb-grovel supported
420 ** asdf-install and sb-posix work somewhat
421 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
422 * improvements to the Solaris/x86 port:
423 ** works on Solaris 11/Solaris Express
424 ** floating-point exception handling support
425 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** REMOVE-METHOD returns its generic function argument even when
428 the method is not one of the generic functions' methods.
429 ** objects declared as MEMBER types can be admissible arguments to
431 ** printing characters should simply be printed by the FORMAT ~:C
433 ** compiler failure when compiling functions with hairy constant
434 defaults for optional parameters.
435 ** compiler produces wrong code when MAYBE-INLINE-expanding a
436 function, which is already optimized.
438 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
439 * new platform: experimental support for SBCL x86/Darwin, including
440 MacOS X 10.4.5 on Intel.
441 * new feature: Unicode character names are now known to the system
442 (through CHAR-NAME and NAME-CHAR).
443 * new feature: the filesystem location of SBCL's core file is
444 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
445 * minor incompatible change: the contrib modules SB-POSIX and
446 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
447 this change is to make it easier to distribute
448 location-independent binaries.
449 * bug fix: as implied by AMOP, standardized classes no longer have
450 slots named by external symbols of public packages. (reported by
452 * optimization: calling functions via constant symbols -- as in
453 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
454 via the function object as in (FUNCALL #'FOO).
455 * optimization: on x86 and x86-64, the calling convention for the
456 case of dealing with an unknown number of values has been altered
457 to be friendlier to the prediction heuristics implemented,
458 particularly with reference to CALL and RET pairing. (thanks to
459 Alastair Bridgewater)
460 * optimization: on x86, the code for access to thread-local symbol
461 values and binding/unbinding of thread-local symbols is smaller.
462 (thanks to Alastair Bridgewater)
463 * enhancement: CONSTANTP is now able to determine constantness of
464 more complex forms, including calls to constant-foldable standardized
465 functions and some special forms beyond QUOTE.
466 * fixed bug: occasional GC crashes on Solaris/x86
467 * optimization: x86-64 supports stack allocation of results of simple
468 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
469 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
470 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
471 (thanks to James Knight)
472 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
473 by the standard (thanks to Stephen Compall)
475 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
476 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
477 be used for bundling the runtime and the core file into one
478 executable binary. This feature is not currently supported on all SBCL
479 platforms. (thanks to James Bielman and NIIMI Satoshi)
480 * new feature: a generational or ephemeral garbage collector is now
481 the default on the PowerPC platform (both Linux and Darwin). The
482 old Cheney (stop and copy) collector is a build-time option.
483 (thanks to Cyrus Harmon, after Raymond Toy)
484 * minor incompatible change: the method by which SBCL finds its
485 contributed modules has changed; it no longer relies on symbolic
486 links from an $SBCL_HOME/systems directory, but searches directly
487 in the subdirectories of $SBCL_HOME.
488 * enhancement: the dynamic heap size on the Linux/PPC platform is
489 markedly larger, even using the older Cheney garbage collector.
490 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
491 certain complicated string types. (reported by Gary King)
492 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
493 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
494 * fixed bug: DOCUMENTATION on structures no longer signals an
495 error if no structure type of the right name exists. (reported by
497 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
498 applicable non-standard primary or :AROUND method on
499 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
500 no longer cause unbound variable errors. (reported by Kevin Reid)
501 * fixed bug: metacircle resolution in cases where methods have slots
502 added before the slots from STANDARD-METHOD. (reported by Jean
504 * fixed bug: the Power PC assembly code for calling into lisp
505 sometimes computed its return address wrongly (depending on the
506 15th bit of the address, fixed at link time). This used to
507 manifest itself in a segmentation violation while building PCL.
508 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
509 many others over the years)
510 * fixed bug: ROOM no longer reports silly numbers for stack usage if
511 the stack pointer (treated as a signed integer) is negative.
512 (thanks to Peter van Eynde)
514 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
515 * new platform: experimental support for the Windows operating
516 system has been added. (thanks to Alastair Bridgewater)
517 * fixed several bugs in and robustified the PPC FFI (including
518 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
519 * bug fix: allow non-simple string symbol names (reported by Paul
521 * bug fix: interrupt handling on NetBSD (thanks to Richard M
523 * bug fix: saving a core corrupted callbacks on x86/x86-64
524 * bug fix: closed a loophole in metacircularity detection and
525 grounding in the PCL implementation of CLOS.
526 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
527 (reported by tomppa on #lisp)
528 * bug fix: clear the direction flag on Lisp -> C transitions, as
529 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
530 (reported by Andrej Grozin and Hendrik Maryns)
531 * optimization: major improvements to GC efficiency on GENCGC platforms
532 * optimization: faster implementation of EQUAL
533 * optimization: emit more efficient opcodes for some common
534 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
535 * optimization: several other minor code-generation improvements on x86-64
536 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
538 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
539 * minor incompatible change: (SETF CLASS-NAME) and (SETF
540 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
541 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
542 by ANSI as a generic function, and for consistency (SETF
543 GENERIC-FUNCTION-NAME) is treated likewise.
544 * fixed bug #233.b: make constraint propagation notice when a variable
545 value is changed after it is referenced but before it is used
546 * fixed bug #296: no more arbitrary behaviour from filenames with
547 odd characters as --load arguments. (reported by Adam Warner)
548 * fixed bug #390: :CHARACTER-SET pathname components now work as
549 expected. (reported by Tim Daly Jr)
550 * fixed bug #391: complicated :TYPE intersections in slot
551 definitions no longer cause an error in PCL internals.
552 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
553 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
554 returning the number of octets which would be written to the
555 file-stream. (thanks to Robert J. Macomber)
556 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
557 arguments to the internal stream functions. (thanks to David
559 * bug fix: stack exhaustion is now handled on the x86/SunOS
560 platform. (thanks to Daisuke Homma)
561 * bug fix: more accurate ROOM results on GENCGC platforms
562 * optimization: improved type inference for arithmetic-for
563 index variables in LOOP
564 * optimization: faster floating-point SQRT on x86-64
565 * fixed some bugs revealed by Paul Dietz' test suite:
566 ** DOCUMENTATION returns NIL instead of "" for method combinations
567 that don't have a docstring
569 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
570 * minor incompatible change: (SETF CLASS-NAME) and (SETF
571 GENERIC-FUNCTION-NAME) are no longer generic functions, and
572 therefore are not customizeable by user code (as seems to be at
573 least permitted and maybe required by AMOP). As a consolation,
574 however, the SBCL implementation of these functions now calls
575 REINITIALIZE-INSTANCE as specified by AMOP.
576 * bug fix: slot-definition documentation is propagated as per ANSI
577 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
578 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
579 argument, as do eslotd initialization methods. (from Pascal
580 Costanza's "Closer" project)
581 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
582 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
584 * bug fix: it is now possible to have more than one subclass of
585 STANDARD-GENERIC-FUNCTION without causing stack overflow.
586 (reported by Bruno Haible, Pascal Costanza and others)
587 * bug fix: the dependent update protocol now works for generic
588 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
590 * bug fix: condition-class instances corresponding to
591 DEFINE-CONDITION forms are now created eagerly. (reported by
592 Kalle Olavi Niemitalo on comp.lang.lisp)
593 * bug fix: floating point printing is more accurate in some
594 circumstances. (thanks to Simon Alexander)
595 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
596 merged with *DEFAULT-PATHNAME-DEFAULTS*.
597 * bug fix: callbacks on OS X now preserve stack-alignment.
598 (thanks to Cyrus Harmon)
599 * enhancement: the x86-64 disassembler is much better at
600 disassembling SSE instructions. (thanks to Lutz Euler)
601 * enhancement: CHECK-TYPE on a variable now allows type-inference
602 (thanks to Vincent Arkesteijn)
603 * optimization: improved performance of EUC-JP external format.
604 (thanks to NIIMI Satoshi)
605 * optimization: performance improvements to IO on file streams of
606 :ELEMENT-TYPE CHARACTER
607 * optimization: much faster memory allocation on x86-64
608 * optimization: faster garbage collections (latency and throughput)
610 * optimization: faster bignum subtraction and division on x86 and x86-64
611 (thanks to Lutz Euler)
612 * optimization: more accurate type-inference for loops that use a
613 floating point index variable or a negative step.
615 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
616 * bug fix: add a workaround to SBCL looping infinitely at startup on
617 Linux kernels with apparently buggy implementations of personality().
618 (thanks to Svein Ove Aas)
619 * bug fix: Unicode symbols are correctly printed in LDB backtraces
620 (thanks to David Lichteblau)
621 * bug fix: local bindings shadow global symbol macros for
622 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
623 * optimization: non-open coded uses of numeric comparison operators
624 (e.g. >) no longer cons when called with more than one parameter
625 on platforms supporting dynamic-extent allocation.
626 * enhancement: saving cores with foreign code loaded is now supported
627 on MIPS/Linux in addition to the previously supported platforms.
628 * bug fix: FIND-RESTART now tests for activity, not applicability when given
629 a restart object as identifier. (reported by Helmut Eller for CMUCL)
630 * bug fix: division by zero in sb-sprof when no samples were collected
631 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
632 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
633 (thanks to Alexander Kjeldaas)
634 * new feature: ignoring the return values of destructive functions
635 like NREVERSE signals a compile-time style-warning.
636 (thanks to Kevin Reid)
637 * new feature: the alignment of alien structure fields can be
638 explicitly specified. (thanks to Cyrus Harmon)
639 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
640 and dump core on SIGQUIT
642 ** incompatible change: threads do not inherit values of specials
643 from their parents (see manual)
644 ** bug fix: threads stacks belonging to dead threads are freed by the
645 next exiting thread, no need to gc to collect thread stacks anymore
646 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
647 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
648 ** bug fix: dynamic variable and thread start related gc lossage
649 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
651 * fixed some bugs revealed by Paul Dietz' test suite:
652 ** SUBTYPEP is slightly more accurate on heinously complicated
653 CONS types where some of the members have uncertain (in the
654 NIL, NIL sense) type relationships to each other.
656 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
657 no longer purified unless :PURIFY T is explicitly specified.
658 ** Non-purified cores are significantly smaller than before
660 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
661 * new feature: timers based on Zach Beane's excellent timer package
662 * added support for the following external formats: koi8-u,
663 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
664 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
665 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
666 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
667 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
668 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
669 * incompatible change: a threaded SBCL will no longer revert to
670 non-threaded mode on non-NPTL systems, but refuse to start entirely.
671 * bug fix: interrupts are disabled until startup is complete; no
672 more sigsegvs when receiving a signal to soon
673 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
675 * bug fix: add a workaround for the memory randomization features in
676 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
677 workaround will only be in effect on systems with the proc filesystem
679 * bug fix: printing objects of type HASH-TABLE signals a
680 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
682 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
683 have been running for over 50 days. (reported by Gilbert Baumann)
684 * bug fix: the logic for getting names of functions gets less
685 confused when confronded with alternate-metaclass
686 funcallable-instances. (reported by Cyrus Harmon)
687 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
688 consistent, even on internal alternate-metaclass objects.
689 * bug fix: flush closure information collected by physical
690 environment analysis prepass before the main pass. (bug reported
692 * bug fix: compiler pack phase does not modify a hashtable, which is
693 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
694 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
695 may be expanded. (reported by Alexander <malishev> on c.l.l)
696 * bug fix: no more style-warnings when compiling code using LAST
697 under high-SPEED low-SPACE settings. (reported by David Wallin
699 * fixed some incompatibilities between SBCL's MOP and the MOP
701 ** the METAOBJECT class is now implemented;
702 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
703 STANDARD-OBJECT, as required;
704 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
705 now compatible (as required by VALIDATE-SUPERCLASS); there
706 remains a consistency requirement over the presence or absence
707 of the FUNCTION class in the superclass list of finalized
708 classes; see the manual for more details;
709 ** the :AROUND method for COMPUTE-SLOTS on
710 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
711 requested slot ordering.
713 ** bug fix: parent thread now can be gc'ed even with a live
715 ** bug fix: binding a special with PROGV to no value is not
716 the same as that symbol not having been bound (thanks to
718 ** bug fix: binding specials is thread safe (thanks to
720 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
721 and SB-SPROF do not die with 'no handler for signal XX in
722 interrupt_handle_now(..)' anymore
723 ** bug fix: WITH-TIMEOUT works with multiple running threads
724 * fixed some bugs revealed by Paul Dietz' test suite:
725 ** ENSURE-GENERIC-FUNCTION should take a method class object for
726 the :method-class keyword argument.
728 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
729 * new port: the Solaris operating system on x86 processors is now
730 mostly supported, though some rough edges in the environment
731 remain. (thanks to Daisuke Homma)
732 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
733 space for its heap. (thanks to Thiemo Seufer)
734 * optimization: SBCL on MIPS platforms now supports dynamic-extent
735 closures. (thanks to Thiemo Seufer)
736 * minor incompatible change: eof selects abort in the debugger.
737 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
738 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
740 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
741 deprecated, and will go away in a future revision of SBCL.
742 * minor incompatible change: GC-ON and GC-OFF are no longer
743 implemented with a counter, it does not matter how many times gc
744 is switched on or off
745 * bug fix: discriminating functions for generic function classes
746 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
747 make invalid assumptions about method precedence order. (reported
749 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
750 platforms. (thanks to Thiemo Seufer)
751 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
752 * bug fix: correct pathname printing with printer escaping is on.
753 (thanks to Kevin Reid)
754 * bug fix: complex VOP definitions in "user-space" no longer trigger
755 package locks. (reported by Zach Beane)
756 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
757 for funcallable-instance functions, and is no different from
759 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
760 characters in the current readtable. (reported by Nicholas Neuss)
761 * bug fix: SUBTYPEP on various CONS types returns more a more
762 accurate acknowledgment of its certainty.
763 * optimizations: REMOVE-DUPLICATES now runs in linear time on
764 lists in some cases. This partially fixes bug 384.
765 * flush all standard streams before prompting in the REPL and the
767 * bug fix: signal handling and triggering gc do not conflict
768 directly anymore, in particular a high frequency sb-sprof does
769 not prevent gc from running
770 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
771 approximation for timezone and DST information between the
772 universal time epoch and the smallest negative 32-bit time_t.
773 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
774 year 1899 so long as the encoded time is non-negative.
775 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
776 the correct answer for non-constant rotate amounts when performing
777 an inline 32-bit rotation.
779 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
780 there is only one thread in the session
781 ** bug fix: memory leak for streams created in one thread and
782 written to in another
783 ** bug fix: lockup when compiled with gcc4
784 ** bug fix: race that allows the gc to be triggered when gc is
786 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
787 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
789 ** bug fix: finished threads are gc'ed properly
790 * fixed some bugs revealed by Paul Dietz' test suite:
791 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
792 has arguments with a different set of applicable methods from
793 the orignal arguments.
794 ** The type error thrown by MAP now has a correct expected-type
796 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
797 name a compiled function.
798 ** Three MISC tests where a large bignum was improperly coerced to
799 a float that couldn't represent that bignum during type
800 derivation were fixed.
801 ** SUBTYPEP can now handle types involving the negation of a
802 list-form FUNCTION type.
803 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
804 as (COMPLEX (AND RATIO (NOT FIXNUM))).
805 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
807 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
808 * New feature: Experimental support for bivalent streams: streams
809 opened with :element-type :default now allow character and binary
810 (unsigned-byte 8) I/O
811 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
812 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
814 * bug fix: no more highly sporadic "couldn't check whether ~S is
815 readable" when reading a stream and an interrupt hits in the middle
816 of a select system call
817 * compiler better recognizes complex arrays (reported by Eduardo
819 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
820 for variables on SBCL built with linkage-tables. (reported by Luis
822 * various error reporting improvements.
823 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
824 (Thanks to Thiemo Seufer)
825 * bug fix: nested reader invokations work correctly
826 * bug fix: it is possible to have simultaneous references to foreign
827 code and foreign data with the same name.
829 ** added x86-64 support
830 ** incompatible change: the threading api now works with thread
831 objects instead of thread ids
832 ** bug fix: threads are protected from signals and interruption when
833 starting up or going down
834 ** bug fix: a race where an exiting thread could lose its stack to gc
835 ** bug fix: don't halt on infinite error in threads if possible
836 ** fixed numerous gc deadlocks introduced in the pthread merge
837 ** bug fix: fixed thread safety issues in read and print
838 ** bug fix: debugger doesn't hang on session lock if interrupted at
839 an inappropriate moment
840 ** bug fix: run-program is now thread safe(r)
841 ** bug fix: inner with-recursive-lock no longer releases the mutex
842 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
843 * fixed some bugs revealed by Paul Dietz' test suite:
844 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
845 EXPECTED-TYPE slots filled.
846 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
848 ** MAKE-CONDITION accepts classes as type-designators.
849 ** COMPILE may never return NIL.
850 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
851 range before calling Unix time functions
853 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
854 * numerous signal handling fixes to increase stability
855 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
856 * minor incompatible change: we now correctly canonize default
857 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
858 as per the MOP, rather than the historical (INITARG INITFUNCTION
859 INITFORM). (reported by Bruno Haible)
860 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
861 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
862 for more information.
863 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
864 pathname is a directory pathname.
865 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
866 * dynamic space size on PPC has been increased to 768Mb. (thanks to
868 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
869 :METACLASS argument in addition to a class name. (reported by
870 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
871 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
872 (reported by Bruno Haible; patch from Gerd Moellmann)
873 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
875 * bug fix: more cleanups to the floating point exception handling on
876 x86-64 (thanks to James Knight)
877 * bug fix: the compiler does not try to constant fold calls of
878 COERCE and BIT-* functions when they return freshly constructed
879 objects. (reported by Nikodemus Siivola and Paul Dietz)
880 * optimization: improved the MIPS versions of generic (in the
881 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
882 * optimization: direct conversion of (unsigned-byte 32) to floats on
883 the PowerPC platform.
884 * optimization: structure instances with raw slots now use less
885 memory, and probably show better memory locality. (thanks to
887 * optimization: DYNAMIC-EXTENT declarations for lists and closures
888 are treated as requests for stack allocation on the x86-64,
889 Alpha-32, PPC and SPARC platforms.
890 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
891 the wrong file descriptor; implementation of SOCKET-OPEN-P.
892 (thanks to Tony Martinez)
894 ** gcing a dead thread can no longer lead to lockups
895 ** threads block signals until they are set up properly
896 ** errno is no longer shared by threads
897 ** interrupt-thread restores the eflags register on x86
898 ** fixed some lockups due to gc/thread interaction
899 * fixed some bugs revealed by Paul Dietz' test suite:
900 ** invalid dotted lists no longer raise a read error when
902 ** signal an error if a symbol that names a declaration is used
903 as the name of a type, or vice versa
904 ** allow using the (declare (typespec var*)) abbreviation for
905 (declare (type typespec var*)) with all type specifiers
906 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
907 ** always return NIL from PPRINT-POP when OBJECT is NIL
908 ** don't signal errors when pretty-printing malformed LABELS,
909 FLET or MACROLET forms
910 ** declarations in a DOLIST body are also in scope for the
912 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
914 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
917 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
918 * fixed cross-compiler leakages that prevented building a 32-bit
919 target with a 64-bit host compiler.
920 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
921 opened with :IF-EXISTS :OVERWRITE.
922 * fixed bug 281: error for an invalid qualifier in a short-form method
923 combination method is not signalled until the faulty method is called.
924 * bug fix: iteration variable type inferrer failed to deal with open
925 intervals. (reported by Alan Shields)
926 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
927 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
928 or more runtime options were provided to the sbcl binary.
929 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
930 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
931 * optimization: there's now a fast-path for fixnum arguments in the
932 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
933 * optimization: the code generated on x86-64 is more compact thanks
934 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
935 * bug fix: floating-point exception handling now partly works on
936 x86-64. (Thanks to James Knight)
937 * improvement to the MIPS backend from Thiemo Seufer: C-style
938 64-bit long long arguments and return values to alien functions
940 * fixed some bugs revealed by Paul Dietz' test suite:
941 ** the type-error signalled from WARN has a filled-in DATUM slot.
942 ** the type-error required when a stream is not associated with
943 a file has the stream as its datum.
944 ** type-errors on single-floats on x86-64 no longer have
945 :INVALID-OBJECT as the datum
946 ** the type-errors signalled for invalid function names now have
947 a correct expected type
948 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
949 for structures defined with a :TYPE. Documentation strings for
950 typed structures are no longer immediately discarded
951 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
952 broadcast streams and synonym streams. FILE-LENGTH now also works
953 on broadcast streams.
955 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
956 * incompatible change: the --noprogrammer option, deprecated since
957 version 0.7.5, has been removed. Please use the equivalent
958 --disable-debugger option instead.
959 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
961 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
962 inoperational for a while now) has been completely removed.
963 * null lexical environments are now printed as #<NULL-LEXENV>,
964 significantly reducing the amount of clutter in typical backtraces.
965 * documentation on weak pointers, finalization, and after GC hooks
966 has been added to the manual.
967 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
968 as other specialized array types, is much faster. SUBSEQ and
969 COPY-SEQ on such arrays have also been sped up.
970 * optimization: EQL is now more efficient when at least other argument
971 is known to be of type (OR FIXNUM (NOT NUMBER)).
972 * fixed bug: the runtime is now less vulnerable to changes in the
973 size of the SBCL object on OS X, and virtual memory is reserved for
974 all spaces that need to be at a fixed address.
975 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
976 * fixed bug: finalizers and after GC hooks that cause consing are now
978 * fixed bug: compiler error messages and summaries are now printed to
979 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
980 * fixed inference of the upper bound of an iteration variable.
981 (reported by Rajat Datta).
982 * fixed bug 211e: calling local functions with duplicated constant
983 keyword argument no longer causes a bogus style warning about an
985 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
986 ftype declaration to be disregarded. (reported by Dave Roberts)
987 * fixed bug 373: caused by erronous compilation of references to alien
988 variables in the runtime on ppc/darwin.
989 * fixed bug 376: CONJUGATE type deriver.
990 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
991 (reported by Baughn on #lisp)
992 * a cross-compiler bug on non-x86 platforms has been identified and
993 fixed. (thanks to Bruno Haible)
994 * improvements to the MIPS runtime code for increased stability.
995 (thanks to Thiemo Seufer)
996 * increased the maximimum compact environment size to allow
997 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
998 * partial workaround for bug 135: don't name the function we're
999 calling for hairy cases of EVAL, so as not to accumulate one environment
1000 entry per call to EVAL. (reported by Kevin Reid)
1001 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1002 * optimization: type testing for non-vector arrays should be faster.
1003 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1005 * bug fix: setting 31st element of a bit vector to zero did not work
1007 * bug fix: redefining a class definition which failed due to a
1008 previous accessor / function clash now works (but see BUGS entry
1009 #380 for more problems in this area). (thanks to Zach Beane)
1010 * the long form of DEFINE-METHOD-COMBINATION disables method group
1011 checking when given a single method group with pattern *. (thanks
1013 * on x86 compiler supports stack allocation of results of simple
1014 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1015 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1016 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1017 * fixed some bugs related to Unicode integration:
1018 ** the restarts for recovering from input and output encoding
1019 errors only appear when there is in fact such an error to
1021 * fixed some bugs revealed by Paul Dietz' test suite:
1022 ** MISC.549 and similar: late transformation of unsafe type
1023 assertions into derived types caused unexpected code
1025 ** SCALE-FLOAT type deriver is less wrong.
1026 ** type derivers for EXP, LOG and similar functions compute result
1027 types for complex arguments better.
1028 ** (MISC.563) CONJUGATE type deriver works for very restricted
1030 ** out-of-line type testers for character strings are available.
1031 ** EQUAL compiler transform understands specialness of objects
1033 ** accessing double-floats stored on the stack now works on x86-64.
1034 ** debugger internals could sometimes create invalid lispobjs,
1035 resulting in GC crashes.
1036 ** MISC.548: type check weakening can convert required type into
1038 ** initialization forms for bindings are not in scope of free special
1041 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1042 * incompatible change: thread support for non-NPTL systems has
1043 been removed - locking is buggy and unreliable. A threaded
1044 SBCL build will now warn at startup and refuse to create
1045 new threads, unless futex support is detected in the kernel
1046 * incompatible change: the top level REPL now has only an ABORT
1047 restart associated with it, not TOPLEVEL and ABORT as it used to.
1048 TOP and TOPLEVEL are now available as debugger commands for
1049 returning to the top level.
1050 * incompatible change: forms evaluated in the REPL now use the
1051 global optimization policy.
1052 * incompatible change: user- and system-initialization files are
1053 no longer processed with LOAD, but by READ and EVAL; hence the
1054 global optimization policy, startup package, readtable, etc,
1056 * The .fasl file format number has been incremented because of
1057 various incompatible changes.
1058 * internal entry point details and argument counts no longer appear
1059 in backtraces unless explicitly requested by setting
1060 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1061 * built-in and standard functions no longer have names like "top
1062 level local call to FOO".
1063 * fixed bug 32: functions defined in non-null lexical environments
1064 now have more legible printed representation
1065 * fixed bug 33: functions defined in non-null lexical environemnts
1066 are now more amenable to inspection by INSPECT.
1067 * workaround for bug 354: XEPs no longer appear in backtraces unless
1068 explicitly requested.
1069 * fixed bug: receiving the signal which results from attempting to
1070 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1071 SIGBUS on other *BSDs) on architectures where the C stack is also
1072 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1073 to debugger confusion or wild instability. (reported by Cheuksan
1075 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1076 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1077 (reported by Lutz Euler)
1078 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1079 compiled code. (reported by Paul Dietz)
1080 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1081 specialized parameter to a method to something that is not TYPEP
1082 the specializer is now possible.
1083 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1084 face of package deletion.
1085 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1086 pathname. (reported by David Tolpin on c.l.l)
1087 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1088 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1089 STRING-STREAM and FILE-STREAM, respectively.
1090 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1091 than previously; in addition, its test suite is now run on build.
1092 (thanks to Luke Gorrie)
1093 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1094 * optimization: added a immediate representation for single-floats
1096 * fixed some bugs revealed by Paul Dietz' test suite:
1097 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1098 ** attempting to create a package with a colliding nickname causes
1099 correctable errors to be signalled.
1100 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1101 ** several x86-64 backend bugs related to sign-extension of immediate
1104 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1105 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1106 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1108 * fixed loading of multiply forward-referenced layouts.
1109 (thanks to Cheuksan Wang)
1110 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1111 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1112 (see COMPILE-FILE documentation for details of :PRINT option)
1113 * fixed bugs 19 and 317: fixed-format floating point printing is
1114 more accurate. This also fixes a bug reported by Adam Warner
1115 related to the ~@F format directive.
1116 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1118 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1119 dispatch table if the from-char is a dispatch macro character.
1120 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1121 by the wordsize no longer ignore the last word. (reported by Lutz
1123 * fixed bug in type checking of dynamic-extent variables. (reported
1125 * optimization: sequence traversal functions use their freedom to
1126 coerce function designators to functions.
1127 * optimization: code with many calls to CLOS methods specialized on
1128 CLOS classes has had redundant type checks removed.
1129 * optimization: added declarations to speed up operations that access
1130 the internal character database (for example STRING-UPCASE)
1131 * optimization: comparison operations between floats and sufficiently small
1132 fixnums no longer create extra rationals
1133 * fixed some bugs related to Unicode integration:
1134 ** portions of multibyte characters at the end of buffers for
1135 character-based file input are correctly transferred to the
1136 start of the buffer at the next read.
1137 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1138 passing it through to OPEN.
1139 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1140 argument given to internal calls to OPEN.
1141 ** the built-in comment readers (introduced by character sequences
1142 ";" and "#|") are more forgiving to encoding errors; they will
1143 STYLE-WARN and then attempt to resync the stream at a character
1144 boundary. (thanks to Teemu Kalvas)
1145 * fixed some bugs in the x86-64 port:
1146 ** Negative short int return values from c-calls are sign-extended
1148 ** The stack is aligned to 16-bytes for c-calls, as required by
1149 the ABI. (thanks to Cheuksan Wang)
1150 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1151 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1152 for enums. (thanks to Vincent Arkesteijn)
1153 ** Multiple small optimizations and bugfixes for floating point
1155 * fixed some bugs revealed by Paul Dietz' test suite:
1156 ** Space, Tab, Linefeed, Return and Page have the invalid
1157 secondary constituent character trait.
1158 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1160 ** WITH-INPUT-FROM-STRING should only update the index place on
1162 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1163 works more reliably.
1164 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1165 with one-dimensional arrays with fill pointers.
1166 ** TYPE-OF failed on a complex with an integer realpart and a
1168 ** compiler failure during type inference for the code of form
1169 (IF (EQL X (THE ...)) ...) (MISC.535).
1171 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1172 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1173 platform. The port supports 61-bit fixnums, large memory spaces
1174 and reloading shared object files.
1175 * enhancement: saving cores with foreign code loaded is now
1176 supported on ppc/Darwin in addition to the previously supported
1178 * enhancement: the statistical profiler now walks deeper into the
1179 call stack for more meaningful call-graphs and accrued time
1180 reports (x86/x86-64 only). It also now reports time spent in
1182 * enhancement: it is now possible to trace most individual methods
1183 of a generic function in addition to tracing the generic function
1185 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1187 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1188 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1189 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1190 produces an error. (thanks to Vincent Arkesteijn)
1191 * bug fix: NAMESTRING on pathnames with :WILD components in their
1192 directories works correctly. (thanks to Artem V. Andreev)
1193 * fixed bug 125: compiler preserves identity of closures. (reported
1195 * bug fixed: functions with &REST arguments sometimes failed with
1196 "Undefined function" when compiled with (DEBUG 3). (reported by
1198 * bug fix: overflow during compiling of setting element of a bit
1199 vector with constant index and value. (reported by Timmy Douglas)
1200 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1201 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1202 * on x86 compiler partially supports stack allocation of dynamic-extent
1204 * GO and RETURN-FROM do not check the extent of their exit points
1205 when compiled with SAFETY 0.
1206 * fixed some bugs related to Unicode integration:
1207 ** encoding and decoding errors are now much more robustly
1208 handled; it should now be possible to recover even from invalid
1209 input or output to the terminal. (thanks to Teemu Kalvas)
1210 ** provided a first cut at implementing STRING-TO-OCTETS and
1211 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1212 ** altered the SB-MD5 contributed module slightly, changing the
1213 interface just enough for it to be supportable for builds where
1214 lisp characters are not eight bits.
1215 * fixed some bugs revealed by Paul Dietz' test suite:
1216 ** the FORMATTER-generated functions for ~V[ conditionals require
1217 the correct number of arguments.
1218 ** READ-FROM-STRING returns the mandated second value when applied
1219 to displaced strings.
1220 ** the #\Rubout and #\Backspace characters are treated as invalid
1221 constituent characters by the tokenizer.
1223 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1224 * new feature: reloading changed shared object files with
1225 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1226 * new feature: references to foreign variables and functions
1227 can now be compiled and loaded before the corresponding shared
1228 object file is loaded, as long as the foreign definitions are
1229 available at runtime.
1230 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1231 supported. (thanks to Dan Debertin)
1232 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1233 just keywords. (thanks to Vincent Arkesteijn)
1234 * fixed bug #331: structure-class instances corresponding to
1235 DEFSTRUCT forms are now created eagerly.
1236 * fixed bug #345: backtraces from calls to undefined functions work
1237 on x86 as well. Related bug #61 is now also partially fixed on x86
1238 (backtraces from throws to unknown catch tags.)
1239 * bug fix: lambda-list parsing is now stricter vrt. order and number
1240 of lambda-list keywords.
1241 * bug fix: as specified by AMOP, an error is signalled if a
1242 class-option appears multiple times in a DEFCLASS form. (reported
1244 * bug fix: RANDOM can be compiled when the compiler derives the type
1245 of its numeric argument as a disjoint set of small integers.
1246 (reported by Paul Dietz)
1247 * bug fix: starting a core saved with shared objects loaded when
1248 those objects are not available no longer causes threaded SBCL to
1249 hang. (reported by Sean Ross)
1250 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1252 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1253 argument, which must always be NIL. (reported by Kalle Niemitalo)
1254 * bug fix: printing 1.0d+23 no longer results in an error.
1255 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1256 * bug fix: structure slot setters preserve evaluation order. (thanks
1258 * fixed some bugs related to Unicode integration:
1259 ** RUN-PROGRAM can allow its child to take input from a Lisp
1260 stream. (reported by Stefan Scholl)
1261 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1262 locales. (reported by Ken Causey)
1263 * fixed some bugs revealed by Paul Dietz' test suite:
1264 ** INCF, DECF and REMF evaluate their place form as specified in
1266 ** FORMATTER expands ~{ ~} iteration directives with V or #
1267 parameters correctly.
1268 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1269 consequent uses no arguments correctly.
1270 ** the system has a partial understanding of the (COMPLEX RATIO)
1272 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1275 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1276 * new feature: a build-time option (controlled by the :SB-UNICODE
1277 keyword feature, enabled by default) for building the system with
1278 support for the entire 21-bit character space defined by the
1280 * new feature: the system now has rudimentary external-format
1281 support; the primary user-visible change at this time is that
1282 characters with the high bit set (such as lower-case-e-acute) will
1283 print correctly to a terminal in a UTF-8 environment.
1284 * minor incompatible change: BASE-CHAR no longer names a class;
1285 however, CHARACTER continues to do so, as required by ANSI.
1286 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1287 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1288 has been moved to the SB-EXT package (temporarily re-exported from
1290 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1291 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1293 * fixed bug #351: better error-handlind and reporting for malformed
1295 * fixed bug #350: bignum-printing is now more memory-efficient,
1296 allowing printing of very large bignums, eg. (expt 2 10000000).
1297 (reported by Bruno Haible)
1298 * fixed bug #302: better primitive-type selection for intersection
1300 * fixed bug #308: non-graphic characters now all have names, as
1301 required. (reported by Bruno Haible)
1302 * bug fix: redefining a class with different superclasses now correctly
1303 removes it from the direct-subclasses of its previous superclasses.
1304 (reported by David Morse)
1305 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1306 new value now works. (reported by Bruno Haible)
1307 * bug fix: correct canonicalization of multiple non-standard slot
1308 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1309 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1310 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1311 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1312 now exists, an signals an error.
1313 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1314 during macroexpansion for non-iteration variables as well. (reported
1315 by Bruno Haible for CMUCL)
1316 * bug fix: Cyclic structures and unprintable objects in compiler
1317 messages no longer cause errors. (reported by Bruno Haible)
1318 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1319 and READ-FROM-STRING all now return a primary value of NIL if
1320 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1321 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1322 (reported by Bruno Haible for CMUCL)
1323 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1324 do not designate a condition. (reported by Bruno Haible for
1326 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1327 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1328 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1329 permitting resolution in favour of any of the conflicting symbols.
1330 (reported by Bruno Haible for CMUCL)
1331 * bug fix: EQUAL compiler optimizations is less aggressive on
1332 strings which can potentially compare true despite having distinct
1333 specialized array element types.
1334 * bug fix: unit enumerations can be defined without dividing by
1335 zero. (thanks to Vincent Arkesteijn)
1336 * FORMAT compile-time argument count checking has been enhanced.
1337 (report from Bruno Haible for CMUCL)
1338 * a partial workaround for the bug 262: the compiler does not try to
1339 inline-expand a local function doing RETURN-FROM from a deleted
1340 BLOCK. (thanks to Peter Denno for the bug report and to David
1341 Wragg for the simple test case)
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** RENAME-PACKAGE allows all package designators as new package
1345 ** constraint propagation and lambda variable substitution are
1346 more cautious in dealing with partially deleted code.
1347 ** compiler sometimes lost reoptimization passes.
1348 ** CERROR, given a condition as condition designator, passes its
1349 remaining arguments to the continue format control without
1351 ** Case-altering FORMAT directives work correctly on non-ASCII
1353 ** The REAL type specifier handles bounds outside the floating
1354 point ranges without signalling FLOATING-POINT-OVERFLOW.
1355 ** Functions with IR1-transformations can create intercomponent
1356 references to global functions.
1357 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1359 ** FORMAT ~R treats a nil value for its first parameter correctly.
1361 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1362 * enhancement: saving cores with foreign code loaded is now
1363 supported on x86/NetBSD and sparc/Linux in addition to the previously
1364 supported platforms.
1365 * bug fix: on some platforms repeated installations caused multiple
1366 copies of HTML documentation to be installed -- should not happen
1367 any more. (reported by Stefan Scholl)
1368 * bug fix: parsing self-recursive alien record types multiple times
1369 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1370 original patch by Helmut Eller for CMUCL)
1371 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1372 (thanks to Richard Kreuter)
1373 * bug fix: defining classes whose accessors are methods on existing
1374 generic functions in other (locked) packages no longer signals
1375 bogus package lock violations. (reported by François-René Rideau)
1376 * bug fix: special variables as DEFMETHOD parameters no longer have
1377 associated bogus type declarations. (reported by David Wragg and
1379 * bug fix: read-write consistency on streams of element-type
1380 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1381 * bug fix: redefiniton of the only method of a generic function with
1382 no DEFGENERIC no longer emits a full WARNING. In addition,
1383 redefinition of generic functions with no DEFGENERIC to an
1384 incompatible lambda list now signals an error. (thanks to Zach
1386 * bug fix: DEFGENERIC now works even when there's a function of the
1387 same name in an enclosing lexical environment. (thanks to Zach
1389 * fixed compiler failure, caused by instrumenting code during
1390 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1391 * optimization: added loop analysis and improved register allocation
1392 * fixed some bugs revealed by Paul Dietz' test suite:
1393 ** POSITION on displaced vectors with non-zero displacement
1394 returns the right answer.
1395 ** (SIMPLE-STRING) is a valid type specifier for sequence
1397 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1399 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1400 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1402 ** The justification version of the FORMAT ~< directive treats
1403 non-zero minpad parameter correctly.
1405 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1406 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1407 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1408 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1409 the supported interface.
1410 * new feature: Single-stepping of code compiled with DEBUG 2 or
1411 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1412 * new feature: saving cores with foreign code loaded is now
1413 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1414 Timothy Moore's work for CMUCL)
1415 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1416 parameters to * instead of NIL if no initform is supplied.
1417 (reported by Johan Bockgaard)
1418 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1419 correctly when FUNCALL appears as the car of the form. Note:
1420 despite this FUNCALL forms are not currently subject to
1421 compiler-macro expansion. (port of Raymond Toy's fix for the
1422 same from CMUCL, reported by Johan Bockgaard)
1423 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1424 (thanks for Teemu Kalvas)
1425 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1426 case instead of terminating the iteration (thanks for Julian
1427 Squires, Sean Champ and Raymond Toy)
1428 * bug fix: incorrect expansion of defgeneric that caused a style
1429 warning. (thanks for Zach Beane)
1430 * bug fix: slot accessor effective method computation works properly
1431 for classes with multiple non-standard applicable methods on
1432 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1433 * on x86 compiler supports stack allocation of results of LIST and
1434 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1435 CMUCL implementation by Gerd Moellmann)
1436 * fixed some bugs revealed by Paul Dietz' test suite:
1437 ** FORMAT strings with both the ~<~:;~> form of the justification
1438 directive and pretty-printing directives cause an error.
1440 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1441 * incompatible change: the internal functions
1442 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1443 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1444 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1445 instead of the old functions.
1446 * new feature: on platforms where "dladdr" is available foreign
1447 function names now appear in backtraces. (based on Helmut Eller's
1449 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1450 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1452 * documentation: ASDF manual is now distributed with SBCL.
1453 * bug fix: SBCL can now load its contributed modules with REQUIRE
1454 even if the system-provided entries have been removed from
1455 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1457 * bug fix: backtraces involving undefined functions or assembly
1458 routines are more informative. (thanks to Brian Downing)
1459 * bug fix: mutually referent alien structures now work correctly.
1460 (reported by Rick Taube)
1461 * bug fix: structures defined by WITH-ALIEN can be referred to
1462 within other definitions in the same WITH-ALIEN.
1463 * bug fix: division operators (MOD, TRUNCATE and the like) with
1464 constant zero divisors and integer dividends no longer generate
1466 * bug fix: provide default methods for INPUT-STREAM-P and
1467 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1468 * bug fix: improve the detection and resolution of MOP metacycles
1469 (where effective method computation depends on the generic function
1470 having its effective method computed). (reported by Bruno Haible)
1471 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1472 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1473 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1474 * bug fix: pathnames with relative directory components can be
1475 represented relative to default pathnames.
1476 * optimization: in taking the GCD of bignums, reduce the two bignums
1477 to approximately the same size (using Euclid's algorithm) before
1478 applying the more sophisticated binary GCD. (thanks to Juho
1480 * optimization: COUNT on bitvectors now operates word-at-a-time.
1481 * optimization: ASH with a positive, but not necessarily constant,
1482 (leftwards) shift, when in a modular context, is compiled to a
1484 * fixed some bugs revealed by Paul Dietz' test suite:
1485 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1486 if the corresponding argument is NIL.
1487 ** FORMAT directives accepting a minpad parameter treat negative
1489 ** literal commas as character parameters to FORMAT directives are
1491 ** literal spaces directly after ~<Newline> directives within a
1492 format-logical-block (~:< ~@:>) do not induce :FILL-style
1493 conditional newlines.
1494 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1495 as specified: it no longer includes conditional newlines).
1496 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1498 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1499 :PER-LINE-PREFIX argument does not evaluate to a string.
1501 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1502 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1503 the manual for details; for now, package locks can be disabled by
1504 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1505 if no major problems are found then it is likely that they will be
1506 compiled in unconditionally.
1507 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1508 unsupported operators on all platforms. To load a shared library
1509 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1510 object file, link it into a shared library outside of SBCL and
1511 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1513 * bug workaround: ROOM T and the GC were not working together
1514 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1515 reporting facility aren't preserved. That reporting has been
1516 disabled, so now until and unless someone figures out how to make it
1517 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1518 * minor incompatible change: as threatened around sbcl-0.8.0, the
1519 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1520 The symbols are also exported from SB-PCL for backwards
1521 compatibility, but more so than before SB-PCL should be treated as
1522 an implementation-internal package.
1523 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1525 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1526 It is known as of this release not to work on the Alpha, however.
1527 * fixed bug #167: errors signalled due to illegal syntax in method
1528 bodies are now more legible.
1529 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1530 designators and can hence be used with TYPEP.
1531 * fixed bug #333: CHECK-TYPE now ensures that the type error
1532 signalled, if any, has the right object to be accessed by
1533 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1534 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1535 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1537 * fixed bug #334: programmatic addition of slots using specialized
1538 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1539 and :ALLOCATION :CLASS slots.
1540 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1541 float exponents. (rereported by Peter Seibel)
1542 * fixed a bug: #\Space (and other whitespace characters) are no
1543 longer considered to be macro characters in standard syntax by
1544 GET-MACRO-CHARACTER.
1545 * fixed bug: initialization of condition class metaobjects no longer
1546 causes an instance of the condition to be created. (reported by
1548 * fixed bug: it is now possible to have slots such that
1549 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1550 description is neither :INSTANCE nor :CLASS.
1551 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1552 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1553 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1554 non-standard SLOT-DEFINITION-ALLOCATION values.
1555 * fixed bugs: various race conditions handling exiting threads.
1556 CL-PPCRE's thread torture test now passes.
1557 * fixed arguably-a-bug: GC time proportional to number of threads in
1558 system even when most of them are idle
1559 * optimization: improved performance of BIT and SBIT on bit-vectors.
1560 * .fasl file incompatibility: The fasl file version number has
1561 been incremented because of changes associated with package locks.
1563 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1564 * minor incompatible change: the system no longer provides
1565 optimization hints (conditions which are TYPEP
1566 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1567 mode; these hints are still emitted when higher SPEED optimization
1569 * new contrib module: a sampling profiler (profiling by statistical
1570 sampling, rather than by instrumenting functions) is available as
1571 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1572 * the behaviour of the standard function ED is now customizeable by
1573 third parties through a hook variable: see ED's documentation
1574 string for information on the protocol.
1575 * the compiler no longer emits efficiency notes for (FUNCALL X)
1576 when the type of X is uncertain under default optimization
1578 * fixed bug 276: mutating a binding of a specialized parameter to a
1579 method to something that is not TYPEP the specializer is now
1581 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1582 exhibit the required behaviour on the x86 platform. (thanks to
1583 Peter van Eynde, Eric Marsden and Bruno Haible)
1584 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1585 even for difficult arguments. (reported by Peter Graves)
1586 * fixed bug 141a: the backquote printer now descends quoted
1588 * fixed another bug in backquote printing: no more destructive
1589 modification of the form's list structure. (reported by Brian
1591 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1592 move between its address being taken and the call to
1593 interrupt_thread, fixing a crashing race condition.
1594 * the SB-POSIX contrib implementation has been adjusted so that it
1595 no longer exhibits ridiculously poor performance when constructing
1596 instances corresponding to C structs.
1598 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1599 * minor incompatible change: the sb-grovel contrib now treats C
1600 structures as alien (in the sense of SB-ALIEN) objects rather than
1601 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1602 has implications for memory management of client code
1603 (sb-grovel-returned objects must now be manually managed) and for
1604 type safety (alien objects now have full types).
1605 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1606 used to control emission of compiler diagnostics, rather than the
1607 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1608 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1609 quality should be considered deprecated.
1610 * install.sh now installs the user manual as well
1611 * (not quite a new documentable feature, but worth considering in
1612 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1613 beginnings of a semantically meaningful condition hierarchy is
1614 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1616 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1617 designator as the defaults argument.
1618 * fixed bug: Displaced arrays whose displaced-to array has become
1619 too small now cause ARRAY-DIMENSION to signal an error, providing
1620 for safer bounds-checking. (reported by Bruno Haible)
1621 * fixed bug: DEFCLASS slot definitions with identical :READER and
1622 :WRITER names now signal a reasonable error. (reported by Thomas
1624 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1626 * fixed bug: Invalid slot specification errors now print correctly.
1627 (thanks to Zach Beane)
1628 * fixed bug 320: Shared to local slot value transfers in class
1629 redefinitions now happen corrently from superclasses as
1630 well. (reported by Bruno Haible)
1631 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1633 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1634 improper lists correctly. (reported by Bruno Haible)
1635 * fixed bug 313: source-transform for <fun-name> was erroneously
1636 applied to a call of a value of a variable with name <fun-name>.
1637 (reported by Antonio Menezes Leitao)
1638 * fixed bug 307: The obsolete instance protocol ensures that
1639 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1640 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1641 remove dead unknown-values globs from the middle of the stack.
1642 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1643 easier to resume long computations after using *BREAK-ON-SIGNALS*
1644 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1645 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1646 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1647 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1648 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1649 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1651 * fixed a (fairly theoretical) bug in string printing: if
1652 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1653 does not have array-element-type equal to the most general string
1655 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1656 function name. (thanks to Nikodemus Siivola)
1657 * fixed bug: compile-time format string checker failed on
1658 non-closed ~{. (reported by Thomas F Burdick)
1659 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1660 don't warn on legal loop code involving more than one aggregate
1661 boolean. (thanks to Nikodemus Siivola)
1662 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1663 in keyword position to MAKE-PACKAGE was not being reported as
1665 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1666 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1668 * fixed bugs 280 and 312: the checking for multiple definitions in a
1669 file is less likely to become confused by uses of inline
1671 * fixed bug: the #S reader macro performs the keyword coercion
1672 specified for slot names. (reported by Kalle Niemitalo)
1673 * fixed bug: lambda lists may contain symbols whose names start with
1674 & but are not lambda-list-keywords; their occurrence triggers a
1676 * fixed bug 321: define-method-combination argument lambda lists do
1677 not cause invalid code to be generated when &optional or &aux
1678 variables have default values. (reported by Bruno Haible)
1679 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1680 have CLOS classes; this appears also to have fixed reported
1681 instability in stack exhaustion detection.
1682 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1683 MUFFLE-WARNING when no associated restart is present is now
1685 * optimization: rearranged the expansion of various defining macros
1686 so that each expands into only one top-level form in a
1687 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1689 * optimization: used a previously little-used slot in symbols to
1690 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1692 * fixed some bugs revealed by Paul Dietz' test suite:
1693 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1695 ** Fixed an optimization bug related to inheritance of initforms
1696 from local to shared slots.
1697 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1698 ** CAST optimizer forgot to flush argument derived type.
1699 ** print/read consistency on floats is now orders of magnitude
1700 more likely. (thanks also to Bruno Haible for a similar report
1702 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1703 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1704 for objects of type REAL. Make it so.
1705 ** SXHASH is specified (infelicitously) to respect similarity,
1706 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1707 it so. (thanks to Markus Ziegler)
1708 ** on the Alpha, the compiler succeeds in compiling functions
1709 returning a known number of arguments greater than 63.
1710 ** fixed handling of invalid NIL arguments in keyword position
1712 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1714 ** provided a readably-printable representation for RANDOM-STATE
1716 ** ensured that pathnames, where they have a namestring, always
1717 print using #P"..." syntax.
1719 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1720 * Support for the forthcoming 2.0 version of the NetBSD kernel
1721 running on x86 hardware has been added. (thanks to Perry
1722 E. Metzger most immediately, and others for their past work)
1723 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1724 current release version; SBCL's OpenBSD support had been broken
1725 since about the time of OpenBSD's switch to ELF binary format.)
1726 * [placeholder for DX summary]
1727 ** user code with &REST lists declared dynamic-extent, under high
1728 speed or space and low safety and debug optimization policy.
1729 * The manual has been converted to Texinfo format and the debugger
1730 chapter from the cmucl manual has been added.
1731 * A facility has been added to extract documentation strings from
1732 sbcl and store them as Texinfo-formatted snippets for inclusion in
1733 the manual (via Texinfo's @include directive)
1734 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1735 the test case to Dave Roberts)
1736 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1737 pointers. (reported by Sean Ross)
1738 * bug fix: PROFILE output is printed nicely even for large numerical
1739 values. (thanks to Zach Beane)
1740 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1741 greater than 32 handle EOF correctly.
1742 * bug fix: on X86 an immediate argument of the IMUL instruction is
1743 correctly printed in disassembly. (thanks to Lutz Euler)
1744 * bug fix: class slots in redefined classes preserve their old
1745 values. (thanks to Bruno Haible and Nikodemus Siivola)
1746 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1747 succeeds. (reported by Marco Baringer)
1748 * fixed some bugs revealed by Paul Dietz' test suite:
1749 ** READ-SEQUENCE now works on ECHO-STREAMs.
1750 ** RATIONALIZE works more according to its specification. (thanks
1753 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1754 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1755 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1756 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1757 should matter to you only if you rebind the printer control
1758 variables and then find you want different bindings in the
1759 debugger than in the ordinary execution of your program.)
1760 * The runtime build system has been tweaked to support building
1761 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1762 assembler and linker. (thanks to Nikodemus Siivola)
1763 * Unbound, undefined, undeclared variables now trigger full
1764 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1765 more often programmer error than deliberate exploitation of undefined
1767 * optimization: the hash algorithm for strings has changed to one
1768 that is less vulnerable to spurious collisions. (thanks to Juho
1770 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1771 needless bounds checking. (thanks to Juho Snellman)
1772 * optimization: implemented multiplication as a modular
1773 (UNSIGNED-BYTE 32) operation on the PPC backend.
1774 * fixed some bugs revealed by Paul Dietz' test suite:
1775 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1777 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1778 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1779 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1781 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1783 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1785 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1786 their output stream on EOF from read.
1787 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1788 have been read to end-of-file.
1789 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1791 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1792 description of determination of which consecutive characters
1794 ** Printing the "Space" character with escaping on now yields "#\\ ",
1795 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1796 ** Reading floating-point numbers with *READ-BASE* set to a number
1797 less than 10 works correctly.
1798 ** Reading floating-point numbers with *READ-BASE* set to a number
1799 more than 10 works correctly.
1800 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1801 the readtable currently in effect.
1803 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1804 * minor incompatible change: parsing of namestrings on a physical
1805 (Unix) host has changed; numbers after the final #\. in a
1806 namestring are no longer interpreted as a version field. This is
1807 intented to be largely invisible to the user, except that the
1808 meaning of the namestring "*.*.*" has changed: it now refers to a
1809 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1810 should usually be replaced by
1811 (make-pathname :name :wild :type :wild :version :wild)
1812 with the added benefit that this is more likely to be portable.
1813 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1814 option to OPEN now signals an error if the file being opened
1815 exists; this may have an impact on existing code.
1816 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1817 signals received, on the PowerPC platforms (both Linux and
1818 Darwin). (thanks to Pierre Mai for pointing out the location of
1820 * several fixes on OS X: The system now builds and runs cleanly on
1821 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1822 (thanks to Brian Mastenbrook)
1823 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1824 second-resolution: integer multiples of 1/3600 between -24 and 24.
1825 (thanks to Vincent Arkesteijn)
1826 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1827 type when called with 1 argument; PEEK-CHAR checked type of
1828 PEEK-TYPE only after having read first character from a
1829 stream. (reported by Peter Graves)
1830 * bug fix: the garbage collector now has much better locality
1831 behaviour, and in particular no longer treats all memory as being
1832 exhausted when many small objects point to each other in a deeply
1834 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1835 recognized as being TYPEP their class.
1836 * bug fix: the PUSHNEW documentation string has been corrected.
1837 (thanks to Vincent Arkesteijn)
1838 * bug fix: defaulting of the value for the last of an atypically
1839 large number of multiple values being bound was not being
1840 performed correctly on the Alpha or PPC platforms
1841 * optimization: implemented multiplication as a modular
1842 (UNSIGNED-BYTE 32) operation on the x86 backend.
1843 * optimization: SEARCH on simple-base-strings can now be open-coded.
1844 (see also contrib/compiler-extras.lisp for inspiration for
1845 teaching the compiler about the Boyer-Moore algorithm).
1846 * value, returned by MAX (and MIN) called with several EQUALP, but
1847 not EQL, arguments now does not depend on compiler settings.
1848 * fixed some bugs revealed by Paul Dietz' test suite:
1849 ** in stack analysis liveness information is propagated from
1850 non-local entry points.
1851 ** pathwise CAST removing failed when the CAST node did not start
1853 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1854 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1856 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1857 host is already defined.
1858 ** RENAME-FILE works on streams instead of signalling an internal
1860 ** PEEK-CHAR uses the current readtable when determining whether
1861 or not a character is whitespace.
1862 ** MERGE-PATHNAMES handles the case when the pathname does not
1863 specify a name while the default-pathname specifies a version.
1864 ** Pathnames now stand a better chance of respecting print/read
1866 ** Attempting to use standardized file system operators with a
1867 pathname with invalid :DIRECTORY components signals a
1869 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1871 ** DIRECTORY on logical pathnames is more correct.
1872 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1873 signal a TYPE-ERROR if their argument is not a stream.
1874 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1875 designator argument does not designate a stream.
1876 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1877 examining the synonym.
1878 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1880 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1881 element-type larger than ([UN]SIGNED-BYTE 32).
1883 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1884 * When built with the :SB-FUTEX feature, threaded builds now take
1885 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1886 for faster/more reliable mutex and condition variable support.
1887 * Incompatible change (but one you probably shouldn't have been using
1888 anyway): the interface and code for arbitrating between multiple
1889 threads in the same user session has been redesigned.
1890 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1891 error when called without an explicit environment argument.
1892 (thanks to Nikodemus Siivola)
1893 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1894 confused about their position. (thanks to Adam Warner and Gerd
1896 * bug fix: Pretty printing backquoted forms with unquotations in the
1897 argument list position of various code constructs such as LAMBDA
1898 now works correctly. (reported by Paul Dietz)
1899 * bug fix: Pretty printing unquotations no longer loses all
1900 stream position information.
1901 * optimization: performance of string output streams is now less
1902 poor for multiple small sequence writes.
1903 * optimization: performance of CSUBTYPEP in the presence of complex
1904 expressions involving CONS and NOT many times has been improved.
1905 (reported by Paul Dietz)
1906 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1908 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1909 streams for the REPL, for the debugger, and so on. (thanks to
1911 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1912 * fixed some bugs revealed by Paul Dietz' test suite:
1913 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1914 is ignored if it is too small, rather than propagating through
1915 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1916 ** extremely complex negations of CONS types were not being
1917 sufficiently canonicalized, leading to inconsistencies in
1919 ** VALUES tranformer lost derived type.
1921 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1922 * fixed a bootstrapping bug: the build process no longer assumes
1923 that the various BOOLE-related constants have the same value in
1924 host and target lisps. (noted by Paul Dietz' test suite on an
1925 SBCL binary built from CLISP)
1926 * The system can now be dynamically linked on the MIPS platform,
1927 which enables dynamic loading of foreign code from Lisp. (thanks
1928 to Ralf Baechle for discussions on the MIPS ABI)
1929 * The system now records debugging information for its own source
1930 files in a filesystem-position-independent manner, relative to
1931 the "SYS" logical host.
1932 * fixed a compiler bug: MV-LET convertion did not check references
1933 to the "max args" entry point. (reported by Brian Downing)
1934 * tweaked disassembly notes to be less confident about proclaiming
1935 some instruction as an LRA. (thanks to Brian Downing)
1936 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1937 now each have their own history, command character, and other
1938 characteristics. (thanks to David Lichteblau)
1939 * fixed some bugs revealed by Paul Dietz' test suite:
1940 ** compiler failure in compiling LOGAND expressions including a
1942 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1943 shift greater than 32.
1944 ** FLUSH-DEST did not mark blocks for type check regeneration.
1945 ** HANDLER-CASE failed to accept declarations in handler clauses
1946 in some circumstances.
1948 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1949 * New code in contrib/sb-introspect (still probably not entirely
1950 stable yet) provides some support for smart Lisp development
1951 environments like SLIME.
1952 * The conditions signalled for errors occurring when loading .fasl
1953 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1954 in a way which should help ASDF recover gracefully.
1955 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1956 stuff has been cleaned up. If you code contrib/ stuff, this might
1957 affect you, and you can look at contrib/README, contrib/STANDARDS,
1958 and/or the 0.8.4.27 diff to check.
1959 * In full calls the compiler now does not generate checks for declared
1960 argument types for all arguments.
1961 * various threading fixes
1962 ** and some experimental patches which didn't make it into
1963 the main tree for this release, but which are shipped in
1964 contrib/experimental-thread.patch as a possible fix for some
1965 failures (deadlock, spinning...) in GC-intensive multithreaded
1967 * fixed PPC build problem (source code incompatibility of different
1968 library versions): added offsetof() hackery which attempts to divine
1969 where glibc maintainers put uc_mcontext today
1970 * fixed bug 282: compiler does not trust type assertions while passing
1971 arguments to a full call.
1972 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1973 &OPTIONAL VALUES type parameter.
1974 * fix bug 214: algorithm for noting rejected templates is now more
1975 similar to that of template seletion. (also reported by rydis on
1977 * fixed bug 141b: printing backquoted information readably and prettily
1978 inserts a space where necessary.
1979 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1980 (CHARACTER 10) are now reported as errors, rather than propagated
1981 as unknown types. (reported by piso on #lisp)
1982 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1983 with values NIL and :ERROR. (thanks to Milan Zamazal)
1984 * fixed bug 191c: CLOS now does proper keyword argument checking as
1985 described in CLHS 7.6.5 and 7.6.5.1.
1986 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1987 longer raise an error; further, using a list as a for-as-arithmetic
1988 counter now raises a meaningful error.
1989 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1990 now understood by sequence creation functions such as MAKE-SEQUENCE
1992 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1993 a STRING-INPUT-STREAM.
1994 * compiler enhancement: SIGNUM is now better able to derive the type
1996 * type declarations inside WITH-SLOTS are checked. (reported by
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** incorrect optimization of TRUNCATE for a positive first
2000 argument and negative second.
2001 ** compiler failure in let-convertion during flushing dead code.
2002 ** compiler failure while deriving type of TRUNCATE on an
2003 interval, containing 0.
2004 ** ASH of a negative bignum by a negative bignum count now returns
2006 ** intersection of CONS types now canonicalizes properly, fixing
2007 inconsistencies in SUBTYPEP.
2009 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2010 * incompatible change: The --disable-debugger command line
2011 option now clobbers the debugger at a more fundamental
2012 level, by redefining #'INVOKE-DEBUGGER instead of by
2013 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2014 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2015 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2016 enter the debugger REPL and then suffer recursive errors
2017 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2018 while under the new system, BREAK in --disable-debugger mode
2019 terminates the system just as an unhandled error would.
2020 * fixed compiler performance when processing loops with a step >1;
2021 * bug fix: DOCUMENTATION now retrieves generic function
2022 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2023 support has been systematized, and now supports the methods
2024 specified by ANSI, along with a default method and a method for
2025 slot documentation. (reported by Nathan Froyd)
2026 * bug fix: effective methods associated with a generic function are
2027 no longer cached over a change of that generic function's method
2028 combination. (reported by Andreas Fuchs)
2029 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2030 names a non-existent directory. (thanks to Andreas Fuchs)
2031 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2032 the PPC platform, lead to stack corruption; this has been fixed.
2033 (reported by Rainer Joswig)
2034 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2035 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2036 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2037 arithmetic in general, is now much more reliable on the PPC
2039 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2040 platform now returns the right answer.
2041 * optimization: restored some effective method precomputation in
2042 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2043 precomputation is now tunable.
2044 * optimization: compiler-internal data structure use has been
2045 reviewed, and changes have been made that should improve the
2046 performance of the compiler by about 20%.
2047 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2048 simple-base-strings and simple-bit-vectors is improved.
2049 * optimization: the optimization of 32-bit logical and arithmetic
2050 functions introduced in version 0.8.3 on the x86 has been
2051 implemented on the mips, ppc and sparc platforms; an
2052 implementation of the same facility, but for 64-bit arithmetic,
2053 has been added for the alpha.
2054 * microoptimization: the compiler is better able to make use of the
2055 x86 LEA instruction for multiplication by constants.
2056 * bug fix: in some situations compiler did not report usage of
2057 generic arithmetic in (SPEED 3) policy.
2058 * bug 145b fix: compiler used wrong type specifier while converting
2059 MEMBER-types to numeric.
2060 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2062 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2063 index argument now works properly on the Alpha platform.
2064 * bug fix: floating point exception treatment on the Alpha platform
2066 * bug fix: FILE-POSITION works much better on string input and
2067 output streams. (thanks to Nikodemus Siivola)
2068 * bug fix: many threading/garbage collection symptoms sorted.
2069 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2070 might be pseudo-atomic.
2071 * internal change: Stopping for GC is now done with signals not
2072 ptrace. GC is now done in whichever thread wanted it, instead of
2074 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2076 * bug fix: result form in DO is not contained in the implicit
2078 * incompatible change: ICR structure is changed; the value part of
2079 CONTINUATION is now called LVAR; corresponding functions are
2080 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2081 * added type deriver for ISQRT (thanks to Robert E. Brown).
2082 * bug fix: better support for loading from the command line when an
2083 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2085 * fixed some bugs revealed by Paul Dietz' test suite:
2086 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2087 ** ROUND and FROUND now give the right answer when given very
2088 small float arguments.
2089 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2091 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2092 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2093 ** LCM should return a non-negative integer.
2094 ** PARSE-INTEGER returned the index of a terminator instead of the
2095 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2096 ** PARSE-INTEGER returned an incorrect index being applied to a
2098 ** LCM with two arguments of 0 returns 0 rather than signalling
2100 ** unsigned addition of a 32-bit constant with the high bit set no
2101 longer causes an internal compiler error on the x86.
2102 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2103 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2104 with negative last argument.
2105 ** byte specifiers with zero size and position no longer cause
2106 an error during type derivation.
2107 ** bignum multiplication on the Alpha platform now returns the
2109 * porting: The system now builds on SuSE AMD64, although it still
2110 generates a 32-bit binary.
2111 * .fasl file incompatibility: The fasl file version number has
2112 been incremented (because of the changes to internal compiler
2113 data structures referred to above).
2115 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2116 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2117 more accurately, on the Darwin kernel running on PowerPC hardware.
2118 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2119 * Compiler code deletion notes now signal a condition of type
2120 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2121 an associated MUFFLE-WARNING restart.
2122 * The compiler now performs limited argument count validation of
2123 constant format strings in FORMAT, and where appropriate in ERROR,
2124 CERROR and WARN. (thanks to Gerd Moellmann)
2125 * New ASDF-INSTALL contrib can be used for automatic download and
2126 installation of third-party Lisp code from CCLAN or other sites
2128 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2129 which forces another thread to execute a function supplied by the
2131 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2132 now accept and act upon their :ELEMENT-TYPE keyword argument.
2133 (reported by Martin Atzmueller, Edi Weitz)
2134 * bug fix: FILE-POSITION now accepts position designators up to
2135 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2136 is the greater. (thanks to Patrik Nordebo)
2137 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2138 Kalle Olavi Niemitalo)
2139 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2140 printable. (reported by Eric Marsden)
2141 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2142 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2143 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2144 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2145 (thanks to Stig Sandoe)
2146 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2147 UNREAD-CHAR now correctly includes the unread character in the
2148 target sequence. (thanks to Gerd Moellmann)
2149 * bug fix in threaded builds: the system can now be suspended and
2150 resumed by shell job control with minimal disruption.
2151 * bug fixes in times and timezones >2038AD
2152 * better handling of "where is GNU make?" problem in build scripts
2153 (thanks to Nikodemus Siivola)
2154 * new optimization: inside a named function any reference to a
2155 function with the same name is considered to be a self-reference;
2156 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2157 optimization quality.
2158 * new optimization on x86: logical functions and + now have
2159 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2160 used when the result is truncated to 32 bits.
2161 * VALUES declaration is partially enabled.
2162 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2163 * fixed some bugs revealed by Paul Dietz' test suite:
2164 ** The system now obeys the constraint imposed by
2165 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2166 types form a lattice under type intersection.
2167 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2168 ** ASSOC now ignores NIL elements in an alist.
2169 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2170 and (1+ MOST-POSITIVE-FIXNUM) answers.
2171 ** The addition of a method with invalid qualifiers to a generic
2172 function does not cause an error to be signalled immediately;
2173 a warning is signalled, and the error is generated only on
2174 calling the generic function.
2175 * changed .fasl file version number, in part to add type codes for
2176 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2177 obscure ANSI requirements
2179 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2180 * fixed bug 148: failure to inline-expand a local function left
2181 garbage, confusing the compiler.
2182 * fixed bugs 3cd: structure slot readers perform type check if the
2183 slot can have an invalid value (i.e. it is either not initialized
2184 or can be written with a less specific slot writer).
2185 * bug fix: the compiler now traps array references to elements off
2186 the end of an array; previously, the bounds checking in some
2187 circumstances could go off-by-one.
2188 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2190 * type declarations for array element types now obey the description
2191 on the CLHS page "Declaration TYPE", as per discussions on
2192 sbcl-help around 2003-05-08. This means that a declaration
2193 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2194 declaration, all references to BAR will be asserted or assumed
2195 (with THE, so dependent on compiler policy) to involve objects of
2196 type FOO. Note that no such declaration is implied in
2197 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2198 * declared types of functions from the "Conditions"
2199 chapter. (reported by Paul Dietz)
2200 * bug fix: CERROR accepts a function as its first argument.
2201 * bug fix: NTH an NTHCDR accept a bignum as index
2202 arguments. (reported by Adam Warner)
2203 * optimization: character compare routines now optimize comparing
2204 against a constant character. (reported by Gilbert Baumann)
2205 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2206 argument now works properly on the MIPS platform.
2207 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2208 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2209 declared to have a functional type, does not cause an error
2211 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2212 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2213 * fixed bug: if last continuation of a deleted block has a
2214 destination, this destination should be deleted too. (reported by
2216 * fixed a bug in the bootstrap process: the host compiler's values
2217 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2218 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2219 test case from Patrik Nordebo)
2220 * improved the ability of the disassembler on the PPC platform to
2221 provide helpful disassembly notes.
2222 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2223 the class in more cases than previously.
2224 * bug fix: FILE-POSITION now understands :START and :END for
2225 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2226 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2227 8) data. (thanks to David Lichteblau)
2228 * bug fix: it is possible to add a method to a generic function
2229 without lambda list.
2230 * bug fix: reader failed to signal END-OF-FILE inside an
2231 object representation. (reported by Nikodemus Siivola)
2232 * fixed some bugs revealed by Paul Dietz' test suite:
2233 ** LAST and [N]BUTLAST should accept a bignum.
2234 ** condition slot accessors are methods.
2235 ** (VECTOR NIL) is a subtype of STRING.
2237 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2238 * minor incompatible change: some nonsensical specialized lambda
2239 lists (used in DEFMETHOD) which were previously ignored now signal
2241 * minor incompatible change: the system is now aware of the types of
2242 variables in the COMMON-LISP package, and will signal errors for
2243 most violations of these type constraints (where previously they
2244 were silently accepted).
2245 * minor incompatible change: COMPILE-FILE now uses the freedom
2246 afforded (ANSI 3.2.2.3) to use derived function types for
2247 functions defined in the same file. This also permits the system
2248 to warn on static type mismatches and function
2249 redefinition. (Currently it does not work with high DEBUG level.)
2250 * minor incompatible change: VALUES declaration is disabled.
2251 * When issuing notes, the compiler now signals a condition of type
2252 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2253 restart for use in user handlers. It is expected that the
2254 COMPILER-NOTE condition will eventually become a condition
2255 supertype to a hierarchy of note types, which will then be
2256 handleable in a similar fashion. However, at the moment, no such
2257 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2258 but it's an implementation detail, not a classification for the
2260 * Changes in type checking closed the following bugs:
2261 ** type checking of unused values (192b, 194d, 203);
2262 ** template selection based on unsafe type assertions (192c, 236);
2263 ** type checking in branches (194bc).
2264 * A short form of VALUES type specifier has ANSI meaning (it has
2265 increased the number of situations when SBCL cannot perform type
2267 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2268 keywords or constants is permissible.
2269 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2270 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2271 classes. (thanks to Antonio Martinez)
2272 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2273 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2274 argument to a pretty-print stream. Instead, it leaves any such
2275 operations to DESCRIBE-OBJECT methods.
2276 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2278 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2279 lambda lists. (thanks to David Lichteblau)
2280 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2281 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2282 (thanks to Gerd Moellmann)
2283 * fixed bug 246: increased compilation speed of long
2284 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2286 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2288 * DEFCONSTANT now throws a condition of type
2289 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2290 constant to a non-EQL value; CONTINUE and ABORT restarts
2291 respectively change and preserve the value.
2292 * fixed bug 63: The code walker, part of the implementation of CLOS,
2293 is now better at handling symbol macros.
2294 * bug fix: There is no longer an internal implementation type named
2295 CL:LENGTH. (reported by Raymond Toy)
2296 * bug fix: In macro-like defining macros/special operators the
2297 implicit block does not enclose the lambda list.
2298 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2299 enclosed in parentheses) are not suitable as type specifiers, and
2300 their use properly signals an error now.
2301 * bug fix: An argument count mismatch for a type specifier in code
2302 being compiled no longer causes an unhandled error at compile
2303 time, but signals a compile-time warning.
2304 * fixed simple vector readable printing
2305 * bug fix: DESCRIBE takes more care over whether the class
2306 precedence list slot of a class is bound before accessing it.
2307 (reported by Markus Krummenacker)
2308 * bug fix: FORMATTER can successfully compile pretty-printer format
2309 strings which use variants of the ~* directive inside.
2310 * bug fix: SEARCH now applies its TEST predicate to the elements of
2311 the arguments in the correct order. (thanks to Wolfhard Buss)
2312 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2313 in type inference. (thanks to Robert E. Brown)
2314 * bug fix: Reading in symbols with an explicit package name of ""
2315 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2316 * fixed some bugs revealed by Paul Dietz' test suite:
2317 ** NIL is now allowed as a structure slot name.
2318 ** Arbitrary numbers, not just REALs, are allowed in certain
2319 circumstances in LOOP for-as-arithmetic clauses.
2320 ** Multiple class redefinitions before slot access no longer
2321 causes a type error.
2322 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2323 association between the name and a class.
2324 ** Generic functions with non-standard method-combination and over
2325 five methods all of which return constants no longer return NIL
2326 after the first few invocations. (thanks to Gerd Moellmann)
2327 ** CALL-NEXT-METHOD with no arguments now passes the original
2328 values of the arguments, even in the presence of assignment.
2329 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2331 ** DISASSEMBLE works with closures and funcallable instances.
2332 ** ADD-METHOD now returns the generic function, not the new method.
2333 ** FIND-METHOD signals an error if the lengths of the specializers
2334 is incompatible with the generic function, even if the ERRORP
2336 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2337 which its argument is a member.
2338 ** DEFCLASS only redefines the class named by its class-name
2339 argument if that name is the proper name of the class;
2340 otherwise, it creates a new class.
2341 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2342 of the UNBOUND-SLOT condition to the name of the slot.
2343 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2344 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2345 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2346 SLOT-MAKUNBOUND in the specified fashion.
2348 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2349 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2350 cross-compilation host. As a consequence, we can now bootstrap our
2351 way up to SBCL starting with a bare gcc toolchain and human-readable
2352 source code (first the source to CLISP, then the source to SBCL).
2353 * A contributed module containing a partial implementation of the
2354 simple-streams interface has been included. (thanks to Rudi
2356 * A contributed module implementing the RFC1321 Message Digest
2357 Algorithm, known as MD5, has been included.
2358 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2359 no longer has any effect, as the code controlled by this feature
2360 has been deleted. (As far as we know, no-one has ever built using
2361 this feature, and its semantics were confused in any case).
2362 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2363 work on conditions (as required by the ANSI specification),
2364 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2365 expected behaviour on conditions. Users should note, however,
2366 that such behaviour is not required by the ANSI specification,
2367 and so use of this behaviour may render their code unportable.
2368 * fixed some bugs revealed by Paul Dietz' test suite:
2369 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2371 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2372 ** SXHASH on condition objects no longer returns NIL.
2373 ** :ALLOCATION :CLASS slots are better treated; their values are
2374 updated on class redefinition, and initforms inherited from
2375 superclasses are applied.
2376 ** REMOVE-METHOD returns its generic function argument even when
2377 no method was removed.
2378 ** SHARED-INITIALIZE now initializes the values of the requested
2379 slots, including those with :ALLOCATION :CLASS.
2380 ** ALLOCATE-INSTANCE now works on structure classes defined via
2381 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2383 ** SLOT-EXISTS-P now works on conditions, as well as structures
2385 ** MAKE-LOAD-FORM now has the required methods on
2386 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2387 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2388 keyword, and now implements the SLOT-NAMES argument.
2389 ** methods with &OPTIONAL arguments no longer allow too many
2390 arguments to be passed in the call without error.
2391 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2392 option is consistent with the required arguments of the generic
2393 function lambda list.
2394 * bug fix: REQUIRE accepts a string designator. (Thanks to
2396 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2397 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2398 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2399 * bug fix: The compiler now checks for duplicated variables in macro
2401 * bug fix: SETQ on globals returns the correct value.
2402 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2403 not a COMPILER-ERROR (followed by some other strange error on
2404 choosing the CONTINUE restart).
2405 * bug fix: make.sh and friends are now more consistent in the way that
2406 they look for GNU "make".
2408 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2409 * experimental native threads support (on x86 Linux >=2.4 only).
2410 This is not compiled in by default: you need to add :SB-THREAD to
2411 the target features. See the "Beyond ANSI" chapter of the manual
2413 * fix for longstanding nonANSIism: The old distinction between
2414 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2415 The return value from CL:FIND-CLASS is now a CLOS class, and
2416 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2417 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2418 * An interface to the MetaObject Protocol, as described in Kiczales,
2419 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2420 MIT Press, 1991, is available from the SB-MOP package.
2421 * incompatible change: the SB-PCL package should now be considered
2422 a private implementation detail, and no longer a semi-private MOP
2424 * minor incompatible change: due to rearrangement for threads, the
2425 control stack and binding stack are now allocated at arbitrary
2426 addresses instead of being hardcoded per-port. Users affected by
2427 this probably have to be doing advanced things with shared
2428 libraries, and will know who they are.
2429 * minor incompatible change: Previously, all --eval forms used were
2430 processed with READ before any of them were processed with EVAL.
2431 Now each --eval form is processed with both READ and EVAL before
2432 the next --eval form is processed. (Thus package operations like
2433 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2434 work as the user might reasonably expect.)
2435 * minor incompatible change: *STANDARD-INPUT* is now only an
2436 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2438 * minor incompatible change: Y-OR-N-P is now character-oriented,
2439 not line oriented. Also, YES-OR-NO-P now works without errors.
2440 (thanks to Antonio Martinez)
2441 * sb-aclrepl module improvements: an integrated inspector, added
2442 repl features, and a bug fix to :trace command.
2443 * Known functions, which cannot be open coded by the backend, are
2444 considered to be able to check types of their arguments. (fixing
2445 a bug report by Nathan J. Froyd)
2446 * fixed a bug in computing method discriminating functions: It is
2447 now possible to define methods specialized on classes which have
2448 forward-referenced superclasses. (thanks to Gerd Moellmann)
2449 * fixed evaluation order in optional entries (reported by Gilbert
2451 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2452 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2453 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2454 required-by-AMOP NAME argument, as well as CLASS and
2455 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2456 * fixed bug 20: DEFMETHOD can define methods using names that are
2457 not the proper names of classes to designate class specializers.
2458 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2459 Unix file descriptors, instead of blowing up. (thanks to Antonio
2461 * Garbage collection refactoring: user-visible change is that a
2462 call to the GC function during WITHOUT-GCING will not do garbage
2463 collection until the end of the WITHOUT-GCING. If you were doing
2464 this you were probably losing anyway.
2465 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2466 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2467 TYPEP the latter but not the former.
2468 * The compiler issues a full WARNING for calls to undefined functions
2469 with names from the CL package.
2470 * MAP-INTO for a vector destination is open coded. (reported by
2471 Brian Downing on c.l.l)
2472 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2473 documentation string.
2474 * fixed some bugs revealed by Paul Dietz' test suite:
2475 ** COPY-ALIST now signals an error if its argument is a dotted
2477 ** Condition slots are now accessed more correctly in the presence
2478 of multiple initargs for a given slot.
2479 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2480 exclude restarts of the same name associated with a different
2482 ** DEFCLASS of forward-referenced classes with another
2483 forward-referenced class in the superclasses list no longer
2485 ** Condition slots are now initialized once each, not multiple
2486 times. (thanks to Gerd Moellmann)
2487 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2489 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2490 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2491 arguments contain duplicated elements.
2492 ** RESTART-CASE understands local macros.
2493 ** RESTART-CASE associates exactly its own restarts with a condition.
2494 ** ENDP in safe mode checks its argument to be of type LIST.
2495 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2496 in question is unbound.
2497 ** Optimized MAKE-INSTANCE functions no longer cause internal
2498 assertion failures in the presence of duplicate initargs.
2499 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2500 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2501 propagated even in the ctor optimized implementation of
2503 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2505 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2506 * a better implementation of SXHASH on (simple) bit vectors,
2507 measured both in execution speed and in distribution of results
2508 over the positive fixnums, has been installed. Likewise, a better
2509 implementation of EQUAL for simple bit vectors is now available.
2510 * fixed CEILING optimization for a divisor of form 2^k.
2511 * fixed bug 240 (emitting extra style warnings "using the lexical
2512 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2513 by Antonio Martinez)
2514 * fixed SXHASH, giving different results for NIL depending on type
2515 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2516 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2517 variable. (found by Rolf Wester)
2518 * SBCL does not ignore type declarations for special
2519 variables. (reported by rif on c.l.l 2003-03-05)
2520 * some bug fixes in contrib/sb-aclrepl/
2521 * fixed some bugs revealed by Paul Dietz' test suite:
2522 ** a bug in the CONS type specifier, whereby the CAR and CDR
2523 types got intertwined, has been fixed;
2524 ** the type system is now able to reason about the interaction
2525 between INTEGER and RATIO types more completely;
2526 ** APPEND, [N]REVERSE and NRECONC check that those their
2527 arguments, which must be proper lists, are really so;
2528 ** An array specialized to be unable to hold elements has been
2529 implemented, as required -- yes, really -- by ANSI;
2530 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2531 on malformed property lists;
2533 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2534 * incompatible packaging change: in line with Unix convention,
2535 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2536 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2537 * REQUIRE and PROVIDE are now optionally capable of doing something
2538 useful. See the documentation string for REQUIRE.
2539 * infrastructure for a managed SBCL contrib system: contributed
2540 modules in this release include:
2541 ** the ASDF system definition facility;
2542 ** an interface to the BSD Sockets API;
2543 ** an ACL-like convenience interface to the repl;
2544 (thanks to Kevin Rosenberg)
2545 ** an implementation of ROTATE-BYTE, with efficient implementation
2547 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2548 gives a closer approximation to the right answer than previously.
2549 (thanks to Raymond Toy)
2550 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2551 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2552 optional environment arguments, as required by ANSI.
2553 * fixed bugs in other functions taking environment objects, allowing
2554 calls with an explicit NIL environment argument to be compiled
2556 * fixed bug 228: primary return values from
2557 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2558 COMPILE or FUNCTION.
2559 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2560 :TYPE will no longer signal errors on innocuous objects.
2561 * fixed bug 231b: SETQ is better at respecting type declarations in
2562 the lexical environment.
2563 * fixed a bug in DEFCLASS: classes named by symbols with no or
2564 unprintable packages can now be defined.
2565 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2566 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2567 * fixed compiler failure related to checking types of functions.
2568 (reported by Robert E. Brown)
2569 * the compiler is now much more consistent in its error-checking
2570 treatment of bounding index arguments to sequence functions: in
2571 (SAFETY 3) code, errors will be signalled in almost all cases if
2572 invalid sequence bounding indices are passed to functions defined
2573 by ANSI to operate on sequences.
2574 * fixed a bug in the build procedure: documentation of SBCL-specific
2575 packages is now preserved and available in the final Lisp image.
2576 * lifted FDEFINITION lookup out of loops in the implementation of
2577 many list operations. (thanks to Robert E. Brown)
2578 * fixed a bug in the reader: the #n# reader macro now works for
2579 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2580 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2581 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2582 E. Brown and Tony Martinez respectively)
2583 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2584 SLOT-BOUNDP: in particular, it is now able to optimize them much
2585 better, and is now not vulnerable to having packages renamed.
2586 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2587 always called when a slot is not present in an instance. (thanks
2589 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2590 longer signal an error on unreading a character following EOF on
2591 the previous constituent stream. (thanks to Tony Martinez)
2592 * fixed some bugs revealed by Paul Dietz' test suite:
2593 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2594 not just nonnegative fixnums;
2595 ** the logical bit-array operators such as BIT-AND now accept an
2596 explicit NIL for their "opt-arg" argument (to indicate a
2597 freshly-consed result bit-array);
2598 ** ELT now signals an error on an invalid sequence index in safe
2600 ** the type system is now cleverer about negations of numeric
2601 types, and consequently understands the BIGNUM and RATIO types
2603 ** the type system is now cleverer about the interaction between
2604 INTEGER and RATIO types: while bugs still remain, many more
2605 cases are accurately computed;
2606 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2607 if it is in the last clause;
2608 ** CONSTANTLY now correctly returns a side-effect-free function in
2610 ** DECLARE is no longer treated as a special-operator; in
2611 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2612 * incremented fasl file version number due to the change in the
2613 DEFSTRUCT-SLOT-DESCRIPTION structure.
2615 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2616 * minor incompatible change: code processed by the "interpreter" or
2617 EVAL now has a compilation optimization policy of (DEBUG 2)
2618 (changed from (DEBUG 1)) to improve debuggability of interactive
2619 development, and to allow the use of the debug RETURN command in
2621 * an experimental implementation of the RETURN command for the
2622 debugger has been included. (thanks to Frederik Kuivinen)
2623 * fixed bug 62: constraints were not propagated into a loop.
2624 * fixed bug in embedded calls of SORT (reported and investigated by
2626 * fixed some bugs revealed by Paul Dietz' test suite:
2627 ** printing and reading of arrays with some dimensions having
2628 length 0 (thanks to Gerd Moellmann);
2629 ** BOA constructor with &AUX argument without a default value does
2630 not cause a type error;
2631 ** CONSTANTP now returns true for all self-evaluating objects.
2633 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2634 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2635 accessors that are related by inheritance, as specified in the
2636 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2637 Valtteri Vuorikoski)
2638 * The compiler is now able to inline functions that were defined in
2639 a complex lexical environment (e.g. inside a MACROLET).
2640 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2642 * Support for the upcoming FreeBSD-5.0 release has been included.
2643 (thanks to Dag-Erling Smorgrav)
2644 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2645 effect when it is not in a toplevel context.
2646 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2647 stand a better chance of being correct. (thanks to Gerd
2649 * fixed bug in COERCE, which now signals an error on coercing a
2650 rational to a bounded real type which excludes the expected
2652 * The compiler is now able to derive types more accurately from the
2653 COERCE and COMPILE functions.
2654 * fixed bug 223: functional binding is considered to be constant
2655 only for symbols in the CL package.
2656 * fixed bug 231: SETQ did not check the type of a variable being set
2657 (reported by Robert E. Brown)
2658 * A new optimization for MAKE-INSTANCE has been included, fixing
2659 various bugs (including relating to :ALLOCATION :CLASS slots and
2660 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2662 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2663 ** As required by ANSI, LOOP now disallows anonymous collection
2664 clauses such as COLLECT I in conjunction with aggregate boolean
2665 clauses such as THEREIS (= I 1);
2666 ** LOOP now signals an error when any variable is reused in the
2667 same loop (including the potentially useful construct analogous
2668 to WITH A = 1 WITH A = (1+ A);
2669 ** IT is only a special loop symbol within the first clause of a
2670 conditional loop clause;
2671 ** LOOP with a typed iteration variable over a hashtable now
2672 signals a type error iff it should.
2673 * fixed some other bugs revealed by Paul Dietz' test suite:
2674 ** FILE-STREAM now names the class previously known as FD-STREAM;
2675 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2676 argument) no longer signals an error;
2677 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2678 of the slot symbol, rather than using the current package
2679 ((:CONC-NAME "") continues to intern the slot's name in the
2681 * incremented fasl file version number, because of the incompatible
2682 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2683 of the new implementation of DEFINE-COMPILER-MACRO.
2685 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2686 * Support for building SBCL for MIPS platforms running in
2687 little-endian mode has now been checked in, and basic
2688 functionality on said platforms verified.
2689 * minor incompatible change: PCL now records the pathname of a file
2690 in which methods and the like are defined, rather than its
2692 * minor incompatible change: TRUENAME now considers the truename of
2693 a file naming a directory to be the pathname with :DIRECTORY
2694 component indicating that directory.
2695 * minor incompatible change: a NAMED clause in the extended form of
2696 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2697 reason for the previous behaviour is unclear.
2698 * more systematization and improvement of CLOS and MOP conformance
2699 in PCL (thanks to Gerd Moellman and Pierre Mai):
2700 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2702 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2703 primary methods with no specializers;
2704 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2706 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2707 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2708 FORWARD-REFERENCED-CLASSes; error reporting on
2709 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2711 ** SXHASH on CLOS instances now uses a slot internal to the
2712 instance to return different numbers on distinct instances,
2713 while preserving the same return value through invocations of
2715 ** DEFMETHOD signals errors when methods with longer incongruent
2716 lambda lists are added to generic functions;
2717 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2718 CLASS, as specified in AMOP;
2719 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2720 on the order returned by the primary method for classes of
2721 class STANDARD-CLASS;
2722 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2723 * fixed some bugs shown by Paul Dietz' test suite:
2724 ** DOLIST puts its body in TAGBODY;
2725 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2727 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2728 value producing form;
2729 ** if more variables are given to PROGV than values, extra
2730 variables are bound and made to have no value;
2731 ** NSUBSTITUTE on list arguments gets the right answer with
2733 ** ELT signals an error of type TYPE-ERROR when the index argument
2734 is not a valid sequence index;
2735 ** LOOP signals (at macroexpansion time) an error of type
2736 PROGRAM-ERROR when duplicate variable names are found;
2737 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2738 ** FUNCALL of special-operators now cause an error of type
2740 ** PSETQ now works as required in the presence of side-effecting
2741 symbol-macro places;
2742 ** NCONC accepts any object as its last argument;
2743 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2745 ** loop-for-as-package does not require a package to be explicitely
2747 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2749 * fixed bug 166: compiler preserves "there is a way to go"
2750 invariant when deleting code.
2751 * fixed bug 172: macro lambda lists with required arguments after
2752 &REST arguments now cause an error to be signalled. (thanks to
2754 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2755 bug. (thanks to Matthew Danish)
2756 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2758 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2759 arguments are lexically rebound. (thanks to Gerd Moellmann and
2761 * fixed bug 194: error messages are now more informative when there
2762 is no primary method applicable in a call to a generic
2763 function. (thanks to Gerd Moellmann)
2764 * fixed bug in command line argument checking (thanks to Julian
2766 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2767 (thanks to Matthew Danish)
2768 * incremented fasl file version number, because of the
2769 SXHASH-related changes in the layout of CLOS data structures
2771 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2772 * minor incompatible change: The runtime (the Unix executable named
2773 "sbcl") is now much pickier about the .core files it will load.
2774 Essentially it now requires .core files to descend from the same
2775 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2776 as the runtime does. (The intent is to prevent the crashes which
2777 can occur, and which can even be reported as mysterious failures,
2778 when people patch the sources or change the build parameters
2779 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2780 sbcl and .core files.)
2781 * fixed bug: VALUES-LIST is no longer optimized away.
2782 * fixed bug 142: The FFI conversion of C string values to Lisp
2783 string values no longer conses excessively. (thanks to Nathan
2784 Froyd porting Raymond Toy's fix to CMU CL)
2785 * began to systematize and improve MOP conformance in PCL (thanks to
2786 Nathan Froyd, Gerd Moellman and Pierre Mai):
2787 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2789 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2790 ** FINALIZE-INHERITANCE is now called on class finalization;
2791 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2792 argument precedence order.
2793 * fixed bug 202: The compiler no longer fails on functions whose
2794 derived types contradict their declared type.
2795 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2796 so it can be non-toplevel.
2797 * The fasl file version number has changed (because of the new
2798 implementation of DEFMACRO).
2799 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2800 safe code, that any length requirement by their type-specifier
2801 argument is valid. The exceptions to this are described in bug
2803 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2804 functions now check the types of their inputs as required by ANSI.
2805 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2806 introduced symbol is DECLAREd to be SPECIAL.
2807 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2808 * fixed bug: :COUNT argument to sequence functions may be negative
2809 * fixed bug: body of DO-SYMBOLS may contain declarations
2810 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2811 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2812 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2813 symbol macro only once
2814 * fixed printing of call frame when argument list is unavailable
2815 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2816 * compiler no longer signals WARNING on unknown keyword
2819 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2820 * A beta-quality port to the mips architecture running Linux,
2821 based on the old CMUCL backend, has been made. It has been tested
2822 on a big-endian kernel, and works sufficiently well to be able to
2823 rebuild itself; it has not been tested in little-endian mode.
2824 * fixed an inconsistency between gencgc.c and purify.c which made
2825 dumping/loading .core files unreliable
2826 * fixed bug 120a: The compiler now deals correctly with IFs where
2827 the consequent is the same as the alternative, instead of
2828 misderiving the return type. (thanks to Alexey Dejneka)
2829 * fixed bug 113: Logical pathnames are now dumpable (the logical
2830 host is resolved at load-time, throwing an error if it is not
2832 * fixed bug 174: FORMAT's error message is slightly clearer when a
2833 non-printing character is used in a format directive.
2834 * fixed several bugs in compiler checking of type declarations, i.e.
2835 violations of the Python "declarations are assertions" principle
2836 (thanks to Alexey Dejneka)
2837 * fixed several bugs in PCL's error checking (thanks to Gerd
2839 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2841 * fixed bug in compilation of functions as first class values
2842 (thanks to Antonio Martinez-Shotton)
2843 * The compiler's handling TYPE-ERRORs which it can prove will
2844 inevitably happen at runtime has been cleaned up and corrected
2845 in several ways. (thanks to Alexey Dejneka)
2846 * improved argument type checking for various basic arithmetic
2847 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2848 lavished on them in the past that they can be compiled in many
2849 ways in different special cases
2850 * fixed bug 181: compiler checks validity of user supplied type
2852 * cleaned up code flushing in optimization: Function calls which
2853 should signal errors for safety purposes (e.g. which ANSI says
2854 should signal errors when their arguments are of incorrect type)
2855 are no longer optimized away.
2856 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2857 * incremented fasl file version number, because changes in the
2858 implementation of sequence functions like COERCE caused
2859 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2860 in old inline expansions) to become undefined. (Actually these
2861 changes were later undone, so we might very well be binary
2862 compatible with 0.7.7 after all, but leaving the version number
2863 incremented seemed like the simplest and most conservative
2866 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2867 * An alpha-quality port to the parisc architecture running Linux,
2868 based on the old CMUCL backend, has been made. This, even more so
2869 than the other backends, should be considered still a work in
2870 progress; known problems include that the Linux kernel in 64-bit
2871 mode does not propagate the correct sigcontext structure to
2872 userspace, and consequently SBCL on a parisc64 kernel will not
2874 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2875 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2876 suppressed.) Also now that the compiler is looking at declarations
2877 in the environment, it checks optimization declarations as well,
2878 and suppresses inlining when (> DEBUG SPEED).
2879 * More fixes have been made to treatment of floating point exception
2880 treatment and other Unix signals. In particular, floating point
2881 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2882 * The detection and handling of control stack exhaustion (infinite
2883 or very deeply nested recursion) has changed. Stack exhaustion
2884 detection is now done by write-protecting pages at the OS level
2885 and applies at all optimization settings; when found, a
2886 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2887 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2888 be caught using IGNORE-ERRORS.
2889 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2890 correctly when transforming recursive LABELS functions to LETs.
2891 (thanks to Alexey Dejneka)
2892 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2893 that are names of constants or global variables.
2894 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2895 alien routines with docstrings.
2896 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2897 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2899 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2900 (thanks to Eric Marsden)
2901 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2902 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2903 * Bug fix: The SPARC backend can now compile functions involving
2904 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2905 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2906 to a functions expecting a PATHNAME-DESIGNATOR.
2907 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2908 lambda lists. (thanks to Alexey Dejneka)
2909 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2910 functions so that EXPORTing the name of the class doesn't cause
2911 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2912 to Antonio Martinez for reporting this)
2913 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2914 constant-folded without failing an assertion. (thanks to Einar
2915 Floystad Dorum for reporting this)
2916 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2917 types are now dealt with more correctly by the compiler.
2918 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2919 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2920 behaviour analogous to RENAME-FILE. This puts its behaviour more
2921 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2923 * The fasl file version number has changed again. (because of the
2924 bug fix involving the names of PCL MAKE-INSTANCE functions)
2926 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2927 * bug fix: Floating point exceptions are treated much more
2928 consistently on the x86/Linux and PPC/Linux platforms.
2929 * Array initialization with :INITIAL-ELEMENT is now much faster for
2930 cases when the compiler cannot open code the array creation, but
2931 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2932 array accesses have also seen a speed increase.
2933 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2934 specified. (This was at the root of some bad interactions between
2935 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2937 * bug fix: Internal error arguments for undefined functions are now
2938 computed correctly on the PPC/Linux platform.
2939 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2940 Raymond Toy's patch for CMU CL)
2941 * Support for the Solaris 9 operating environment has been included
2942 (thanks to Daniel Merritt)
2943 * A very ugly but hopefully complete draft of the missing FFI chapter
2944 of the manual has been created by reformatting the corresponding
2945 CMU CL manual chapter into (currently very ugly and incoherent)
2946 DocBook and bringing it up to date for SBCL behavior. Thus, the
2947 manual is now essentially complete, at least by my extreme
2948 once-and-only-once standards, whereby it's acceptable to refer to
2949 the doc strings of SB-EXT functions as the primary documentation.
2950 * The fasl file version number has changed again, due to cleanup of
2951 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2952 is no longer a static symbol.)
2954 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2955 * SBCL now builds with OpenMCL (version 0.12) as the
2956 cross-compilation host; also, more progress has been made toward
2957 bootstrapping under CLISP.
2958 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2960 * bug 158 fixed: The compiler can now deal with integer loop
2961 increments different from 1; fixing this turned out also to fix
2963 * bug 169 fixed: no more bogus warnings about using lexical bindings
2964 despite the presence of perfectly good SPECIAL declarations (thanks
2965 to David Lichteblau)
2966 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2967 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2968 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2970 * bug 180 fixed: Method combination specifications no longer ignore
2971 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2972 * bug fix: Structure type predicate functions now check their argument
2973 count as they should.
2974 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2975 correctly. (thanks to Pierre Mai)
2976 * minor incompatible change: The --noprogrammer option is deprecated
2977 in favor of the new --disable-debugger option, which is very similar.
2978 (The major difference is that it takes effect at a slightly different
2979 time at startup, causing handling of errors in --sysinit and
2980 --userinit files will be affected differently.) The
2981 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2982 been added to allow this functionality to be controlled from ordinary
2983 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2984 maintainers, who might want to run non-interactive scripts to
2985 build SBCL cores which will later be used interactively.)
2986 * minor incompatible change: The LOAD function no longer, when given
2987 a wild pathname to load, loads all files matching that pathname.
2988 Instead, an error of type FILE-ERROR is signalled.
2990 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2991 * bug 147 fixed: The compiler preserves its block link/count
2992 invariants more correctly now so that it doesn't crash. (thanks
2994 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2996 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2997 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2998 no-value-for-this-character as NIL (as specified by ANSI).
2999 * HOST-NAMESTRING on physical pathnames now returns a string that is
3000 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3001 (thanks to Christophe Rhodes)
3002 * The Alpha port handles icache flushing more correctly. (thanks to
3004 * More progress has been made toward bootstrapping under CLISP. (thanks
3005 to Christophe Rhodes)
3006 * The fasl file format has changed again, because dynamic loading
3007 on OpenBSD (which has non-ELF object files) motivated some cleanups
3008 in the way that foreign symbols are transformed and passed around.
3009 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3010 is no longer treated as whitespace by the reader, but instead as
3011 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3012 |A
\7fB|, instead of A as it used to.
3014 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3015 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3016 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3017 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3018 this as of 0.7.1.45, but was left out of the previous news section
3019 (thanks to Dan Barlow)
3020 * SBCL now runs on the Solaris operating system on SPARC architectures
3021 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3022 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3023 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3024 * SPARC backend cleanups, allowing builds of cores optimized for V8
3025 and V9 SPARCS, and also emission of code targeted to a particular
3026 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3028 * SBCL is closer to bootstrapping under CLISP, thanks to various
3029 fixes by Christophe Rhodes.
3030 * The fasl file format has changed again, to allow the compiler's
3031 INFO database to support symbol macros.
3032 * The user manual (in doc/) is formatted into HTML more nicely.
3033 (thanks to coreythomas)
3034 * The system is smarter about SUBTYPEP relationships, especially
3035 those involving NOT types (including types such as ATOM which are
3036 represented internally using NOT types). Thus SUBTYPEP is less
3037 likely to return (VALUES NIL NIL) in general, and in particular
3038 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3040 * The fasl file format has changed again, because the internal
3041 representation of types now includes a new slot to support the new
3042 SUBTYPEP-of-NOT-types logic.
3043 * (not a change in the main branch of SBCL, but a related prototype
3044 which can hopefully be merged into the main branch of SBCL in the
3045 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3046 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3047 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3048 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3049 dispatch macro character combinations. (thanks to Alexey Dejneka)
3050 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3051 that superficially look like logical namestrings correctly.
3052 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3054 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3055 default supplied-p) for &optional and &key arguments. (thanks to
3058 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3059 * incompatible change: The compiler is now less aggressive about
3060 tail call optimization, doing it only when (> SPACE DEBUG) or
3061 (> SPEED DEBUG). (This is an incompatible change because there are
3062 programs which relied on the old CMU-CL-style behavior to optimize
3063 away their unbounded recursion which will now die of stack overflow.)
3064 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3065 for non-GENCGC systems has been increased to 20M (since that
3066 seems much closer to the likely performance optimum for modern
3067 systems than the old 4M value was)
3068 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3069 and *DEBUG-PRINT-LEVEL*
3070 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3071 of CMU CL's support for SPARC, and various endianness and other
3072 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3073 * new syntactic sugar for the Unix command line: --load foo.bar is now
3074 an alternate notation for --eval '(load "foo.bar")'.
3076 ** The system now detects stack overflow and handles it gracefully,
3077 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3078 optimization settings. (This is a good thing in general, and
3079 its introduction in this version should be particularly timely
3080 for anyone whose code fails because of suppression of tail
3082 ** The system now hunts for the C variable "environ" in a more
3083 devious way, to avoid segfaults when the C library version differs
3084 between compile time and run time. (thanks to Christophe Rhodes)
3085 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3086 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3087 ** The compiler no longer issues bogus style warnings for undefined
3088 classes in the same source file as the DEFCLASSes which defined
3089 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3091 ** fixes in CONDITION class precedence list for undefined function
3092 errors (thanks to Alexei Dejneka)
3093 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3094 correctly. (thanks to Dan Barlow)
3095 ** portability fixes aiming at bootstrapping under CLISP (thanks
3096 to Dave McDonald and Christophe Rhodes)
3097 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3098 ** fixes in type translation and and type inference (thanks to
3100 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3101 ** various fixes in the new ports (thanks to Dan Barlow)
3102 * several changes related to debugging:
3103 ** suppression of tail recursion, as noted above
3104 ** stack overflow detection, as noted above
3105 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3106 is now the default. (For some time encapsulation has been more
3107 reliable than the breakpoint-based :ENCAPSULATE NIL
3108 implementation, at least on X86 systems; and I just noticed that
3109 encapsulation also seems closer to the spirit of the ANSI
3112 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3114 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3115 up properly again. (There was a packaging bug in 0.7.0 which
3116 left their definitions in SB-SYS::LOAD-FOREIGN and
3117 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3118 vital for most things which interface to C-level interfaces,
3119 like extensions working with sockets or databases or
3120 Perl-compatible regexes or whatever, and the need to fix
3121 this bug is the main reason that 0.7.1 was released so
3123 ** DEFGENERIC is now choosier about the methods it redefines, so that
3124 reLOADing a previously-LOADed file containing DEFGENERICs does
3125 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3126 cycle now works as it should. (thanks to Alexey Dejneka)
3127 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3128 fixed by Christophe Rhodes. (He actually submitted this patch
3129 months ago, and I delayed until after 0.7.0.)
3130 ** Bug 111 (internal compiler confusion about runtime checks on
3131 FUNCTION types) was fixed by Alexey Dejneka.
3132 * Some internal cleanups (getting rid of variables which aren't
3133 needed now that the byte interpreter is gone) caused the fasl
3134 file format number to change again.
3136 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3137 * major incompatible change: The default fasl file extension, i.e. the
3138 default extension for files produced by COMPILE-FILE, has changed
3139 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3141 ** There are many changes in the implementation of the compiler.
3142 SBCL is now essentially a compiler-only implementation of ANSI
3143 Common Lisp. EVAL still "interprets" a few special cases, but
3144 almost all the interesting cases are handled by creating
3145 a LAMBDA expression, calling COMPILE on it, then calling
3146 FUNCALL on the result.
3147 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3148 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3149 Since the code is newer, there might still be some new bugs
3150 (though not as many as before Martin Atzmueller's fixes:-). But
3151 the new code is substantially simpler and clearer, and hopefully
3152 any remaining bugs will be simpler, less fundamental, and more
3153 fixable then the bugs in the old code.
3154 ** The revised compiler is still a little unsteady on its feet.
3156 *** The debugging information it produces (particularly the names
3157 of FUNCTION objects) is sometimes much less useful than what
3158 the old compiler produced.
3159 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3160 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3161 within a MACROLET) has been temporarily weakened.
3162 ** There are new compiler optimizations for various functions:
3163 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3164 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3165 *** the math functions TRUNCATE, FLOOR, and CEILING
3166 *** the function-of-all-trades COERCE
3167 Mostly these should be transparent, but there's one
3168 potentially-annoying problem (bug 117): when the compiler
3169 inline-expands a function and does type analysis on the result,
3170 it can create control paths which have type mismatches, and
3171 when it can't prove that those control paths aren't taken,
3172 it will issue WARNINGs about the type mismatches. This is
3173 a particular problem in practice for the new sequence functions.
3174 It's not clear how this should be fixed, and for now, a
3175 workaround is given in the entry for 117 in the BUGS file.
3176 ** (Because of the interaction between the two previous items --
3177 occasional inlining problems and new inline expansions -- some
3178 of the new sequence function optimizations won't really kick in
3179 completely until debugging information, and then inlining, are
3180 straightened out in some future version.)
3181 * minor incompatible changes:
3182 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3183 DIRECTORY no longer implicitly promotes NIL slots of its
3184 pathname argument to :WILD. In particular, when you ask for the
3185 contents of a directory (which you used to be able to do without
3186 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3187 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3188 ** changes in behavior that ANSI explicitly defines to be
3189 implementation dependent:
3190 *** The new compiler-only implementation still conforms with ANSI,
3191 but acts a little different than before. Besides the obvious
3192 changes in performance tradeoffs (that the cost per form passed
3193 to EVAL has gone up, and the cost per form executed by EVAL
3194 has gone down), the behavior of the system changes a little
3195 because there are no longer any interpreted function objects.
3196 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3197 e.g. doing COMPILE on the output of interactive DEFUN is
3199 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3201 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3202 has changed from (:CL) to NIL.
3203 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3204 e.g. #\Newline and #\Tab, don't have names specified in the
3205 ANSI Common Lisp standard, is now based on their ASCII symbolic
3206 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3207 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3208 are no longer used for output.
3209 ** changes in internal implementation constants:
3210 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3211 4 million. (If your application spends a lot of time GCing and
3212 you have a lot of RAM, you might want to experiment with
3213 increasing it even more.)
3214 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3215 However, almost all old code should still continue to work without
3216 immediate update, as SB-C-CALL is now a (deprecated) nickname
3218 ** Old operator names in the style DEF-FOO are now deprecated in
3219 favor of new corresponding names DEFINE-FOO, for consistency with
3220 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3221 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3222 internal symbols, but a few supported extensions like
3223 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3224 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3225 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3226 etc. as you get deeper into recursive calls to the debugger
3227 command loop, instead of the old "5]", "5]]", "5]]]"
3228 sequence. (I was motivated to do this when squabbles between
3229 ILISP and SBCL left me very deeply nested in the debugger. In the
3230 short term, this change will probably provoke more ILISP/SBCL
3231 squabbles, but hopefully it will be an improvement in the long run.)
3232 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3233 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3234 function, since declaiming return types involving aliens is
3235 (1) annoyingly messy to do by hand and (2) vital to efficient
3236 compilation of code which calls such functions.
3237 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3238 longer reexported by the SB-EXT package. They're solely useful
3239 for alien code, so it seems more logical that you should get
3240 them from the SB-ALIEN package, not in SB-EXT.
3241 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3242 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3243 features. Instead, the code that they used to control is always
3244 built into the system.
3245 * many other bug fixes
3246 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3247 advantage of the new EVAL-WHEN stuff and to clean them up in
3248 general, and they are now more ANSI-compliant in a number of
3249 ways. Martin Atzmueller is responsible for a lot of this.
3250 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3252 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3253 *** correct ERROR type for various file operations
3254 *** some fixes for Lisp streams
3255 *** DEFMETHOD syntax checking
3256 *** changing old weird representation of debug information as
3257 strings (which, among their other deficiencies, don't transform
3258 correctly when you rename packages, and don't change their
3259 print representation when you change things like *PACKAGE*
3260 and *PRINT-LENGTH*) to symbols and lists of symbols
3261 He also made several improvements and fixed several bugs in DESCRIBE.
3262 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3264 *** misbehavior of WRITE-STRING/WRITE-LINE
3265 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3266 and several other LOOP problems as well
3267 *** DIRECTORY when similar filenames are present
3268 *** DEFGENERIC with :METHOD options
3269 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3270 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3271 *** argument ordering in FIND with :TEST option
3272 *** mishandled package designator argument in APROPOS-LIST
3273 *** various problems in the backquote readmacro
3274 *** a bug in APROPOS
3275 *** probably some others that I'm not describing very well here,
3276 since the CVS log documents them by reference to sbcl-devel
3277 messages, and the SourceForge archives aren't working well.:-(
3278 ** Dan Barlow improved the Alpha port (and is making progress on the
3279 PPC port, for those of you who think different).
3280 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3281 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3282 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3283 continues to work on the SPARC port (for those of you in a position
3284 to look down upon our little PC-compatible boxes from a great height).
3285 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3286 and uses that copy, rather than the current dynamic value, when
3287 it's trying to decide whether to truncate output. Thus e.g.
3288 (let ((*print-lines* 50))
3289 (pprint-logical-block (stream nil)
3291 (let ((*print-lines* 8))
3292 (print (aref possiblybigthings i) stream)))))
3293 should now truncate the logical block only at 50 lines, instead of
3294 often truncating it at 8 lines, as it did before.
3295 * The doc/cmucl/ directory, containing old CMU CL documentation
3296 from the time of the fork, is no longer part of the base system.
3297 SourceForge has shut down its anonymous FTP service, and with it
3298 my original plan for distributing the old CMU CL documentation
3299 there. For now, if you need these files you can download an old
3300 SBCL source release and extract them from it.
3301 * The fasl file version number changed again, for dozens of reasons,
3302 some of which are apparent above.
3304 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3305 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3306 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3307 patch, so that DISASSEMBLE output is much nicer.
3308 * The code in the SB-PROFILE package now seems reasonably stable.
3309 I still haven't decided what the final interface should look like
3310 (I'd like PROFILE to interact cleanly with TRACE, since both
3311 facilities use function encapsulation) but if you have a need
3312 for profiling now, you can probably use it successfully with
3313 the current CMU-CL-style interface.
3314 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3315 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3316 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3317 link now returns the dangling link itself, and for similar
3318 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3319 link itself. (In these cases the old code signalled an error and
3320 looped endlessly, respectively.) Thus, DIRECTORY now works even
3321 in the presence of dangling and cyclic symbolic links.
3322 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3323 is now a supported extension again, since the consensus on
3324 sbcl-devel was that it can be useful for ordinary development
3325 work, not just for debugging SBCL itself.
3326 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3327 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3328 that currently-defined functions might be redefined later with
3329 different return types.
3330 * Hash tables can be printed readably, as inspired by CMU CL code
3331 of Eric Marsden and SBCL code of Martin Atzmueller.
3332 * better error handling in CLOS method combination, thanks to
3333 Martin Atzmueller porting Pierre Mai's CMU CL patches
3334 * more overflow fixes for >16Mbyte I/O buffers
3335 * A bug in READ has been fixed, so that now a single Ctrl-D
3336 character suffices to cause end-of-file on character streams.
3337 In particular, now you only need one Ctrl-D at the command
3338 line (not two) to exit SBCL.
3339 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3341 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3342 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3343 does the right thing.
3344 * The implementation of some type tests, especially for CONDITION
3345 types, is now tidier and maybe faster, due to CMU CL code
3346 originally by Douglas Crosher, ported by Martin Atzmueller.
3347 * Some math functions have been fixed, and there are new
3348 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3349 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3350 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3351 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3352 likely to be built into sbcl-0.7.0.)
3353 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3354 CL, ported to SBCL by Martin Atzmueller)
3355 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3356 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3357 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3358 * There's a new slam.sh hack to shorten the edit/compile/debug
3359 cycle for low-level changes to SBCL itself, and a new
3360 :SB-AFTER-XC-CORE target feature to control the generation of
3361 the after-xc.core file needed by slam.sh.
3362 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3363 COMPILE-FILE is no longer supported, so that now every function
3364 gets an entry point, so that block compilation looks a little
3365 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3366 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3367 returns the number of bytes consed since the system started,
3368 rather than the number consed since the first time the function
3369 was called. (The new definition parallels ANSI functions like
3370 CL:GET-INTERNAL-RUN-TIME.)
3371 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3372 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3373 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3374 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3375 * Fasl file version numbers are now independent of the target CPU,
3376 since historically most system changes which required version
3377 number changes have affected all CPUs equally. Similarly,
3378 the byte fasl file version is now equal to the ordinary
3381 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3382 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3383 is no longer recognized. I apologize for this, because it was
3384 listed in SB-EXT as a supported extension, but I found that
3385 its existing behavior was poorly specified, as well as incorrectly
3386 specified, and it looked like too much of a mess to straighten it
3387 out. I have enough on my hands trying to get ANSI stuff to work..
3388 * many patches ported from CMU CL by Martin Atzmueller, with
3389 half a dozen bug fixes in pretty-printing and the debugger, and
3390 half a dozen others elsewhere
3391 * fixed bug 13: Floating point infinities are now supported again.
3392 They might still be a little bit flaky, but thanks to bug reports
3393 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3394 as flaky as they were.
3395 * The --noprogrammer command line option is now supported. (Its
3396 behavior is slightly different in detail from what the old man
3397 page claimed it would do, but it's still appropriate under the
3398 same circumstances that the man page talks about.)
3399 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3400 are now supported, and enabled by default. Thus, the compiler can
3401 handle many floating point and complex operations much less
3402 inefficiently. (Thus e.g. you can implement a complex FFT
3404 * The compiler now detects type mismatches between DECLAIM FTYPE
3405 and DEFUN better, and implements CHECK-TYPE more correctly, and
3406 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3407 thanks to patches from Martin Atzmueller.
3408 * various fixes to make the cross-compiler more portable to
3409 ANSI-conforming-but-different cross-compilation hosts (notably
3410 Lispworks for Windows, following bug reports from Arthur Lemmens)
3411 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3412 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3413 fixed thanks to Pierre Mai's CMU CL patches.
3414 * fixes to keep the system from overflowing internal counters when
3415 it tries to use i/o buffers larger than 16M bytes
3416 * fixed bug 45a: Various internal functions required to support
3417 complex special functions have been merged from CMU CL sources.
3418 (When I was first setting up SBCL, I misunderstood a compile-time
3419 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3420 * improved support for type intersection and union, fixing bug 12
3421 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3422 more obscure bugs as well
3423 * some steps toward byte-compiling non-performance-critical
3424 parts of the system, courtesy of patches from Martin Atzmueller
3425 * Christophe Rhodes has made some debian packages of sbcl at
3426 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3427 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3428 stable, but are nonetheless usable. When he's ready, I'd be happy
3429 to add them to the SourceForge "File Releases" section. (And if
3430 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3431 * new fasl file format version number (because of changes in
3432 internal representation of (OR ..) types to accommodate the new
3433 support for (AND ..) types, among other things)
3435 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3436 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3438 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3440 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3441 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3442 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3443 * DESCRIBE now works on structure objects again.
3444 * Most function call argument type mismatches are now handled as
3445 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3446 know whether the function will be redefined before the call is
3447 executed. (The compiler could flag local calls with full WARNINGs,
3448 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3449 it doesn't keep track of enough information to know whether calls
3450 are local in this sense.)
3451 * Compiler output is now more verbose, with messages truncated
3452 later than before. (There should be some supported way for users
3453 to override the default verbosity, but I haven't decided how to
3454 provide it yet, so this behavior is still controlled by the internal
3455 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3456 src/compiler/ir1util.lisp.)
3457 * Fasl file format version numbers have increased again, because
3458 support for the Gray streams extension changes the layout of the
3459 system's STREAM objects.
3460 * The Gray subclassable streams extension now works, thanks to a
3461 patch from Martin Atzmueller.
3462 * The full LOAD-FOREIGN extension (not just the primitive
3463 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3464 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3465 but like most other programs, it defaults to copying the Unix
3466 environment from the original process instead of starting the
3467 new process in an empty environment.
3468 * Extensions which manipulate the Unix environment now support
3469 an :ENVIRONMENT keyword option which doesn't smash case or
3470 do other bad things. The CMU-CL-style :ENV option is retained
3471 for porting convenience.
3472 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3473 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3475 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3477 * A patch from Martin Atzmueller seems to have solved the SIGINT
3478 problem, and as far as we know, signal-handling now works cleanly.
3479 (If you find any new bugs, please report them!)
3480 * The system no longer defaults Lisp source file names to types
3481 ".l", ".cl", or ".lsp", but only to ".lisp".
3482 * The compiler no longer uses special default file extensions for
3483 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3484 seems to expect a single default extension for all compiled code,
3485 and there's no compelling reason to try to stretch the standard
3486 to allow two different extensions.) Instead, byte-compiled files
3487 default to the same extension as native-compiled files.
3488 * Fasl file format version numbers have increased again, because
3489 a rearrangement of internal implementation packages made some
3490 dumped symbols in old fasl files unreadable in new cores.
3491 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3492 many fewer weird special cases.
3493 * Bug #17 (differing COMPILE-FILE behavior between logical and
3494 physical pathnames) has been fixed, and some related misbehavior too,
3495 thanks to a patch from Martin Atzmueller.
3496 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3497 by Tim Moore, ported to SBCL by Martin Atzmueller.
3498 * Martin Atzmueller fixed several filesystem-related problems,
3499 including bug #36, in part by porting CMU CL patches, which were
3500 written in part by Paul Werkowski.
3501 * More compiler warnings in src/runtime/ are gone, thanks to
3502 more patches from Martin Atzmueller.
3503 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3506 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3508 * DESCRIBE now works on CONDITION objects.
3509 * The debugger now handles errors which arise when trying to print
3510 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3512 * The build system now uses an additional file, customize-target-features.lisp,
3513 to allow local modifications to the target *FEATURES* list. (The point of
3514 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3515 debugging features enabled, without having to worry about propagating it
3516 into everyone's system when I do a "cvs update".) When no
3517 customize-target-features.lisp file exists, the target *FEATURES* list
3518 should be constructed the same way as before.
3519 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3520 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3521 warn about behavior which is undefined under the ANSI spec. Note: This
3522 is specified by ANSI, but it's not very popular with programmers.
3523 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3524 macro in the SBCL sources for an example of a workaround which you
3525 might use to make portable ANSI-standard code which does what you want.
3526 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3527 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3528 when it's used as a non-toplevel form. (This is required in order
3529 to implement the DEFCONSTANT-EQX macro.)
3530 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3531 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3532 funny things, due to bugs in the implementation of EVAL-WHEN
3533 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3534 non-toplevel DEFCONSTANTs are uncommon.)
3535 * The core file version number and fasl file version number have been
3536 incremented, because the old noncompliant DEFCONSTANT behavior involved
3537 calling functions which no longer exist, and because I also took the
3538 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3539 * fixed bug 1 (error handling before read-eval-print loop starts), and
3540 redid debugger restarts and related debugger commands somewhat while
3542 ** The QUIT debugger command is gone, since it did something
3543 rather different than the SB-EXT:QUIT command, and since it never
3544 worked properly outside the main toplevel read/eval/print loop.
3545 Invoking the new TOPLEVEL restart provides the same functionality.
3546 ** The GO debugger command is also gone, since you can just invoke
3547 the CONTINUE restart directly instead.
3548 ** The TOP debugger command is also gone, since it's redundant with the
3549 FRAME 0 command, and since it interfered with abbreviations for the
3551 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3553 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3554 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3555 some porting work by Martin Atzmueller)
3556 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3557 with a boatload of patches which he ported to SBCL. Now that those
3559 ** The system tries to make sure that its low-priority messages
3560 are prefixed by semicolons, to help people who like to use
3561 syntax highlighting in their ILISP buffer. (This patch
3562 was originally due to Raymond Toy.)
3563 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3564 patches originally written by Raymond Toy.
3565 ** The compiler understands coercion between single-value and
3566 multiple-VALUES type expressions better, getting rid of some very
3567 weird behavior, thanks to patches originally by Robert MacLachlan
3568 and Douglas Crosher.
3569 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3570 lambda lists, thanks to a patch originally by Pierre Mai.
3571 ** The system no longer bogusly warns about "abbreviated type
3573 ** The compiler gets less confused by inlining and RETURN-FROM,
3574 thanks to some patches originally by Tim Moore.
3575 ** The system no longer hangs when dumping circular lists to fasl
3576 files, thanks to a patch originally from Douglas Crosher.
3577 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3578 undefined function error.
3579 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3580 in structure slot initforms) for now, documented workaround instead:-|
3581 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3582 * fixed bug 5: added stubs for various Gray stream functions called
3583 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3584 installed, at least appropriate type errors are generated
3585 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3586 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3587 consistently in DEFMETHOD forms.
3588 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3589 it doesn't seem to affect SBCL after all
3590 * The C runtime system now builds with better optimization and many
3591 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3593 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3595 * The system is now under CVS at SourceForge (instead of the
3596 CVS repository on my home machine).
3597 * The new signal handling code has been tweaked to treat register
3598 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3599 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3600 downstream, e.g. in debug-int.lisp, has implicit dependencies
3601 on the unsignedness of integer representation of machine words,
3602 and that caused the system to bomb out with infinite regress
3603 when trying to recover from type errors involving signed values,
3604 e.g. (BUTLAST '(1 2 3) -1).)
3605 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3606 (This was one of the bugs Peter Van Eynde reported back in July.)
3607 * The system now uses code inspired by Colin Walters' O(N)
3608 implementation of MAP (from the cmucl-imp@cons.org mailing
3609 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3610 inline the MAP operation, and there is more than one
3611 sequence argument to the MAP call (so that it can't just
3612 do ETYPECASE once and for all based on the type of the
3613 single sequence argument). (The old non-inline implementation
3614 of the general M-argument sequence-of-length-N case required
3615 O(M*N*N) time when any of the sequence arguments were LISTs.)
3616 * The QUIT :UNIX-CODE keyword argument has been renamed to
3617 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3618 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3620 * Raymond Wiker's patches to port dynamic loading from Linux to
3621 FreeBSD have been added.
3622 * The BUGS file is now more nearly up to date, thanks in large part
3623 to Martin Atzmueller's review of it.
3624 * The debugger now flushes standard output streams before it begins
3625 its output ("debugger invoked" and so forth).
3626 * The core version number and fasl file version number have both
3627 been incremented, because of incompatible changes in the layout
3629 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3630 used to not be called for a saved Lisp image.)
3631 * Martin Atzmueller's version of a patch to fix a compiler crash,
3632 as posted on sbcl-devel 13 September 2000, has been installed.
3633 * Instead of installing Martin Atzmueller's patch for the
3634 compiler transform for SUBSEQ, I deleted the compiler transform,
3635 and transforms for some similar consing operations.
3636 * A bug in signal handling which kept TRACE from working on OpenBSD
3638 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3639 away by constant folding
3640 * The system now defines its address space constants in one place
3641 (in the Lisp sources), and propagates them automatically elsewhere
3642 (through GENESIS and the sbcl.h file). Therefore, patching the
3643 address map is less unnecessarily tedious and error-prone. The
3644 Lisp names of address space constants have also been systematized.
3645 * CVS tags like dollar-Header-dollar have been removed from
3646 the sources, because they have never saved me trouble and
3647 they've been source of trouble working with patches and other
3648 diff-related operations.
3649 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3650 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3652 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3654 * The system has been ported to OpenBSD.
3655 * The system now compiles with a simple "sh make.sh" on the systems
3656 that it's supported on. I.e., now you no longer need to tweak
3657 text in the target-features.lisp-expr and symlinks in src/runtime/
3658 by hand, the make.sh takes care of it for you.
3659 * The system is no longer so grossly inefficient when compiling code
3660 involving vectors implemented as general (not simple) vectors (VECTOR T),
3661 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3662 which dares to use the various sequence functions on non-simple
3663 vectors, takes less of a performance hit.
3664 * There is now a primitive type predicate VECTOR-T-P
3665 to test for the (VECTOR T) type, so that e.g.
3666 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3667 can now be compiled with some semblance of efficiency. (The old code
3668 turned the type declaration into a full call to %TYPEP at runtime!)
3669 * AREF on (VECTOR T) is still not fast, since it's still compiled
3670 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3671 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3672 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3673 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3674 without making full calls to SUBTYPEP at runtime.
3675 (Some analogous efficiency issues for non-simple vectors specialized to
3676 element types other than T, or for non-simple multidimensional arrays,
3677 have not been addressed. They could almost certainly be handled the
3678 same way if anyone is motivated to do so.)
3679 * The changes in array handling break binary compatibility, so
3680 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3681 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3682 instead of (VALUES T T).
3683 * By following the instructions that Dan Barlow posted to sbcl-devel
3684 on 2 July 2000, I was able to enable primitive dynamic object
3685 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3686 functionality is not implemented (since it calls ld to resolve
3687 library references automatically, requiring RUN-PROGRAM for its
3688 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3689 try to resolve library references) is now supported.
3690 * The system now flushes the standard output streams when it terminates,
3691 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3692 them at several other probably-convenient times, e.g. in each pass of
3693 the toplevel read-eval-print loop, and after evaluating a form given
3694 as an "--eval" command-line option. (These changes were motivated by a
3695 discussion of stream flushing issues on cmucl-imp in August 2000.)
3696 * The source transform for TYPEP of array types no longer assumes
3697 that an array whose element type is a not-yet-defined type
3698 is implemented as an array of T, but instead punts, so that the
3699 type will be interpreted at runtime.
3700 * There is now some support for cross-compiling in make.sh: each of
3701 the phases of make.sh has its own script. (This should be transparent
3702 to people doing ordinary, non-cross-compile builds.)
3703 * Since my laptop doesn't have hundreds of megabytes of memory like
3704 my desktop machine, I became more motivated to do some items on
3705 my to-do list in order to reduce the size of the system a little:
3706 ** Arrange for various needed-only-at-cold-init things to be
3707 uninterned after cold init. To support this, those things have
3708 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3709 symbols with such names are now uninterned after cold init).
3710 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3711 fasl files for cold load.
3712 ** Remove the old compiler structure pooling code (which used to
3713 be conditional on the target feature :SB-ALLOC) completely.
3714 ** Redo the representation of some data in cold init to be more compact.
3715 (I also looked into supporting byte compiled code at bootstrap time,
3716 which would probably reduce the size of the system a lot, but that
3717 looked too complicated, so I punted for now.)
3718 * The maximum signal nesting depth in the src/runtime/ support code has
3719 been reduced from 4096 to 256. (I don't know any reason for the very
3720 large old value. If the new smaller value turns out to break something,
3721 I'll probably just bump it back up.)
3722 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3724 * Many, many bugs reported by Peter Van Eynde have been added to
3725 the BUGS list; some have even been fixed.
3726 * While enabling dynamic object file loading, I tried to make the
3727 code easier to understand, renaming various functions and variables
3728 with less ambiguous names, and changing some function calling
3729 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3730 * While trying to figure out how to do the OpenBSD port, I tried to
3731 clean up some of the code in src/runtime/. In particular, I dropped
3732 support for non-POSIX signal handling, added various comments,
3733 tweaked the code to reduce the number of compilation warnings, and
3734 renamed some files to increase consistency.
3735 * To support the new automatic configuration functionality in make.sh,
3736 the source file target-features.lisp-expr has been replaced with the
3737 source file base-target-features.lisp-expr and the machine-generated
3738 file local-target-features.lisp-expr.
3739 * fixed a stupid quoting error in make.sh so that using CMU CL
3740 "lisp -batch" as cross-compilation host works again
3742 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3744 * DESCRIBE no longer tries to call itself recursively to describe
3745 bound/fbound values, so that it no longer fails on symbols which are
3746 bound to themselves (like keywords, T, and NIL).
3747 * DESCRIBE now works on generic functions.
3748 * The printer now prints less-screwed-up representations of closures
3749 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3750 * A private symbol is used instead of the :EMPTY keyword previously
3751 used to mark empty slots in hash tables. Thus
3752 (DEFVAR *HT* (MAKE-HASH-TABLE))
3753 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3754 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3755 now does what ANSI says that it should. (You can still get
3756 similar noncompliant behavior if bang on the hash table
3757 implementation with all the symbols you get back from
3758 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3759 This breaks binary compatibility, since tests for equality to
3760 :EMPTY are wired into things like the macroexpansion of
3761 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3763 * There's now a minimal placeholder implementation for CL:STEP,
3764 as required by ANSI.
3765 * An obscure bug in the interaction of the normal compiler, the byte
3766 compiler, inlining, and structure predicates has been patched
3767 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3768 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3770 * Missing ordinary arguments in a macro call are now detected even
3771 when the macro lambda list contains &KEY or &REST.
3772 * The debugger no longer complains about encountering the top of the
3773 stack when you type "FRAME 0" to explicitly instruct it to go to
3774 the top of the stack. And it now prints the frame you request even
3775 if it's the current frame (instead of saying "You are here.").
3776 * As specified by ANSI, the system now always prints keywords
3777 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3779 * The default initial SIZE of HASH-TABLEs is now smaller.
3780 * Type information from CLOS class dispatch is now propagated
3781 into DEFMETHOD bodies, so that e.g.
3782 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3784 is now basically equivalent to
3785 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3786 (DECLARE (TYPE SINGLE-FLOAT X))
3788 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3789 operation, without having to do run-time type dispatch.
3790 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3791 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3792 * Since I don't seem to be making big file reorganizations very often
3793 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3794 is overflowing my ability to conveniently back them up), I've finally
3795 checked the system into CVS. (The CVS repository is on my home system,
3796 not at SourceForge -- putting it on SourceForge might come later.)
3797 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3798 high-level GC-NOTIFY-FOO functions send their output. (There's
3799 still very little control of where low-level verbose GC functions
3800 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3801 less than it used to -- the GC-NOTIFY-FOO functions are now under
3802 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3803 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3804 in only one place in the source code, and propagates it automatically
3805 everywhere that it's needed. Thus e.g. when I bump the version from
3806 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3807 * The C source files now include boilerplate legalese and documentation
3808 at the head of each file (just as the Lisp source files already did).
3809 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3810 to his page will be replaced with a link to his new CLiki service.
3812 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3814 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3815 * The build process now looks for GNU make under the default name "gmake",
3816 instead of "make" as it used to. If GNU make is not available as "gmake"
3817 on your system, you can change this default behavior by setting the
3818 GNUMAKE environment variable.
3819 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3820 can build without error under CMU CL.
3822 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3824 * There is now a partial SBCL user manual (with some new text and some
3825 text cribbed from the CMU CL manual).
3826 * The beginnings of a profiler have been added (starting with the
3827 CMU CL profiler and simplifying and cleaning up). Eventually the
3828 main interface should be through the TRACE macro, but for now,
3829 it's still accessed through vaguely CMU-CL-style functions and macros
3830 exported from the package SB-PROFILE.
3831 * Some problems left over from porting CMU CL to the new
3832 cross-compilation bootstrap process have been cleaned up:
3833 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3834 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3835 part when getting rid of the compiler *BACKEND* stuff.)
3836 ** The value of *NULL-TYPE* was screwed up, because it was
3837 being initialized before the type system knew the final
3838 definition of the 'NULL type. This screwed up several key
3839 optimizations in the compiler, causing inefficiency in all sorts
3840 of places. (I found it because I wanted to understand why
3841 GET-INTERNAL-RUN-TIME was consing.)
3842 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3843 PROCLAIM FTYPE information. Unfortunately this broke binary
3844 compatibility again, since now the forms output by DEFGENERIC
3845 to refer to functions which didn't exist in 0.6.3.
3846 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3847 can use the new (as of 0.6.3) transform for SOME into MAP into
3849 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3850 values to (INTEGER 0 1000000), so that the time code will no longer
3851 occasionally get blown up by Linux returning 1000000 microseconds
3852 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3853 its output conform to the ANSI spec. (Alas, this makes its output
3854 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3855 * A full call to MAP NIL with a single sequence argument no longer conses.
3856 * fixes to problems pointed out by Martin Atzmueller:
3857 * The manual page no longer talks about multiprocessing as though
3858 it were currently supported.
3859 * The ILISP support patches have been removed from the distribution,
3860 because as of version 5.10.1, ILISP now supports SBCL without us
3861 having to maintain patches.
3862 * added a modified version of Raymond Toy's recent CMU CL patch for
3863 EQUALP comparison of HASH-TABLE
3865 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3867 * The system still can't cross-compile itself with
3868 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3869 entails), but at least it can compile more of itself that way
3870 than it used to be able to, and various buglets which were uncovered
3871 by trying to cross-compile itself that way have now been fixed.
3872 * This release breaks binary compatibility again. This time
3873 at least I've incremented the FASL file format version to 2, so that the
3874 problem can be detected reliably instead of just causing weird errors.
3875 * various new style warnings:
3876 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3877 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3878 ** using the lexical binding of a variable named in the *FOO* style
3879 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3880 as specified by ANSI.
3881 * *RANDOM-STATE* is no longer automatically initialized from
3882 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3883 default behavior of the system is to repeat its behavior every time
3884 it's run. If you'd like to change this behavior, you can always
3885 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3886 old convention there was no comparably easy way to get the system to
3887 repeat its behavior every time it was run.
3888 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3889 FTYPE declarations has been removed, in favor of their ANSI
3890 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3891 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3892 the types of their sequence arguments can be determined at compile time.
3893 This is done through a new open code expansion for MAP which eliminates
3894 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3895 argument types can be determined at compile time.
3896 * The optimizer now transforms COERCE into an identity operation when it
3897 can prove that the coerced object is already of the correct type. (This
3898 can be a win for machine generated code, including the output of other
3899 optimization transforms, such as the MAP transform above.)
3900 * Credit information has been moved from source file headers into CREDITS.
3901 * Source file headers have been made more standard.
3902 * The CASE macro now compiles without complaining even when it has
3905 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3907 * (Note that the way that the PCL macroexpansions were rewritten
3908 to accommodate the change in DEFGENERIC below breaks binary
3909 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3910 not run under sbcl-0.6.2. Once we get out of alpha releases,
3911 i.e. hit release 1.0.0, we'll probably try to maintain binary
3912 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3913 and sbcl-1.4.4. Until then, however, it might be fairly common
3914 for maintenance releases to break binary compatibility.)
3915 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3917 * The reporting of unhandled signals has been changed to print some
3918 explanatory text as well as the report form. (Previously only
3919 the report form was printed.)
3920 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3921 it defines, so that the compiler no longer issues undefined function
3922 warnings for compiled-but-not-yet-loaded generic functions.
3923 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3924 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3925 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3926 section of the ANSI spec are supported.
3927 * The "" nickname for the "KEYWORD" package has been removed.
3928 The reader still handles symbol tokens which begin with a package marker
3929 as keywords, but it doesn't expose its mechanism for doing so in the
3930 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3931 * The system now issues STYLE-WARNINGs for contradictory TYPE
3932 proclamations. (Warnings for contradictory FTYPE proclamations would
3933 be nice too, but those can't be done usefully unless the type system
3934 is made smarter about FUNCTION types.)
3935 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3936 systematized, so that "*target-*.lisp is supposed to exist only on the
3937 target and imply that there's a related file which exists on the
3938 host, and *host-*.lisp is supposed to exist only on the host and imply
3939 that there's a related file which exists on the target. This involves a
3940 lot of renaming. Hopefully the acute confusion caused by the renaming
3941 will be justified by the reduction in chronic confusion..
3942 ** runtime-type.lisp -> early-target-type.lisp
3943 ** target-type.lisp -> late-target-type.lisp
3944 ** early-host-format.lisp -> early-format.lisp
3945 ** late-host-format.lisp -> late-format.lisp
3946 ** host-error.lisp -> misc-error.lisp
3947 ** early-error.lisp -> early-target-error.lisp
3948 ** late-error.lisp -> late-target-error.lisp
3949 ** host-defboot.lisp -> early-defboot.lisp
3950 ** code/misc.lisp -> code/target-misc.lisp
3951 ** code/host-misc.lisp -> code/misc.lisp
3952 ** code/numbers.lisp -> code/target-numbers.lisp
3953 ** code/early-numbers.lisp -> numbers.lisp
3954 ** early-host-type.lisp -> early-type.lisp
3955 ** late-host-type.lisp -> late-type.lisp
3956 ** host-typep.lisp -> typep.lisp
3957 ** load.lisp -> target-load.lisp
3958 ** host-load.lisp -> load.lisp
3959 ** host-disassem.lisp -> disassem.lisp
3960 ** host-insts.lisp -> insts.lisp
3961 ** byte-comp.lisp -> target-byte-comp.lisp
3962 ** host-byte-comp.lisp -> byte-comp.lisp
3963 ** host-signal.lisp -> signal.lisp
3964 ** host-defstruct.lisp -> defstruct.lisp
3965 ** late-target-type.lisp -> deftypes-for-target.lisp
3966 Furthermore, several other previously target-only files foo.lisp (e.g.
3967 hash-table.lisp and random.lisp) have been split into a target-and-host
3968 foo.lisp file and a target-only target-foo.lisp file, with their key type
3969 definitions in the target-and-host part, so that the cross-compiler will
3970 know more about target types.
3971 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3972 gone away. In their place are various *BACKEND-FOO* variables
3973 corresponding to the slots of the old structure.
3974 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3975 into the target SBCL has been fixed.
3976 * The chill.lisp system for loading cold code into a running SBCL
3978 * Support for the CMU CL "scavenger hook" extension has been removed.
3979 (It was undocumented and unused in the CMU CL sources that SBCL was
3980 derived from, and stale in sbcl-0.6.1.)
3981 * Various errors in the cross-compiler type system were detected
3982 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3983 (enabling various consistency checks). Many of them were fixed,
3984 but some hard problems remain, so the compiler is back to
3985 running without *TYPE-SYSTEM-INITIALIZED* for now.
3986 * As part of the cross-compiler type system cleanup, I implemented
3987 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3988 * I have started adding UNCROSS calls throughout the type system
3989 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3990 on cross-compiler input files will be able to go away, and various
3992 * CONSTANTP now returns true for quoted forms (as explicitly required
3995 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3997 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3998 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3999 (MAKE-LIST -1). In the longer term, it ought to become true
4000 that declarations are assertions even at SAFETY 1. For now, it's not
4001 quite true even at SAFETY 3, but it's at least more nearly true..
4002 (Note that this change seems to increases the size of the system by
4003 O(5%) and to decrease the speed of the compiler by 20% or more.)
4004 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4005 for the problem of printing dozens of lines of distracting information
4006 about low-level system machinery as part of the top stack frame
4007 on entry to the debugger when an undefined function was called.
4008 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4009 is not reset to COMMON-LISP-USER.
4010 * Compilation of stuff related to dyncount.lisp has been made conditional
4011 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4012 smaller. The various dyncount-related symbols have been moved into
4013 a new "SB-DYNCOUNT" package.
4014 * tty-inspect.lisp has been renamed to inspect.lisp.
4015 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4016 feature has gone away. (When we eventually port to other flavors of
4017 libc and/or Unix, we'll try to make the differences between flavors
4018 invisible at the user level.)
4019 * Various other *FEATURES* tags, and/or their associated conditionals,
4020 have been removed if obsolescent, or given better documentation, or
4021 sometimes given more-mnemonic names.
4023 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4025 * tidied up "make.sh" script
4026 * tidied up system directory structure
4027 * better "clean.sh" behavior
4028 * added doc/FOR-CMUCL-DEVELOPERS
4029 * many many small tweaks to output format, e.g. removing possibly-confusing
4030 trailing #\. character in DESCRIBE-INSTANCE
4031 * (EQUALP #\A 'A) no longer signals an error.
4032 * new hashing code, including EQUALP hashing
4033 * tidied up Lisp initialization and toplevel
4034 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4035 * command line argument processing
4036 * added POSIX-GETENV function to deal with Unix-ish environment variables
4037 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4038 terminating SBCL on EOF
4039 * non-verbose GC by default
4040 * There is no more "sbcl" shell script; the sbcl file is now the C
4041 runtime executable (just like CMU CL).
4042 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4044 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4045 symbol and package information as Lisp native symbol and package objects
4046 instead of strings naming symbols and strings naming packages. This way,
4047 whenever packages are renamed (as in warm init), debug information is
4048 transformed along with everything else.
4049 * tweaked the optimization policy declarations which control the building
4050 of SBCL itself. Now, among other things, the system no longer saves
4051 source location debugging information. (This helps two problems at once
4052 by reducing SBCL size and by keeping SBCL from trying to look for its
4053 sources -- which may not exist -- when reporting errors.)
4054 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4055 debugging and testing purposes
4056 * cleaned up printing, making the printer call PRINT-OBJECT for
4057 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4058 methods, giving nearly-ANSI behavior
4059 * converted almost all special variables to use *FOO* naming convention
4060 * deleted PARSE-TIME functionality, since it can be done portably
4061 * moved some files out of cold init into warm init
4062 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4064 * regularized formatting of source files
4065 * added an install.sh script
4066 * fixed ridiculous memory usage of cross-compiler by making
4067 compiler/alloc.lisp not try to do pooling unless it can hook
4068 itself into the GC of the cross-compilation host. Now the system
4069 builds nicely on my old laptop.
4070 * added :SB-ALLOC in target-features.lisp-expr
4071 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4072 was not implemented)
4073 * re-did condition handling and note reporting in the compiler. Notes
4074 are no longer handled by signalling conditions. Style warnings
4075 and warnings are handled more correctly and reported in such a way
4076 that it's easy to find one or the other in your output (so that you
4077 can e.g. figure out which of many problems caused COMPILE-FILE to
4079 * changed the severity of several compiler warnings from full WARNING
4080 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4081 compiler note reporting so that it doesn't use the condition system
4082 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4083 COMPILE-FILE command)
4084 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4085 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4086 no longer supported, since their implementation was deeply intertwingled
4087 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4088 * removed bogus "support" for compiler macros named (SETF FOO), and
4089 removed the compiler macro for SETF INFO (but only after making a fool
4090 of myself on the cmucl-imp mailing list by posting a bogus patch for
4091 DEFINE-COMPILER-MACRO..)
4092 * Compiled files containing forms which have side effects on the Lisp
4093 reader (such as DEFPACKAGE forms) are now handled more correctly.
4094 (Compiler queuing of top level lambdas has been suppressed by setting
4095 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4096 * deleted various currently-unused source files, e.g. gengc.lisp. They
4097 may be added back at some point e.g. when porting to other architectures,
4098 but until they are it's distracting to distribute them and to try to
4100 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4101 there were so many of them they're just distractions, and UNCROSS is
4102 known to be able to handle the current sources
4103 * moved PROFILE functionality into TRACE, so that it will be clear
4104 how the wrapping and unwrapping of functions when you profile them
4105 interacts with the wrapping and unwrapping of functions when you
4106 trace them. (Actually, the functionality isn't there yet, but at least
4107 the interface specification is there. Hopefully, the functionality will
4108 arrive with some future maintenance release.)
4109 * removed host-oops.lisp
4110 * changed signature of QUIT function to allow UNIX-CODE argument
4111 * fixed READ-SEQUENCE bug
4112 * tweaked verbose GC output so that it looks more like the progress
4113 output that ANSI specifies for functions like LOAD
4114 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4115 * added <http://sbcl.sourceforge.com> to the banner information printed by