1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.19 (1.0.0?) relative to sbcl-0.9.18:
3 * improvement: floating point modes in effect are now saved in
4 core, and restored on startup.
5 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
6 startup, not time since first call to GET-INTERNAL-REAL-TIME.
7 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
8 threads are not running after *SAVE-HOOKS* have run.
9 * bug fix: compiler bug triggered by a (non-standard) VALUES
10 declaration in a LET* was fixed. (reported by Kaersten Poeck)
11 * bug fix: file compiler no longer confuses validated and already
12 dumped structurres. (reported by Kaersten Poeck)
13 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
14 fill-pointer signals a type-error as required. (thanks to
16 * bug fix: disassemly of funcallable instances works.
17 * bug fix: single stepping on PPC.
18 * bug fix: fix thread-safety problems in the type system (generally
19 manifesting as nonsensical errors like "STRING is a bad type specifier
20 for sequences" or "The value 1 is not of type FIXNUM")
21 * Improvements to the Windows port:
22 ** floating point exceptions are now reported correctly.
23 ** stack exhaustion detection works partially.
24 ** more accurate GET-INTERNAL-REAL-TIME.
25 ** better RUN-PROGRAM behaviour in case of errors in the child
27 ** PROBE-FILE now simplifies pathnames correctly.
28 ** DIRECTORY now works correctly with :WILD-INFERIORS.
29 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
30 1st 1970 (although time timezone data will be unreliable).
31 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
33 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
34 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
35 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
37 * bug fix: two potential GC deadlocks affecting threaded builds.
38 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
39 returns T (reported by Anton Kazennikov)
40 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
41 with non-variable places
42 * bug fix: remove a race condition in the setting of
43 funcallable-instance functions, this should make threaded CLOS
44 code more stable against memory faults.
45 * bug fix: corruption of specials when unbinding is interrupted by an
46 asynchronous unwind (reported by Hannu Koivisto)
47 * improvement: the debugger will now also display local variables that
48 are only used once, for code compiled with a DEBUG optimization quality
51 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
52 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
53 * incompatible change: External-format support for FFI calls. The
54 SB-ALIEN:C-STRING no longer implies an ASCII
55 external-format. Instead the string is subject to external-format
56 conversion. Additionally return values of type SB-ALIEN:C-STRING
57 are by default of type (SIMPLE-ARRAY CHARACTER), not
58 SIMPLE-BASE-STRING. For an alien type that behaves like the old
59 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
60 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
61 * incompatible change: SB-EXT package no longer contains the
62 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
63 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
65 * incompatible change: the single-stepper is no longer available
66 on Alpha, Mips and Sparc platforms.
67 * minor incompatible change: the direct superclasses of
68 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
69 not (STANDARD-OBJECT FUNCTION). This makes the
70 class-precedence-lists of GENERIC-FUNCTION and
71 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
73 * minor incompatible change: the default stream external format on
74 non-unicode SBCL is no longer detected from the locale, but is always
76 * new feature: Add a version of evaluator that uses an interpreter instead
77 of the compiler. EVAL still uses the compiler by default, to switch it
78 to use the interpreter, set the value of the variable
79 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
80 * minor incompatible change: the single-stepper REPL has been merged
81 with the normal debugger (see the "Stepping" heading of the debugger help
82 for more details). The debugger command STEP will no longer switch
83 to the single-stepper REPL.
84 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
86 * bug fix: loading of default sysinit file works. (thanks to Leonid
88 * bug fix: better detection of circularities in the file-compiler.
89 (reported by Marco Monteiro)
90 * bug fix: the CL pathname functions now work with files that have
91 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
92 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
93 systems with Unix98 pty semantics.
94 * bug fix: ASDF-INSTALL will now work with bsd tar.
95 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
97 * bug fix: timers expiring in dead threads no longer cause a
98 type-error (reported by Paul "Nonny Mouse").
99 * bug fix: thanks to more lightweight single-stepper instrumentation,
100 code compiled with (DEBUG 3) will compile and execute significantly faster,
101 and will have more accurate type-inferencing than before
102 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
103 whose bindings are modified
104 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
105 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
106 ** replace dummy implementations of CL:MACHINE-INSTANCE and
107 CL:SOFTWARE-VERSION with proper version
109 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
110 * feature: implemented the READER-METHOD-CLASS and
111 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
112 as specified by AMOP.
113 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
115 * optimization: faster LOGCOUNT implementation on x86 and x86-64
116 (thanks to Lutz Euler)
117 * optimization: hashing of general arrays and vectors has been
118 improved. (reported by Any Fingerhut)
119 * enhancement: SB-INTROSPECT is now able to find definitions of
120 profiled functions. (thanks to Troels Henriksen)
121 * enhancement: compiler-macro expansion applies now to FUNCALL forms
122 as well, allowing compiler-macros for SETF-functions to expand.
123 * enhancement: step-instrumentation no longer wraps known
124 single-value functions with multiple-value context, allowing
125 better type inference.
126 * fixed bug #337: use of MAKE-METHOD in method combination now works
127 even in the presence of user-defined method classes. (reported by
128 Bruno Haible and Pascal Costanza)
129 * fixed bug #339(c): if there are applicable methods not part of any
130 long-form method-combination group, call INVALID-METHOD-ERROR.
131 (reported by Bruno Haible)
132 * fixed bug #361: the :FUNCTION initarg in the protocol for
133 initialization of methods can now be used to override
134 internally-produced optimized functions. (reported by Bruno
136 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
137 system-provided lambda expression no longer cause warnings about
138 unbound #:|pv-table| symbols.
139 * bug fix: improved the handling of type declarations and the
140 detection of violations for keyword arguments with non-constant
142 * bug fix: potentially erronous calls to PATHNAME and
143 MERGE-PATHNAMES were being flushed in some cases.
144 (reported by Richard Kreuter)
145 * bug fix: compiled calls to TYPEP were mishandling obsolete
146 instances. (reported by James Bielman and Attila Lendvai)
147 * bug fix: format strings with ~> without matching ~< no longer
148 trigger an AVER, but signal an understandable error instead.
149 (reported by Antonio Martinez)
150 * bug fix: specifying an output-file without a pathname-type for
151 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
152 of input-file instead of "fasl". (reported by Robert Dodier)
153 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
154 (reported by James Y Knight).
155 * bug fix: compiler-macros-function did not consider the environment
156 argument for shadowing by local functions.
157 * bug fix: compiler-macros expansion was inhibited by local INLINE
159 * bug fix: inline expansions of known functions were subject to
160 step-instrumentation in high debug policies, leading to problems
162 * bug fix: compiler failed to differentiate between different CONS
164 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
165 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
166 * bug fix: #368: incorrect use of expressed vs. upgraded array
168 * bug fix: #306a: more precise unions of array types.
169 * thread-safety improvements:
170 ** CONDITION-WAIT could return early on Linux, if the thread was
171 interrupted and subsequently continued with SIGCONT.
172 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
174 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
175 * added support for the ucs-2 external format. (contributed by Ivan
177 * minor incompatible change: pretty printing of objects of type
178 (cons symbol) is, in the default pprint-dispatch-table, now
179 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
181 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
182 called later in a class's lifetime, possibly as late as when the
183 first instance of the class is created. Previously,
184 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
185 class became finalizeable.
186 * fixed bug: FILE-POSITION sometimes returned inconsistent results
187 for multibyte external-format streams. (thanks to "vbzoli")
188 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
189 with :ALLOCATION :CLASS inherited from superclasses of the
191 * fixed bug: anonymous classes can now be created using the :NAME
192 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
193 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
194 * fixed bug: core-files saved with :EXECUTABLE T can again be
195 executed when SBCL_HOME isn't set. (reported by James Knight)
196 * fixed bug: toplevel LOCALLY forms with declarations could
197 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
198 * fixed bug: printing from several different threads using different
199 values of *print-case* could cause invalid output, due to
200 some internal special variables of the printer not being bound
201 thread-locally (reported by Max Mikhanosha)
202 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
203 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
204 (reported by Pascal Costanza)
205 * fixed bug: FINALIZE-INHERITANCE is called from
206 REINITIALIZE-INSTANCE on classes when the class has previously
207 been finalized, as required by AMOP.
208 * minor code generation optimizations:
209 ** better register allocation in CLOS dispatching functions
210 ** overflow detection when coercing signed bytes to fixnums on x86-64
211 ** is now implemented with one IMUL instruction instead of three shifts
212 ** more efficient bit-vector access on x86 and x86-64
213 ** more efficient access to raw structure slots on x86 and x86-64
214 * fixed some bugs revealed by Paul Dietz' test suite:
215 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
218 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
219 * feature: thread support on Solaris/x86, and experimental thread support
221 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
223 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
224 default method for SLOT-UNBOUND.
225 * minor incompatible change: prevent the user from specializing the
226 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
227 somewhat counter to the intent of the protocol, I (CSR) think, and
228 additionally it just doesn't work in SBCL as currently
229 implemented, thanks to optimizations (that are always valid for
230 the other three SLOT-VALUEish functions, but not for the setter).
231 * minor incompatible change: the :SB-LDB feature is now enabled by
232 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
233 the low-level debugger.
234 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
235 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
236 down to an SB-SYS:FD-STREAM, allowing direct communication in
237 more cases. Temporary files and pipes are still used as a fallback
239 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
240 on the non-reentrant C functions localtime(3) and gmtime(3).
241 * bug fix: global optimization policy was not visible in LOCALLY and
243 * bug fix: class objects can be used as specializers in methods.
244 (reported by Pascal Costanza)
245 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
246 with a type error. (reported by blitz_ on #lisp)
247 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
248 failed with a type error. (reported by Pascal Bourguignon)
249 * bug fix: merging pathnames against defaults with :DIRECTORY
250 starting with '(:RELATIVE :BACK) should preserve the :BACK.
251 (reported by James Y Knight)
252 * bug fix: saving large (>2GB) cores on x86-64 now works
253 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
254 constant index and a (simple-array (signed-byte 32)) array
255 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
256 error (patch by Robert J. Macomber)
257 * feature: TIME now displays the amount of run-time spent in GC
258 * bug fix: The debugger now does a better job of respecting
259 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
260 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
261 workaround for bug 403.)
262 * bug fix: floating point exception handling now works on FreeBSD
263 (thanks to NIIMI Satoshi)
264 * fixed some bugs revealed by Paul Dietz' test suite:
265 ** MISC.641: LET-conversion were not supposed to work in late
267 * improvements to the Win32/x86 port:
268 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
269 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
271 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
272 * new feature: source path information is generated for macro-expansion
273 errors for use in IDE's like Slime (thanks to Helmut Eller)
274 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
275 * bug fix: compiler does not loop forever on an invalid type in
277 * improvement: compilation of most CLOS applications is significantly
279 * optimization: added a limited bytecode compiler for simple toplevel
280 forms, speeding up compilation and FASL loading
281 * bug fix: the statistical profiler now properly distinguishes anonymous
284 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
285 * minor incompatible change: in sbcl-0.9.11 (but not earlier
286 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
287 into subclasses of the finalized class. Now user calls to
288 FINALIZE-INHERITANCE finalize just the one class, and calls by the
289 system return before any subclasses are finalized.
290 * minor incompatible change: The reader no longer ignores errors
291 regarding non-existent packages in #+ and #- feature tests.
292 * new feature: command line options --no-sysinit, --no-userinit to
293 inhibit loading the corresponding init files
294 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
295 for sending data through UDP sockets (thanks to François-René Rideau)
296 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
297 error is signalled instead (thanks to François-René Rideau)
298 * minor incompatible change: Error signalling behaviour of lexical
299 operations violating package locks has changed slightly. Refer to
300 documentation on package locks for details.
301 * enhancement: EVAL can process IF-expressions without resorting to the
303 * bug fix: Some operations involving SIN, COS, or TAN weren't being
304 constant-folded properly due to missing out-of-line functions.
305 (reported by Mika Pihlajamäki)
306 * bug fix: LISTEN sometimes returned T even in cases where no data was
307 immediately available from the stream
308 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
309 were reversed. (reported by Levente Mészáros)
310 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
311 when given a vector displaced to an adjustable array. (reported
313 * bug fix: derivation of float boundaries from numbers outside the
314 appropriate float range (reported by John Wiseman)
315 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
317 * bug fix: precomputing cache entries for generic functions with
318 some subclasses of specializers as yet invalid does not attempt to
319 fill a cache line with a negative offset. (reported by Levente
321 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
322 allow condition class objects as arguments to DOCUMENTATION and
323 (SETF DOCUMENTATION); only find and set documentation for
324 structure names for the STRUCTURE doc-type. (suggested by Gary
326 * improvements to the Win32/x86 port:
327 ** (user-homedir-pathname) and default initialization file
328 locations now know about the user's "Documents and Settings"
329 directory (thanks to Yaroslav Kavenchuk)
330 ** run-program is implemented (thanks to Mike Thomas)
331 ** sockets support (thanks to Timothy Ritchey)
332 ** better backtrace support (thanks to Alastair Bridgewater)
333 ** sb-grovel supported
334 ** asdf-install and sb-posix work somewhat
335 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
336 * improvements to the Solaris/x86 port:
337 ** works on Solaris 11/Solaris Express
338 ** floating-point exception handling support
339 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
340 * fixed some bugs revealed by Paul Dietz' test suite:
341 ** REMOVE-METHOD returns its generic function argument even when
342 the method is not one of the generic functions' methods.
343 ** objects declared as MEMBER types can be admissible arguments to
345 ** printing characters should simply be printed by the FORMAT ~:C
347 ** compiler failure when compiling functions with hairy constant
348 defaults for optional parameters.
349 ** compiler produces wrong code when MAYBE-INLINE-expanding a
350 function, which is already optimized.
352 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
353 * new platform: experimental support for SBCL x86/Darwin, including
354 MacOS X 10.4.5 on Intel.
355 * new feature: Unicode character names are now known to the system
356 (through CHAR-NAME and NAME-CHAR).
357 * new feature: the filesystem location of SBCL's core file is
358 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
359 * minor incompatible change: the contrib modules SB-POSIX and
360 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
361 this change is to make it easier to distribute
362 location-independent binaries.
363 * bug fix: as implied by AMOP, standardized classes no longer have
364 slots named by external symbols of public packages. (reported by
366 * optimization: calling functions via constant symbols -- as in
367 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
368 via the function object as in (FUNCALL #'FOO).
369 * optimization: on x86 and x86-64, the calling convention for the
370 case of dealing with an unknown number of values has been altered
371 to be friendlier to the prediction heuristics implemented,
372 particularly with reference to CALL and RET pairing. (thanks to
373 Alastair Bridgewater)
374 * optimization: on x86, the code for access to thread-local symbol
375 values and binding/unbinding of thread-local symbols is smaller.
376 (thanks to Alastair Bridgewater)
377 * enhancement: CONSTANTP is now able to determine constantness of
378 more complex forms, including calls to constant-foldable standardized
379 functions and some special forms beyond QUOTE.
380 * fixed bug: occasional GC crashes on Solaris/x86
381 * optimization: x86-64 supports stack allocation of results of simple
382 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
383 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
384 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
385 (thanks to James Knight)
386 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
387 by the standard (thanks to Stephen Compall)
389 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
390 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
391 be used for bundling the runtime and the core file into one
392 executable binary. This feature is not currently supported on all SBCL
393 platforms. (thanks to James Bielman and NIIMI Satoshi)
394 * new feature: a generational or ephemeral garbage collector is now
395 the default on the PowerPC platform (both Linux and Darwin). The
396 old Cheney (stop and copy) collector is a build-time option.
397 (thanks to Cyrus Harmon, after Raymond Toy)
398 * minor incompatible change: the method by which SBCL finds its
399 contributed modules has changed; it no longer relies on symbolic
400 links from an $SBCL_HOME/systems directory, but searches directly
401 in the subdirectories of $SBCL_HOME.
402 * enhancement: the dynamic heap size on the Linux/PPC platform is
403 markedly larger, even using the older Cheney garbage collector.
404 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
405 certain complicated string types. (reported by Gary King)
406 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
407 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
408 * fixed bug: DOCUMENTATION on structures no longer signals an
409 error if no structure type of the right name exists. (reported by
411 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
412 applicable non-standard primary or :AROUND method on
413 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
414 no longer cause unbound variable errors. (reported by Kevin Reid)
415 * fixed bug: metacircle resolution in cases where methods have slots
416 added before the slots from STANDARD-METHOD. (reported by Jean
418 * fixed bug: the Power PC assembly code for calling into lisp
419 sometimes computed its return address wrongly (depending on the
420 15th bit of the address, fixed at link time). This used to
421 manifest itself in a segmentation violation while building PCL.
422 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
423 many others over the years)
424 * fixed bug: ROOM no longer reports silly numbers for stack usage if
425 the stack pointer (treated as a signed integer) is negative.
426 (thanks to Peter van Eynde)
428 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
429 * new platform: experimental support for the Windows operating
430 system has been added. (thanks to Alastair Bridgewater)
431 * fixed several bugs in and robustified the PPC FFI (including
432 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
433 * bug fix: allow non-simple string symbol names (reported by Paul
435 * bug fix: interrupt handling on NetBSD (thanks to Richard M
437 * bug fix: saving a core corrupted callbacks on x86/x86-64
438 * bug fix: closed a loophole in metacircularity detection and
439 grounding in the PCL implementation of CLOS.
440 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
441 (reported by tomppa on #lisp)
442 * bug fix: clear the direction flag on Lisp -> C transitions, as
443 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
444 (reported by Andrej Grozin and Hendrik Maryns)
445 * optimization: major improvements to GC efficiency on GENCGC platforms
446 * optimization: faster implementation of EQUAL
447 * optimization: emit more efficient opcodes for some common
448 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
449 * optimization: several other minor code-generation improvements on x86-64
450 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
452 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
453 * minor incompatible change: (SETF CLASS-NAME) and (SETF
454 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
455 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
456 by ANSI as a generic function, and for consistency (SETF
457 GENERIC-FUNCTION-NAME) is treated likewise.
458 * fixed bug #233.b: make constraint propagation notice when a variable
459 value is changed after it is referenced but before it is used
460 * fixed bug #296: no more arbitrary behaviour from filenames with
461 odd characters as --load arguments. (reported by Adam Warner)
462 * fixed bug #390: :CHARACTER-SET pathname components now work as
463 expected. (reported by Tim Daly Jr)
464 * fixed bug #391: complicated :TYPE intersections in slot
465 definitions no longer cause an error in PCL internals.
466 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
467 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
468 returning the number of octets which would be written to the
469 file-stream. (thanks to Robert J. Macomber)
470 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
471 arguments to the internal stream functions. (thanks to David
473 * bug fix: stack exhaustion is now handled on the x86/SunOS
474 platform. (thanks to Daisuke Homma)
475 * bug fix: more accurate ROOM results on GENCGC platforms
476 * optimization: improved type inference for arithmetic-for
477 index variables in LOOP
478 * optimization: faster floating-point SQRT on x86-64
479 * fixed some bugs revealed by Paul Dietz' test suite:
480 ** DOCUMENTATION returns NIL instead of "" for method combinations
481 that don't have a docstring
483 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
484 * minor incompatible change: (SETF CLASS-NAME) and (SETF
485 GENERIC-FUNCTION-NAME) are no longer generic functions, and
486 therefore are not customizeable by user code (as seems to be at
487 least permitted and maybe required by AMOP). As a consolation,
488 however, the SBCL implementation of these functions now calls
489 REINITIALIZE-INSTANCE as specified by AMOP.
490 * bug fix: slot-definition documentation is propagated as per ANSI
491 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
492 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
493 argument, as do eslotd initialization methods. (from Pascal
494 Costanza's "Closer" project)
495 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
496 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
498 * bug fix: it is now possible to have more than one subclass of
499 STANDARD-GENERIC-FUNCTION without causing stack overflow.
500 (reported by Bruno Haible, Pascal Costanza and others)
501 * bug fix: the dependent update protocol now works for generic
502 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
504 * bug fix: condition-class instances corresponding to
505 DEFINE-CONDITION forms are now created eagerly. (reported by
506 Kalle Olavi Niemitalo on comp.lang.lisp)
507 * bug fix: floating point printing is more accurate in some
508 circumstances. (thanks to Simon Alexander)
509 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
510 merged with *DEFAULT-PATHNAME-DEFAULTS*.
511 * bug fix: callbacks on OS X now preserve stack-alignment.
512 (thanks to Cyrus Harmon)
513 * enhancement: the x86-64 disassembler is much better at
514 disassembling SSE instructions. (thanks to Lutz Euler)
515 * enhancement: CHECK-TYPE on a variable now allows type-inference
516 (thanks to Vincent Arkesteijn)
517 * optimization: improved performance of EUC-JP external format.
518 (thanks to NIIMI Satoshi)
519 * optimization: performance improvements to IO on file streams of
520 :ELEMENT-TYPE CHARACTER
521 * optimization: much faster memory allocation on x86-64
522 * optimization: faster garbage collections (latency and throughput)
524 * optimization: faster bignum subtraction and division on x86 and x86-64
525 (thanks to Lutz Euler)
526 * optimization: more accurate type-inference for loops that use a
527 floating point index variable or a negative step.
529 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
530 * bug fix: add a workaround to SBCL looping infinitely at startup on
531 Linux kernels with apparently buggy implementations of personality().
532 (thanks to Svein Ove Aas)
533 * bug fix: Unicode symbols are correctly printed in LDB backtraces
534 (thanks to David Lichteblau)
535 * bug fix: local bindings shadow global symbol macros for
536 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
537 * optimization: non-open coded uses of numeric comparison operators
538 (e.g. >) no longer cons when called with more than one parameter
539 on platforms supporting dynamic-extent allocation.
540 * enhancement: saving cores with foreign code loaded is now supported
541 on MIPS/Linux in addition to the previously supported platforms.
542 * bug fix: FIND-RESTART now tests for activity, not applicability when given
543 a restart object as identifier. (reported by Helmut Eller for CMUCL)
544 * bug fix: division by zero in sb-sprof when no samples were collected
545 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
546 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
547 (thanks to Alexander Kjeldaas)
548 * new feature: ignoring the return values of destructive functions
549 like NREVERSE signals a compile-time style-warning.
550 (thanks to Kevin Reid)
551 * new feature: the alignment of alien structure fields can be
552 explicitly specified. (thanks to Cyrus Harmon)
553 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
554 and dump core on SIGQUIT
556 ** incompatible change: threads do not inherit values of specials
557 from their parents (see manual)
558 ** bug fix: threads stacks belonging to dead threads are freed by the
559 next exiting thread, no need to gc to collect thread stacks anymore
560 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
561 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
562 ** bug fix: dynamic variable and thread start related gc lossage
563 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
565 * fixed some bugs revealed by Paul Dietz' test suite:
566 ** SUBTYPEP is slightly more accurate on heinously complicated
567 CONS types where some of the members have uncertain (in the
568 NIL, NIL sense) type relationships to each other.
570 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
571 no longer purified unless :PURIFY T is explicitly specified.
572 ** Non-purified cores are significantly smaller than before
574 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
575 * new feature: timers based on Zach Beane's excellent timer package
576 * added support for the following external formats: koi8-u,
577 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
578 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
579 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
580 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
581 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
582 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
583 * incompatible change: a threaded SBCL will no longer revert to
584 non-threaded mode on non-NPTL systems, but refuse to start entirely.
585 * bug fix: interrupts are disabled until startup is complete; no
586 more sigsegvs when receiving a signal to soon
587 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
589 * bug fix: add a workaround for the memory randomization features in
590 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
591 workaround will only be in effect on systems with the proc filesystem
593 * bug fix: printing objects of type HASH-TABLE signals a
594 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
596 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
597 have been running for over 50 days. (reported by Gilbert Baumann)
598 * bug fix: the logic for getting names of functions gets less
599 confused when confronded with alternate-metaclass
600 funcallable-instances. (reported by Cyrus Harmon)
601 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
602 consistent, even on internal alternate-metaclass objects.
603 * bug fix: flush closure information collected by physical
604 environment analysis prepass before the main pass. (bug reported
606 * bug fix: compiler pack phase does not modify a hashtable, which is
607 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
608 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
609 may be expanded. (reported by Alexander <malishev> on c.l.l)
610 * bug fix: no more style-warnings when compiling code using LAST
611 under high-SPEED low-SPACE settings. (reported by David Wallin
613 * fixed some incompatibilities between SBCL's MOP and the MOP
615 ** the METAOBJECT class is now implemented;
616 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
617 STANDARD-OBJECT, as required;
618 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
619 now compatible (as required by VALIDATE-SUPERCLASS); there
620 remains a consistency requirement over the presence or absence
621 of the FUNCTION class in the superclass list of finalized
622 classes; see the manual for more details;
623 ** the :AROUND method for COMPUTE-SLOTS on
624 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
625 requested slot ordering.
627 ** bug fix: parent thread now can be gc'ed even with a live
629 ** bug fix: binding a special with PROGV to no value is not
630 the same as that symbol not having been bound (thanks to
632 ** bug fix: binding specials is thread safe (thanks to
634 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
635 and SB-SPROF do not die with 'no handler for signal XX in
636 interrupt_handle_now(..)' anymore
637 ** bug fix: WITH-TIMEOUT works with multiple running threads
638 * fixed some bugs revealed by Paul Dietz' test suite:
639 ** ENSURE-GENERIC-FUNCTION should take a method class object for
640 the :method-class keyword argument.
642 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
643 * new port: the Solaris operating system on x86 processors is now
644 mostly supported, though some rough edges in the environment
645 remain. (thanks to Daisuke Homma)
646 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
647 space for its heap. (thanks to Thiemo Seufer)
648 * optimization: SBCL on MIPS platforms now supports dynamic-extent
649 closures. (thanks to Thiemo Seufer)
650 * minor incompatible change: eof selects abort in the debugger.
651 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
652 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
654 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
655 deprecated, and will go away in a future revision of SBCL.
656 * minor incompatible change: GC-ON and GC-OFF are no longer
657 implemented with a counter, it does not matter how many times gc
658 is switched on or off
659 * bug fix: discriminating functions for generic function classes
660 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
661 make invalid assumptions about method precedence order. (reported
663 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
664 platforms. (thanks to Thiemo Seufer)
665 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
666 * bug fix: correct pathname printing with printer escaping is on.
667 (thanks to Kevin Reid)
668 * bug fix: complex VOP definitions in "user-space" no longer trigger
669 package locks. (reported by Zach Beane)
670 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
671 for funcallable-instance functions, and is no different from
673 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
674 characters in the current readtable. (reported by Nicholas Neuss)
675 * bug fix: SUBTYPEP on various CONS types returns more a more
676 accurate acknowledgment of its certainty.
677 * optimizations: REMOVE-DUPLICATES now runs in linear time on
678 lists in some cases. This partially fixes bug 384.
679 * flush all standard streams before prompting in the REPL and the
681 * bug fix: signal handling and triggering gc do not conflict
682 directly anymore, in particular a high frequency sb-sprof does
683 not prevent gc from running
684 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
685 approximation for timezone and DST information between the
686 universal time epoch and the smallest negative 32-bit time_t.
687 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
688 year 1899 so long as the encoded time is non-negative.
689 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
690 the correct answer for non-constant rotate amounts when performing
691 an inline 32-bit rotation.
693 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
694 there is only one thread in the session
695 ** bug fix: memory leak for streams created in one thread and
696 written to in another
697 ** bug fix: lockup when compiled with gcc4
698 ** bug fix: race that allows the gc to be triggered when gc is
700 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
701 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
703 ** bug fix: finished threads are gc'ed properly
704 * fixed some bugs revealed by Paul Dietz' test suite:
705 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
706 has arguments with a different set of applicable methods from
707 the orignal arguments.
708 ** The type error thrown by MAP now has a correct expected-type
710 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
711 name a compiled function.
712 ** Three MISC tests where a large bignum was improperly coerced to
713 a float that couldn't represent that bignum during type
714 derivation were fixed.
715 ** SUBTYPEP can now handle types involving the negation of a
716 list-form FUNCTION type.
717 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
718 as (COMPLEX (AND RATIO (NOT FIXNUM))).
719 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
721 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
722 * New feature: Experimental support for bivalent streams: streams
723 opened with :element-type :default now allow character and binary
724 (unsigned-byte 8) I/O
725 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
726 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
728 * bug fix: no more highly sporadic "couldn't check whether ~S is
729 readable" when reading a stream and an interrupt hits in the middle
730 of a select system call
731 * compiler better recognizes complex arrays (reported by Eduardo
733 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
734 for variables on SBCL built with linkage-tables. (reported by Luis
736 * various error reporting improvements.
737 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
738 (Thanks to Thiemo Seufer)
739 * bug fix: nested reader invokations work correctly
740 * bug fix: it is possible to have simultaneous references to foreign
741 code and foreign data with the same name.
743 ** added x86-64 support
744 ** incompatible change: the threading api now works with thread
745 objects instead of thread ids
746 ** bug fix: threads are protected from signals and interruption when
747 starting up or going down
748 ** bug fix: a race where an exiting thread could lose its stack to gc
749 ** bug fix: don't halt on infinite error in threads if possible
750 ** fixed numerous gc deadlocks introduced in the pthread merge
751 ** bug fix: fixed thread safety issues in read and print
752 ** bug fix: debugger doesn't hang on session lock if interrupted at
753 an inappropriate moment
754 ** bug fix: run-program is now thread safe(r)
755 ** bug fix: inner with-recursive-lock no longer releases the mutex
756 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
757 * fixed some bugs revealed by Paul Dietz' test suite:
758 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
759 EXPECTED-TYPE slots filled.
760 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
762 ** MAKE-CONDITION accepts classes as type-designators.
763 ** COMPILE may never return NIL.
764 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
765 range before calling Unix time functions
767 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
768 * numerous signal handling fixes to increase stability
769 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
770 * minor incompatible change: we now correctly canonize default
771 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
772 as per the MOP, rather than the historical (INITARG INITFUNCTION
773 INITFORM). (reported by Bruno Haible)
774 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
775 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
776 for more information.
777 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
778 pathname is a directory pathname.
779 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
780 * dynamic space size on PPC has been increased to 768Mb. (thanks to
782 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
783 :METACLASS argument in addition to a class name. (reported by
784 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
785 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
786 (reported by Bruno Haible; patch from Gerd Moellmann)
787 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
789 * bug fix: more cleanups to the floating point exception handling on
790 x86-64 (thanks to James Knight)
791 * bug fix: the compiler does not try to constant fold calls of
792 COERCE and BIT-* functions when they return freshly constructed
793 objects. (reported by Nikodemus Siivola and Paul Dietz)
794 * optimization: improved the MIPS versions of generic (in the
795 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
796 * optimization: direct conversion of (unsigned-byte 32) to floats on
797 the PowerPC platform.
798 * optimization: structure instances with raw slots now use less
799 memory, and probably show better memory locality. (thanks to
801 * optimization: DYNAMIC-EXTENT declarations for lists and closures
802 are treated as requests for stack allocation on the x86-64,
803 Alpha-32, PPC and SPARC platforms.
804 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
805 the wrong file descriptor; implementation of SOCKET-OPEN-P.
806 (thanks to Tony Martinez)
808 ** gcing a dead thread can no longer lead to lockups
809 ** threads block signals until they are set up properly
810 ** errno is no longer shared by threads
811 ** interrupt-thread restores the eflags register on x86
812 ** fixed some lockups due to gc/thread interaction
813 * fixed some bugs revealed by Paul Dietz' test suite:
814 ** invalid dotted lists no longer raise a read error when
816 ** signal an error if a symbol that names a declaration is used
817 as the name of a type, or vice versa
818 ** allow using the (declare (typespec var*)) abbreviation for
819 (declare (type typespec var*)) with all type specifiers
820 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
821 ** always return NIL from PPRINT-POP when OBJECT is NIL
822 ** don't signal errors when pretty-printing malformed LABELS,
823 FLET or MACROLET forms
824 ** declarations in a DOLIST body are also in scope for the
826 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
828 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
831 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
832 * fixed cross-compiler leakages that prevented building a 32-bit
833 target with a 64-bit host compiler.
834 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
835 opened with :IF-EXISTS :OVERWRITE.
836 * fixed bug 281: error for an invalid qualifier in a short-form method
837 combination method is not signalled until the faulty method is called.
838 * bug fix: iteration variable type inferrer failed to deal with open
839 intervals. (reported by Alan Shields)
840 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
841 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
842 or more runtime options were provided to the sbcl binary.
843 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
844 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
845 * optimization: there's now a fast-path for fixnum arguments in the
846 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
847 * optimization: the code generated on x86-64 is more compact thanks
848 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
849 * bug fix: floating-point exception handling now partly works on
850 x86-64. (Thanks to James Knight)
851 * improvement to the MIPS backend from Thiemo Seufer: C-style
852 64-bit long long arguments and return values to alien functions
854 * fixed some bugs revealed by Paul Dietz' test suite:
855 ** the type-error signalled from WARN has a filled-in DATUM slot.
856 ** the type-error required when a stream is not associated with
857 a file has the stream as its datum.
858 ** type-errors on single-floats on x86-64 no longer have
859 :INVALID-OBJECT as the datum
860 ** the type-errors signalled for invalid function names now have
861 a correct expected type
862 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
863 for structures defined with a :TYPE. Documentation strings for
864 typed structures are no longer immediately discarded
865 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
866 broadcast streams and synonym streams. FILE-LENGTH now also works
867 on broadcast streams.
869 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
870 * incompatible change: the --noprogrammer option, deprecated since
871 version 0.7.5, has been removed. Please use the equivalent
872 --disable-debugger option instead.
873 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
875 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
876 inoperational for a while now) has been completely removed.
877 * null lexical environments are now printed as #<NULL-LEXENV>,
878 significantly reducing the amount of clutter in typical backtraces.
879 * documentation on weak pointers, finalization, and after GC hooks
880 has been added to the manual.
881 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
882 as other specialized array types, is much faster. SUBSEQ and
883 COPY-SEQ on such arrays have also been sped up.
884 * optimization: EQL is now more efficient when at least other argument
885 is known to be of type (OR FIXNUM (NOT NUMBER)).
886 * fixed bug: the runtime is now less vulnerable to changes in the
887 size of the SBCL object on OS X, and virtual memory is reserved for
888 all spaces that need to be at a fixed address.
889 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
890 * fixed bug: finalizers and after GC hooks that cause consing are now
892 * fixed bug: compiler error messages and summaries are now printed to
893 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
894 * fixed inference of the upper bound of an iteration variable.
895 (reported by Rajat Datta).
896 * fixed bug 211e: calling local functions with duplicated constant
897 keyword argument no longer causes a bogus style warning about an
899 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
900 ftype declaration to be disregarded. (reported by Dave Roberts)
901 * fixed bug 373: caused by erronous compilation of references to alien
902 variables in the runtime on ppc/darwin.
903 * fixed bug 376: CONJUGATE type deriver.
904 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
905 (reported by Baughn on #lisp)
906 * a cross-compiler bug on non-x86 platforms has been identified and
907 fixed. (thanks to Bruno Haible)
908 * improvements to the MIPS runtime code for increased stability.
909 (thanks to Thiemo Seufer)
910 * increased the maximimum compact environment size to allow
911 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
912 * partial workaround for bug 135: don't name the function we're
913 calling for hairy cases of EVAL, so as not to accumulate one environment
914 entry per call to EVAL. (reported by Kevin Reid)
915 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
916 * optimization: type testing for non-vector arrays should be faster.
917 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
919 * bug fix: setting 31st element of a bit vector to zero did not work
921 * bug fix: redefining a class definition which failed due to a
922 previous accessor / function clash now works (but see BUGS entry
923 #380 for more problems in this area). (thanks to Zach Beane)
924 * the long form of DEFINE-METHOD-COMBINATION disables method group
925 checking when given a single method group with pattern *. (thanks
927 * on x86 compiler supports stack allocation of results of simple
928 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
929 * support for building and running on Mac OS X 10.4 "Tiger" has been added
930 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
931 * fixed some bugs related to Unicode integration:
932 ** the restarts for recovering from input and output encoding
933 errors only appear when there is in fact such an error to
935 * fixed some bugs revealed by Paul Dietz' test suite:
936 ** MISC.549 and similar: late transformation of unsafe type
937 assertions into derived types caused unexpected code
939 ** SCALE-FLOAT type deriver is less wrong.
940 ** type derivers for EXP, LOG and similar functions compute result
941 types for complex arguments better.
942 ** (MISC.563) CONJUGATE type deriver works for very restricted
944 ** out-of-line type testers for character strings are available.
945 ** EQUAL compiler transform understands specialness of objects
947 ** accessing double-floats stored on the stack now works on x86-64.
948 ** debugger internals could sometimes create invalid lispobjs,
949 resulting in GC crashes.
950 ** MISC.548: type check weakening can convert required type into
952 ** initialization forms for bindings are not in scope of free special
955 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
956 * incompatible change: thread support for non-NPTL systems has
957 been removed - locking is buggy and unreliable. A threaded
958 SBCL build will now warn at startup and refuse to create
959 new threads, unless futex support is detected in the kernel
960 * incompatible change: the top level REPL now has only an ABORT
961 restart associated with it, not TOPLEVEL and ABORT as it used to.
962 TOP and TOPLEVEL are now available as debugger commands for
963 returning to the top level.
964 * incompatible change: forms evaluated in the REPL now use the
965 global optimization policy.
966 * incompatible change: user- and system-initialization files are
967 no longer processed with LOAD, but by READ and EVAL; hence the
968 global optimization policy, startup package, readtable, etc,
970 * The .fasl file format number has been incremented because of
971 various incompatible changes.
972 * internal entry point details and argument counts no longer appear
973 in backtraces unless explicitly requested by setting
974 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
975 * built-in and standard functions no longer have names like "top
976 level local call to FOO".
977 * fixed bug 32: functions defined in non-null lexical environments
978 now have more legible printed representation
979 * fixed bug 33: functions defined in non-null lexical environemnts
980 are now more amenable to inspection by INSPECT.
981 * workaround for bug 354: XEPs no longer appear in backtraces unless
982 explicitly requested.
983 * fixed bug: receiving the signal which results from attempting to
984 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
985 SIGBUS on other *BSDs) on architectures where the C stack is also
986 the Lisp stack (x86 and x86-64) from foreign code no longer leads
987 to debugger confusion or wild instability. (reported by Cheuksan
989 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
990 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
991 (reported by Lutz Euler)
992 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
993 compiled code. (reported by Paul Dietz)
994 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
995 specialized parameter to a method to something that is not TYPEP
996 the specializer is now possible.
997 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
998 face of package deletion.
999 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1000 pathname. (reported by David Tolpin on c.l.l)
1001 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1002 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1003 STRING-STREAM and FILE-STREAM, respectively.
1004 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1005 than previously; in addition, its test suite is now run on build.
1006 (thanks to Luke Gorrie)
1007 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1008 * optimization: added a immediate representation for single-floats
1010 * fixed some bugs revealed by Paul Dietz' test suite:
1011 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1012 ** attempting to create a package with a colliding nickname causes
1013 correctable errors to be signalled.
1014 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1015 ** several x86-64 backend bugs related to sign-extension of immediate
1018 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1019 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1020 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1022 * fixed loading of multiply forward-referenced layouts.
1023 (thanks to Cheuksan Wang)
1024 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1025 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1026 (see COMPILE-FILE documentation for details of :PRINT option)
1027 * fixed bugs 19 and 317: fixed-format floating point printing is
1028 more accurate. This also fixes a bug reported by Adam Warner
1029 related to the ~@F format directive.
1030 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1032 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1033 dispatch table if the from-char is a dispatch macro character.
1034 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1035 by the wordsize no longer ignore the last word. (reported by Lutz
1037 * fixed bug in type checking of dynamic-extent variables. (reported
1039 * optimization: sequence traversal functions use their freedom to
1040 coerce function designators to functions.
1041 * optimization: code with many calls to CLOS methods specialized on
1042 CLOS classes has had redundant type checks removed.
1043 * optimization: added declarations to speed up operations that access
1044 the internal character database (for example STRING-UPCASE)
1045 * optimization: comparison operations between floats and sufficiently small
1046 fixnums no longer create extra rationals
1047 * fixed some bugs related to Unicode integration:
1048 ** portions of multibyte characters at the end of buffers for
1049 character-based file input are correctly transferred to the
1050 start of the buffer at the next read.
1051 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1052 passing it through to OPEN.
1053 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1054 argument given to internal calls to OPEN.
1055 ** the built-in comment readers (introduced by character sequences
1056 ";" and "#|") are more forgiving to encoding errors; they will
1057 STYLE-WARN and then attempt to resync the stream at a character
1058 boundary. (thanks to Teemu Kalvas)
1059 * fixed some bugs in the x86-64 port:
1060 ** Negative short int return values from c-calls are sign-extended
1062 ** The stack is aligned to 16-bytes for c-calls, as required by
1063 the ABI. (thanks to Cheuksan Wang)
1064 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1065 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1066 for enums. (thanks to Vincent Arkesteijn)
1067 ** Multiple small optimizations and bugfixes for floating point
1069 * fixed some bugs revealed by Paul Dietz' test suite:
1070 ** Space, Tab, Linefeed, Return and Page have the invalid
1071 secondary constituent character trait.
1072 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1074 ** WITH-INPUT-FROM-STRING should only update the index place on
1076 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1077 works more reliably.
1078 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1079 with one-dimensional arrays with fill pointers.
1080 ** TYPE-OF failed on a complex with an integer realpart and a
1082 ** compiler failure during type inference for the code of form
1083 (IF (EQL X (THE ...)) ...) (MISC.535).
1085 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1086 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1087 platform. The port supports 61-bit fixnums, large memory spaces
1088 and reloading shared object files.
1089 * enhancement: saving cores with foreign code loaded is now
1090 supported on ppc/Darwin in addition to the previously supported
1092 * enhancement: the statistical profiler now walks deeper into the
1093 call stack for more meaningful call-graphs and accrued time
1094 reports (x86/x86-64 only). It also now reports time spent in
1096 * enhancement: it is now possible to trace most individual methods
1097 of a generic function in addition to tracing the generic function
1099 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1101 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1102 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1103 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1104 produces an error. (thanks to Vincent Arkesteijn)
1105 * bug fix: NAMESTRING on pathnames with :WILD components in their
1106 directories works correctly. (thanks to Artem V. Andreev)
1107 * fixed bug 125: compiler preserves identity of closures. (reported
1109 * bug fixed: functions with &REST arguments sometimes failed with
1110 "Undefined function" when compiled with (DEBUG 3). (reported by
1112 * bug fix: overflow during compiling of setting element of a bit
1113 vector with constant index and value. (reported by Timmy Douglas)
1114 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1115 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1116 * on x86 compiler partially supports stack allocation of dynamic-extent
1118 * GO and RETURN-FROM do not check the extent of their exit points
1119 when compiled with SAFETY 0.
1120 * fixed some bugs related to Unicode integration:
1121 ** encoding and decoding errors are now much more robustly
1122 handled; it should now be possible to recover even from invalid
1123 input or output to the terminal. (thanks to Teemu Kalvas)
1124 ** provided a first cut at implementing STRING-TO-OCTETS and
1125 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1126 ** altered the SB-MD5 contributed module slightly, changing the
1127 interface just enough for it to be supportable for builds where
1128 lisp characters are not eight bits.
1129 * fixed some bugs revealed by Paul Dietz' test suite:
1130 ** the FORMATTER-generated functions for ~V[ conditionals require
1131 the correct number of arguments.
1132 ** READ-FROM-STRING returns the mandated second value when applied
1133 to displaced strings.
1134 ** the #\Rubout and #\Backspace characters are treated as invalid
1135 constituent characters by the tokenizer.
1137 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1138 * new feature: reloading changed shared object files with
1139 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1140 * new feature: references to foreign variables and functions
1141 can now be compiled and loaded before the corresponding shared
1142 object file is loaded, as long as the foreign definitions are
1143 available at runtime.
1144 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1145 supported. (thanks to Dan Debertin)
1146 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1147 just keywords. (thanks to Vincent Arkesteijn)
1148 * fixed bug #331: structure-class instances corresponding to
1149 DEFSTRUCT forms are now created eagerly.
1150 * fixed bug #345: backtraces from calls to undefined functions work
1151 on x86 as well. Related bug #61 is now also partially fixed on x86
1152 (backtraces from throws to unknown catch tags.)
1153 * bug fix: lambda-list parsing is now stricter vrt. order and number
1154 of lambda-list keywords.
1155 * bug fix: as specified by AMOP, an error is signalled if a
1156 class-option appears multiple times in a DEFCLASS form. (reported
1158 * bug fix: RANDOM can be compiled when the compiler derives the type
1159 of its numeric argument as a disjoint set of small integers.
1160 (reported by Paul Dietz)
1161 * bug fix: starting a core saved with shared objects loaded when
1162 those objects are not available no longer causes threaded SBCL to
1163 hang. (reported by Sean Ross)
1164 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1166 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1167 argument, which must always be NIL. (reported by Kalle Niemitalo)
1168 * bug fix: printing 1.0d+23 no longer results in an error.
1169 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1170 * bug fix: structure slot setters preserve evaluation order. (thanks
1172 * fixed some bugs related to Unicode integration:
1173 ** RUN-PROGRAM can allow its child to take input from a Lisp
1174 stream. (reported by Stefan Scholl)
1175 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1176 locales. (reported by Ken Causey)
1177 * fixed some bugs revealed by Paul Dietz' test suite:
1178 ** INCF, DECF and REMF evaluate their place form as specified in
1180 ** FORMATTER expands ~{ ~} iteration directives with V or #
1181 parameters correctly.
1182 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1183 consequent uses no arguments correctly.
1184 ** the system has a partial understanding of the (COMPLEX RATIO)
1186 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1189 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1190 * new feature: a build-time option (controlled by the :SB-UNICODE
1191 keyword feature, enabled by default) for building the system with
1192 support for the entire 21-bit character space defined by the
1194 * new feature: the system now has rudimentary external-format
1195 support; the primary user-visible change at this time is that
1196 characters with the high bit set (such as lower-case-e-acute) will
1197 print correctly to a terminal in a UTF-8 environment.
1198 * minor incompatible change: BASE-CHAR no longer names a class;
1199 however, CHARACTER continues to do so, as required by ANSI.
1200 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1201 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1202 has been moved to the SB-EXT package (temporarily re-exported from
1204 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1205 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1207 * fixed bug #351: better error-handlind and reporting for malformed
1209 * fixed bug #350: bignum-printing is now more memory-efficient,
1210 allowing printing of very large bignums, eg. (expt 2 10000000).
1211 (reported by Bruno Haible)
1212 * fixed bug #302: better primitive-type selection for intersection
1214 * fixed bug #308: non-graphic characters now all have names, as
1215 required. (reported by Bruno Haible)
1216 * bug fix: redefining a class with different superclasses now correctly
1217 removes it from the direct-subclasses of its previous superclasses.
1218 (reported by David Morse)
1219 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1220 new value now works. (reported by Bruno Haible)
1221 * bug fix: correct canonicalization of multiple non-standard slot
1222 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1223 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1224 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1225 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1226 now exists, an signals an error.
1227 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1228 during macroexpansion for non-iteration variables as well. (reported
1229 by Bruno Haible for CMUCL)
1230 * bug fix: Cyclic structures and unprintable objects in compiler
1231 messages no longer cause errors. (reported by Bruno Haible)
1232 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1233 and READ-FROM-STRING all now return a primary value of NIL if
1234 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1235 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1236 (reported by Bruno Haible for CMUCL)
1237 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1238 do not designate a condition. (reported by Bruno Haible for
1240 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1241 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1242 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1243 permitting resolution in favour of any of the conflicting symbols.
1244 (reported by Bruno Haible for CMUCL)
1245 * bug fix: EQUAL compiler optimizations is less aggressive on
1246 strings which can potentially compare true despite having distinct
1247 specialized array element types.
1248 * bug fix: unit enumerations can be defined without dividing by
1249 zero. (thanks to Vincent Arkesteijn)
1250 * FORMAT compile-time argument count checking has been enhanced.
1251 (report from Bruno Haible for CMUCL)
1252 * a partial workaround for the bug 262: the compiler does not try to
1253 inline-expand a local function doing RETURN-FROM from a deleted
1254 BLOCK. (thanks to Peter Denno for the bug report and to David
1255 Wragg for the simple test case)
1256 * fixed some bugs revealed by Paul Dietz' test suite:
1257 ** RENAME-PACKAGE allows all package designators as new package
1259 ** constraint propagation and lambda variable substitution are
1260 more cautious in dealing with partially deleted code.
1261 ** compiler sometimes lost reoptimization passes.
1262 ** CERROR, given a condition as condition designator, passes its
1263 remaining arguments to the continue format control without
1265 ** Case-altering FORMAT directives work correctly on non-ASCII
1267 ** The REAL type specifier handles bounds outside the floating
1268 point ranges without signalling FLOATING-POINT-OVERFLOW.
1269 ** Functions with IR1-transformations can create intercomponent
1270 references to global functions.
1271 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1273 ** FORMAT ~R treats a nil value for its first parameter correctly.
1275 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1276 * enhancement: saving cores with foreign code loaded is now
1277 supported on x86/NetBSD and sparc/Linux in addition to the previously
1278 supported platforms.
1279 * bug fix: on some platforms repeated installations caused multiple
1280 copies of HTML documentation to be installed -- should not happen
1281 any more. (reported by Stefan Scholl)
1282 * bug fix: parsing self-recursive alien record types multiple times
1283 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1284 original patch by Helmut Eller for CMUCL)
1285 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1286 (thanks to Richard Kreuter)
1287 * bug fix: defining classes whose accessors are methods on existing
1288 generic functions in other (locked) packages no longer signals
1289 bogus package lock violations. (reported by François-René Rideau)
1290 * bug fix: special variables as DEFMETHOD parameters no longer have
1291 associated bogus type declarations. (reported by David Wragg and
1293 * bug fix: read-write consistency on streams of element-type
1294 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1295 * bug fix: redefiniton of the only method of a generic function with
1296 no DEFGENERIC no longer emits a full WARNING. In addition,
1297 redefinition of generic functions with no DEFGENERIC to an
1298 incompatible lambda list now signals an error. (thanks to Zach
1300 * bug fix: DEFGENERIC now works even when there's a function of the
1301 same name in an enclosing lexical environment. (thanks to Zach
1303 * fixed compiler failure, caused by instrumenting code during
1304 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1305 * optimization: added loop analysis and improved register allocation
1306 * fixed some bugs revealed by Paul Dietz' test suite:
1307 ** POSITION on displaced vectors with non-zero displacement
1308 returns the right answer.
1309 ** (SIMPLE-STRING) is a valid type specifier for sequence
1311 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1313 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1314 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1316 ** The justification version of the FORMAT ~< directive treats
1317 non-zero minpad parameter correctly.
1319 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1320 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1321 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1322 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1323 the supported interface.
1324 * new feature: Single-stepping of code compiled with DEBUG 2 or
1325 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1326 * new feature: saving cores with foreign code loaded is now
1327 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1328 Timothy Moore's work for CMUCL)
1329 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1330 parameters to * instead of NIL if no initform is supplied.
1331 (reported by Johan Bockgaard)
1332 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1333 correctly when FUNCALL appears as the car of the form. Note:
1334 despite this FUNCALL forms are not currently subject to
1335 compiler-macro expansion. (port of Raymond Toy's fix for the
1336 same from CMUCL, reported by Johan Bockgaard)
1337 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1338 (thanks for Teemu Kalvas)
1339 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1340 case instead of terminating the iteration (thanks for Julian
1341 Squires, Sean Champ and Raymond Toy)
1342 * bug fix: incorrect expansion of defgeneric that caused a style
1343 warning. (thanks for Zach Beane)
1344 * bug fix: slot accessor effective method computation works properly
1345 for classes with multiple non-standard applicable methods on
1346 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1347 * on x86 compiler supports stack allocation of results of LIST and
1348 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1349 CMUCL implementation by Gerd Moellmann)
1350 * fixed some bugs revealed by Paul Dietz' test suite:
1351 ** FORMAT strings with both the ~<~:;~> form of the justification
1352 directive and pretty-printing directives cause an error.
1354 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1355 * incompatible change: the internal functions
1356 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1357 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1358 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1359 instead of the old functions.
1360 * new feature: on platforms where "dladdr" is available foreign
1361 function names now appear in backtraces. (based on Helmut Eller's
1363 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1364 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1366 * documentation: ASDF manual is now distributed with SBCL.
1367 * bug fix: SBCL can now load its contributed modules with REQUIRE
1368 even if the system-provided entries have been removed from
1369 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1371 * bug fix: backtraces involving undefined functions or assembly
1372 routines are more informative. (thanks to Brian Downing)
1373 * bug fix: mutually referent alien structures now work correctly.
1374 (reported by Rick Taube)
1375 * bug fix: structures defined by WITH-ALIEN can be referred to
1376 within other definitions in the same WITH-ALIEN.
1377 * bug fix: division operators (MOD, TRUNCATE and the like) with
1378 constant zero divisors and integer dividends no longer generate
1380 * bug fix: provide default methods for INPUT-STREAM-P and
1381 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1382 * bug fix: improve the detection and resolution of MOP metacycles
1383 (where effective method computation depends on the generic function
1384 having its effective method computed). (reported by Bruno Haible)
1385 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1386 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1387 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1388 * bug fix: pathnames with relative directory components can be
1389 represented relative to default pathnames.
1390 * optimization: in taking the GCD of bignums, reduce the two bignums
1391 to approximately the same size (using Euclid's algorithm) before
1392 applying the more sophisticated binary GCD. (thanks to Juho
1394 * optimization: COUNT on bitvectors now operates word-at-a-time.
1395 * optimization: ASH with a positive, but not necessarily constant,
1396 (leftwards) shift, when in a modular context, is compiled to a
1398 * fixed some bugs revealed by Paul Dietz' test suite:
1399 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1400 if the corresponding argument is NIL.
1401 ** FORMAT directives accepting a minpad parameter treat negative
1403 ** literal commas as character parameters to FORMAT directives are
1405 ** literal spaces directly after ~<Newline> directives within a
1406 format-logical-block (~:< ~@:>) do not induce :FILL-style
1407 conditional newlines.
1408 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1409 as specified: it no longer includes conditional newlines).
1410 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1412 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1413 :PER-LINE-PREFIX argument does not evaluate to a string.
1415 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1416 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1417 the manual for details; for now, package locks can be disabled by
1418 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1419 if no major problems are found then it is likely that they will be
1420 compiled in unconditionally.
1421 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1422 unsupported operators on all platforms. To load a shared library
1423 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1424 object file, link it into a shared library outside of SBCL and
1425 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1427 * bug workaround: ROOM T and the GC were not working together
1428 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1429 reporting facility aren't preserved. That reporting has been
1430 disabled, so now until and unless someone figures out how to make it
1431 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1432 * minor incompatible change: as threatened around sbcl-0.8.0, the
1433 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1434 The symbols are also exported from SB-PCL for backwards
1435 compatibility, but more so than before SB-PCL should be treated as
1436 an implementation-internal package.
1437 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1439 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1440 It is known as of this release not to work on the Alpha, however.
1441 * fixed bug #167: errors signalled due to illegal syntax in method
1442 bodies are now more legible.
1443 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1444 designators and can hence be used with TYPEP.
1445 * fixed bug #333: CHECK-TYPE now ensures that the type error
1446 signalled, if any, has the right object to be accessed by
1447 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1448 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1449 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1451 * fixed bug #334: programmatic addition of slots using specialized
1452 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1453 and :ALLOCATION :CLASS slots.
1454 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1455 float exponents. (rereported by Peter Seibel)
1456 * fixed a bug: #\Space (and other whitespace characters) are no
1457 longer considered to be macro characters in standard syntax by
1458 GET-MACRO-CHARACTER.
1459 * fixed bug: initialization of condition class metaobjects no longer
1460 causes an instance of the condition to be created. (reported by
1462 * fixed bug: it is now possible to have slots such that
1463 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1464 description is neither :INSTANCE nor :CLASS.
1465 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1466 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1467 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1468 non-standard SLOT-DEFINITION-ALLOCATION values.
1469 * fixed bugs: various race conditions handling exiting threads.
1470 CL-PPCRE's thread torture test now passes.
1471 * fixed arguably-a-bug: GC time proportional to number of threads in
1472 system even when most of them are idle
1473 * optimization: improved performance of BIT and SBIT on bit-vectors.
1474 * .fasl file incompatibility: The fasl file version number has
1475 been incremented because of changes associated with package locks.
1477 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1478 * minor incompatible change: the system no longer provides
1479 optimization hints (conditions which are TYPEP
1480 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1481 mode; these hints are still emitted when higher SPEED optimization
1483 * new contrib module: a sampling profiler (profiling by statistical
1484 sampling, rather than by instrumenting functions) is available as
1485 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1486 * the behaviour of the standard function ED is now customizeable by
1487 third parties through a hook variable: see ED's documentation
1488 string for information on the protocol.
1489 * the compiler no longer emits efficiency notes for (FUNCALL X)
1490 when the type of X is uncertain under default optimization
1492 * fixed bug 276: mutating a binding of a specialized parameter to a
1493 method to something that is not TYPEP the specializer is now
1495 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1496 exhibit the required behaviour on the x86 platform. (thanks to
1497 Peter van Eynde, Eric Marsden and Bruno Haible)
1498 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1499 even for difficult arguments. (reported by Peter Graves)
1500 * fixed bug 141a: the backquote printer now descends quoted
1502 * fixed another bug in backquote printing: no more destructive
1503 modification of the form's list structure. (reported by Brian
1505 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1506 move between its address being taken and the call to
1507 interrupt_thread, fixing a crashing race condition.
1508 * the SB-POSIX contrib implementation has been adjusted so that it
1509 no longer exhibits ridiculously poor performance when constructing
1510 instances corresponding to C structs.
1512 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1513 * minor incompatible change: the sb-grovel contrib now treats C
1514 structures as alien (in the sense of SB-ALIEN) objects rather than
1515 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1516 has implications for memory management of client code
1517 (sb-grovel-returned objects must now be manually managed) and for
1518 type safety (alien objects now have full types).
1519 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1520 used to control emission of compiler diagnostics, rather than the
1521 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1522 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1523 quality should be considered deprecated.
1524 * install.sh now installs the user manual as well
1525 * (not quite a new documentable feature, but worth considering in
1526 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1527 beginnings of a semantically meaningful condition hierarchy is
1528 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1530 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1531 designator as the defaults argument.
1532 * fixed bug: Displaced arrays whose displaced-to array has become
1533 too small now cause ARRAY-DIMENSION to signal an error, providing
1534 for safer bounds-checking. (reported by Bruno Haible)
1535 * fixed bug: DEFCLASS slot definitions with identical :READER and
1536 :WRITER names now signal a reasonable error. (reported by Thomas
1538 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1540 * fixed bug: Invalid slot specification errors now print correctly.
1541 (thanks to Zach Beane)
1542 * fixed bug 320: Shared to local slot value transfers in class
1543 redefinitions now happen corrently from superclasses as
1544 well. (reported by Bruno Haible)
1545 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1547 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1548 improper lists correctly. (reported by Bruno Haible)
1549 * fixed bug 313: source-transform for <fun-name> was erroneously
1550 applied to a call of a value of a variable with name <fun-name>.
1551 (reported by Antonio Menezes Leitao)
1552 * fixed bug 307: The obsolete instance protocol ensures that
1553 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1554 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1555 remove dead unknown-values globs from the middle of the stack.
1556 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1557 easier to resume long computations after using *BREAK-ON-SIGNALS*
1558 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1559 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1560 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1561 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1562 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1563 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1565 * fixed a (fairly theoretical) bug in string printing: if
1566 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1567 does not have array-element-type equal to the most general string
1569 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1570 function name. (thanks to Nikodemus Siivola)
1571 * fixed bug: compile-time format string checker failed on
1572 non-closed ~{. (reported by Thomas F Burdick)
1573 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1574 don't warn on legal loop code involving more than one aggregate
1575 boolean. (thanks to Nikodemus Siivola)
1576 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1577 in keyword position to MAKE-PACKAGE was not being reported as
1579 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1580 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1582 * fixed bugs 280 and 312: the checking for multiple definitions in a
1583 file is less likely to become confused by uses of inline
1585 * fixed bug: the #S reader macro performs the keyword coercion
1586 specified for slot names. (reported by Kalle Niemitalo)
1587 * fixed bug: lambda lists may contain symbols whose names start with
1588 & but are not lambda-list-keywords; their occurrence triggers a
1590 * fixed bug 321: define-method-combination argument lambda lists do
1591 not cause invalid code to be generated when &optional or &aux
1592 variables have default values. (reported by Bruno Haible)
1593 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1594 have CLOS classes; this appears also to have fixed reported
1595 instability in stack exhaustion detection.
1596 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1597 MUFFLE-WARNING when no associated restart is present is now
1599 * optimization: rearranged the expansion of various defining macros
1600 so that each expands into only one top-level form in a
1601 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1603 * optimization: used a previously little-used slot in symbols to
1604 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1606 * fixed some bugs revealed by Paul Dietz' test suite:
1607 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1609 ** Fixed an optimization bug related to inheritance of initforms
1610 from local to shared slots.
1611 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1612 ** CAST optimizer forgot to flush argument derived type.
1613 ** print/read consistency on floats is now orders of magnitude
1614 more likely. (thanks also to Bruno Haible for a similar report
1616 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1617 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1618 for objects of type REAL. Make it so.
1619 ** SXHASH is specified (infelicitously) to respect similarity,
1620 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1621 it so. (thanks to Markus Ziegler)
1622 ** on the Alpha, the compiler succeeds in compiling functions
1623 returning a known number of arguments greater than 63.
1624 ** fixed handling of invalid NIL arguments in keyword position
1626 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1628 ** provided a readably-printable representation for RANDOM-STATE
1630 ** ensured that pathnames, where they have a namestring, always
1631 print using #P"..." syntax.
1633 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1634 * Support for the forthcoming 2.0 version of the NetBSD kernel
1635 running on x86 hardware has been added. (thanks to Perry
1636 E. Metzger most immediately, and others for their past work)
1637 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1638 current release version; SBCL's OpenBSD support had been broken
1639 since about the time of OpenBSD's switch to ELF binary format.)
1640 * [placeholder for DX summary]
1641 ** user code with &REST lists declared dynamic-extent, under high
1642 speed or space and low safety and debug optimization policy.
1643 * The manual has been converted to Texinfo format and the debugger
1644 chapter from the cmucl manual has been added.
1645 * A facility has been added to extract documentation strings from
1646 sbcl and store them as Texinfo-formatted snippets for inclusion in
1647 the manual (via Texinfo's @include directive)
1648 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1649 the test case to Dave Roberts)
1650 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1651 pointers. (reported by Sean Ross)
1652 * bug fix: PROFILE output is printed nicely even for large numerical
1653 values. (thanks to Zach Beane)
1654 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1655 greater than 32 handle EOF correctly.
1656 * bug fix: on X86 an immediate argument of the IMUL instruction is
1657 correctly printed in disassembly. (thanks to Lutz Euler)
1658 * bug fix: class slots in redefined classes preserve their old
1659 values. (thanks to Bruno Haible and Nikodemus Siivola)
1660 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1661 succeeds. (reported by Marco Baringer)
1662 * fixed some bugs revealed by Paul Dietz' test suite:
1663 ** READ-SEQUENCE now works on ECHO-STREAMs.
1664 ** RATIONALIZE works more according to its specification. (thanks
1667 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1668 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1669 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1670 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1671 should matter to you only if you rebind the printer control
1672 variables and then find you want different bindings in the
1673 debugger than in the ordinary execution of your program.)
1674 * The runtime build system has been tweaked to support building
1675 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1676 assembler and linker. (thanks to Nikodemus Siivola)
1677 * Unbound, undefined, undeclared variables now trigger full
1678 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1679 more often programmer error than deliberate exploitation of undefined
1681 * optimization: the hash algorithm for strings has changed to one
1682 that is less vulnerable to spurious collisions. (thanks to Juho
1684 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1685 needless bounds checking. (thanks to Juho Snellman)
1686 * optimization: implemented multiplication as a modular
1687 (UNSIGNED-BYTE 32) operation on the PPC backend.
1688 * fixed some bugs revealed by Paul Dietz' test suite:
1689 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1691 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1692 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1693 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1695 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1697 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1699 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1700 their output stream on EOF from read.
1701 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1702 have been read to end-of-file.
1703 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1705 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1706 description of determination of which consecutive characters
1708 ** Printing the "Space" character with escaping on now yields "#\\ ",
1709 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1710 ** Reading floating-point numbers with *READ-BASE* set to a number
1711 less than 10 works correctly.
1712 ** Reading floating-point numbers with *READ-BASE* set to a number
1713 more than 10 works correctly.
1714 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1715 the readtable currently in effect.
1717 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1718 * minor incompatible change: parsing of namestrings on a physical
1719 (Unix) host has changed; numbers after the final #\. in a
1720 namestring are no longer interpreted as a version field. This is
1721 intented to be largely invisible to the user, except that the
1722 meaning of the namestring "*.*.*" has changed: it now refers to a
1723 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1724 should usually be replaced by
1725 (make-pathname :name :wild :type :wild :version :wild)
1726 with the added benefit that this is more likely to be portable.
1727 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1728 option to OPEN now signals an error if the file being opened
1729 exists; this may have an impact on existing code.
1730 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1731 signals received, on the PowerPC platforms (both Linux and
1732 Darwin). (thanks to Pierre Mai for pointing out the location of
1734 * several fixes on OS X: The system now builds and runs cleanly on
1735 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1736 (thanks to Brian Mastenbrook)
1737 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1738 second-resolution: integer multiples of 1/3600 between -24 and 24.
1739 (thanks to Vincent Arkesteijn)
1740 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1741 type when called with 1 argument; PEEK-CHAR checked type of
1742 PEEK-TYPE only after having read first character from a
1743 stream. (reported by Peter Graves)
1744 * bug fix: the garbage collector now has much better locality
1745 behaviour, and in particular no longer treats all memory as being
1746 exhausted when many small objects point to each other in a deeply
1748 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1749 recognized as being TYPEP their class.
1750 * bug fix: the PUSHNEW documentation string has been corrected.
1751 (thanks to Vincent Arkesteijn)
1752 * bug fix: defaulting of the value for the last of an atypically
1753 large number of multiple values being bound was not being
1754 performed correctly on the Alpha or PPC platforms
1755 * optimization: implemented multiplication as a modular
1756 (UNSIGNED-BYTE 32) operation on the x86 backend.
1757 * optimization: SEARCH on simple-base-strings can now be open-coded.
1758 (see also contrib/compiler-extras.lisp for inspiration for
1759 teaching the compiler about the Boyer-Moore algorithm).
1760 * value, returned by MAX (and MIN) called with several EQUALP, but
1761 not EQL, arguments now does not depend on compiler settings.
1762 * fixed some bugs revealed by Paul Dietz' test suite:
1763 ** in stack analysis liveness information is propagated from
1764 non-local entry points.
1765 ** pathwise CAST removing failed when the CAST node did not start
1767 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1768 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1770 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1771 host is already defined.
1772 ** RENAME-FILE works on streams instead of signalling an internal
1774 ** PEEK-CHAR uses the current readtable when determining whether
1775 or not a character is whitespace.
1776 ** MERGE-PATHNAMES handles the case when the pathname does not
1777 specify a name while the default-pathname specifies a version.
1778 ** Pathnames now stand a better chance of respecting print/read
1780 ** Attempting to use standardized file system operators with a
1781 pathname with invalid :DIRECTORY components signals a
1783 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1785 ** DIRECTORY on logical pathnames is more correct.
1786 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1787 signal a TYPE-ERROR if their argument is not a stream.
1788 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1789 designator argument does not designate a stream.
1790 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1791 examining the synonym.
1792 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1794 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1795 element-type larger than ([UN]SIGNED-BYTE 32).
1797 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1798 * When built with the :SB-FUTEX feature, threaded builds now take
1799 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1800 for faster/more reliable mutex and condition variable support.
1801 * Incompatible change (but one you probably shouldn't have been using
1802 anyway): the interface and code for arbitrating between multiple
1803 threads in the same user session has been redesigned.
1804 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1805 error when called without an explicit environment argument.
1806 (thanks to Nikodemus Siivola)
1807 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1808 confused about their position. (thanks to Adam Warner and Gerd
1810 * bug fix: Pretty printing backquoted forms with unquotations in the
1811 argument list position of various code constructs such as LAMBDA
1812 now works correctly. (reported by Paul Dietz)
1813 * bug fix: Pretty printing unquotations no longer loses all
1814 stream position information.
1815 * optimization: performance of string output streams is now less
1816 poor for multiple small sequence writes.
1817 * optimization: performance of CSUBTYPEP in the presence of complex
1818 expressions involving CONS and NOT many times has been improved.
1819 (reported by Paul Dietz)
1820 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1822 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1823 streams for the REPL, for the debugger, and so on. (thanks to
1825 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1826 * fixed some bugs revealed by Paul Dietz' test suite:
1827 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1828 is ignored if it is too small, rather than propagating through
1829 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1830 ** extremely complex negations of CONS types were not being
1831 sufficiently canonicalized, leading to inconsistencies in
1833 ** VALUES tranformer lost derived type.
1835 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1836 * fixed a bootstrapping bug: the build process no longer assumes
1837 that the various BOOLE-related constants have the same value in
1838 host and target lisps. (noted by Paul Dietz' test suite on an
1839 SBCL binary built from CLISP)
1840 * The system can now be dynamically linked on the MIPS platform,
1841 which enables dynamic loading of foreign code from Lisp. (thanks
1842 to Ralf Baechle for discussions on the MIPS ABI)
1843 * The system now records debugging information for its own source
1844 files in a filesystem-position-independent manner, relative to
1845 the "SYS" logical host.
1846 * fixed a compiler bug: MV-LET convertion did not check references
1847 to the "max args" entry point. (reported by Brian Downing)
1848 * tweaked disassembly notes to be less confident about proclaiming
1849 some instruction as an LRA. (thanks to Brian Downing)
1850 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1851 now each have their own history, command character, and other
1852 characteristics. (thanks to David Lichteblau)
1853 * fixed some bugs revealed by Paul Dietz' test suite:
1854 ** compiler failure in compiling LOGAND expressions including a
1856 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1857 shift greater than 32.
1858 ** FLUSH-DEST did not mark blocks for type check regeneration.
1859 ** HANDLER-CASE failed to accept declarations in handler clauses
1860 in some circumstances.
1862 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1863 * New code in contrib/sb-introspect (still probably not entirely
1864 stable yet) provides some support for smart Lisp development
1865 environments like SLIME.
1866 * The conditions signalled for errors occurring when loading .fasl
1867 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1868 in a way which should help ASDF recover gracefully.
1869 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1870 stuff has been cleaned up. If you code contrib/ stuff, this might
1871 affect you, and you can look at contrib/README, contrib/STANDARDS,
1872 and/or the 0.8.4.27 diff to check.
1873 * In full calls the compiler now does not generate checks for declared
1874 argument types for all arguments.
1875 * various threading fixes
1876 ** and some experimental patches which didn't make it into
1877 the main tree for this release, but which are shipped in
1878 contrib/experimental-thread.patch as a possible fix for some
1879 failures (deadlock, spinning...) in GC-intensive multithreaded
1881 * fixed PPC build problem (source code incompatibility of different
1882 library versions): added offsetof() hackery which attempts to divine
1883 where glibc maintainers put uc_mcontext today
1884 * fixed bug 282: compiler does not trust type assertions while passing
1885 arguments to a full call.
1886 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1887 &OPTIONAL VALUES type parameter.
1888 * fix bug 214: algorithm for noting rejected templates is now more
1889 similar to that of template seletion. (also reported by rydis on
1891 * fixed bug 141b: printing backquoted information readably and prettily
1892 inserts a space where necessary.
1893 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1894 (CHARACTER 10) are now reported as errors, rather than propagated
1895 as unknown types. (reported by piso on #lisp)
1896 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1897 with values NIL and :ERROR. (thanks to Milan Zamazal)
1898 * fixed bug 191c: CLOS now does proper keyword argument checking as
1899 described in CLHS 7.6.5 and 7.6.5.1.
1900 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1901 longer raise an error; further, using a list as a for-as-arithmetic
1902 counter now raises a meaningful error.
1903 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1904 now understood by sequence creation functions such as MAKE-SEQUENCE
1906 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1907 a STRING-INPUT-STREAM.
1908 * compiler enhancement: SIGNUM is now better able to derive the type
1910 * type declarations inside WITH-SLOTS are checked. (reported by
1912 * fixed some bugs revealed by Paul Dietz' test suite:
1913 ** incorrect optimization of TRUNCATE for a positive first
1914 argument and negative second.
1915 ** compiler failure in let-convertion during flushing dead code.
1916 ** compiler failure while deriving type of TRUNCATE on an
1917 interval, containing 0.
1918 ** ASH of a negative bignum by a negative bignum count now returns
1920 ** intersection of CONS types now canonicalizes properly, fixing
1921 inconsistencies in SUBTYPEP.
1923 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1924 * incompatible change: The --disable-debugger command line
1925 option now clobbers the debugger at a more fundamental
1926 level, by redefining #'INVOKE-DEBUGGER instead of by
1927 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1928 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1929 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1930 enter the debugger REPL and then suffer recursive errors
1931 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1932 while under the new system, BREAK in --disable-debugger mode
1933 terminates the system just as an unhandled error would.
1934 * fixed compiler performance when processing loops with a step >1;
1935 * bug fix: DOCUMENTATION now retrieves generic function
1936 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1937 support has been systematized, and now supports the methods
1938 specified by ANSI, along with a default method and a method for
1939 slot documentation. (reported by Nathan Froyd)
1940 * bug fix: effective methods associated with a generic function are
1941 no longer cached over a change of that generic function's method
1942 combination. (reported by Andreas Fuchs)
1943 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1944 names a non-existent directory. (thanks to Andreas Fuchs)
1945 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1946 the PPC platform, lead to stack corruption; this has been fixed.
1947 (reported by Rainer Joswig)
1948 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1949 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1950 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1951 arithmetic in general, is now much more reliable on the PPC
1953 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1954 platform now returns the right answer.
1955 * optimization: restored some effective method precomputation in
1956 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1957 precomputation is now tunable.
1958 * optimization: compiler-internal data structure use has been
1959 reviewed, and changes have been made that should improve the
1960 performance of the compiler by about 20%.
1961 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1962 simple-base-strings and simple-bit-vectors is improved.
1963 * optimization: the optimization of 32-bit logical and arithmetic
1964 functions introduced in version 0.8.3 on the x86 has been
1965 implemented on the mips, ppc and sparc platforms; an
1966 implementation of the same facility, but for 64-bit arithmetic,
1967 has been added for the alpha.
1968 * microoptimization: the compiler is better able to make use of the
1969 x86 LEA instruction for multiplication by constants.
1970 * bug fix: in some situations compiler did not report usage of
1971 generic arithmetic in (SPEED 3) policy.
1972 * bug 145b fix: compiler used wrong type specifier while converting
1973 MEMBER-types to numeric.
1974 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1976 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1977 index argument now works properly on the Alpha platform.
1978 * bug fix: floating point exception treatment on the Alpha platform
1980 * bug fix: FILE-POSITION works much better on string input and
1981 output streams. (thanks to Nikodemus Siivola)
1982 * bug fix: many threading/garbage collection symptoms sorted.
1983 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1984 might be pseudo-atomic.
1985 * internal change: Stopping for GC is now done with signals not
1986 ptrace. GC is now done in whichever thread wanted it, instead of
1988 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1990 * bug fix: result form in DO is not contained in the implicit
1992 * incompatible change: ICR structure is changed; the value part of
1993 CONTINUATION is now called LVAR; corresponding functions are
1994 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1995 * added type deriver for ISQRT (thanks to Robert E. Brown).
1996 * bug fix: better support for loading from the command line when an
1997 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1999 * fixed some bugs revealed by Paul Dietz' test suite:
2000 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2001 ** ROUND and FROUND now give the right answer when given very
2002 small float arguments.
2003 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2005 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2006 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2007 ** LCM should return a non-negative integer.
2008 ** PARSE-INTEGER returned the index of a terminator instead of the
2009 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2010 ** PARSE-INTEGER returned an incorrect index being applied to a
2012 ** LCM with two arguments of 0 returns 0 rather than signalling
2014 ** unsigned addition of a 32-bit constant with the high bit set no
2015 longer causes an internal compiler error on the x86.
2016 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2017 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2018 with negative last argument.
2019 ** byte specifiers with zero size and position no longer cause
2020 an error during type derivation.
2021 ** bignum multiplication on the Alpha platform now returns the
2023 * porting: The system now builds on SuSE AMD64, although it still
2024 generates a 32-bit binary.
2025 * .fasl file incompatibility: The fasl file version number has
2026 been incremented (because of the changes to internal compiler
2027 data structures referred to above).
2029 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2030 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2031 more accurately, on the Darwin kernel running on PowerPC hardware.
2032 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2033 * Compiler code deletion notes now signal a condition of type
2034 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2035 an associated MUFFLE-WARNING restart.
2036 * The compiler now performs limited argument count validation of
2037 constant format strings in FORMAT, and where appropriate in ERROR,
2038 CERROR and WARN. (thanks to Gerd Moellmann)
2039 * New ASDF-INSTALL contrib can be used for automatic download and
2040 installation of third-party Lisp code from CCLAN or other sites
2042 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2043 which forces another thread to execute a function supplied by the
2045 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2046 now accept and act upon their :ELEMENT-TYPE keyword argument.
2047 (reported by Martin Atzmueller, Edi Weitz)
2048 * bug fix: FILE-POSITION now accepts position designators up to
2049 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2050 is the greater. (thanks to Patrik Nordebo)
2051 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2052 Kalle Olavi Niemitalo)
2053 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2054 printable. (reported by Eric Marsden)
2055 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2056 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2057 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2058 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2059 (thanks to Stig Sandoe)
2060 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2061 UNREAD-CHAR now correctly includes the unread character in the
2062 target sequence. (thanks to Gerd Moellmann)
2063 * bug fix in threaded builds: the system can now be suspended and
2064 resumed by shell job control with minimal disruption.
2065 * bug fixes in times and timezones >2038AD
2066 * better handling of "where is GNU make?" problem in build scripts
2067 (thanks to Nikodemus Siivola)
2068 * new optimization: inside a named function any reference to a
2069 function with the same name is considered to be a self-reference;
2070 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2071 optimization quality.
2072 * new optimization on x86: logical functions and + now have
2073 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2074 used when the result is truncated to 32 bits.
2075 * VALUES declaration is partially enabled.
2076 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2077 * fixed some bugs revealed by Paul Dietz' test suite:
2078 ** The system now obeys the constraint imposed by
2079 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2080 types form a lattice under type intersection.
2081 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2082 ** ASSOC now ignores NIL elements in an alist.
2083 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2084 and (1+ MOST-POSITIVE-FIXNUM) answers.
2085 ** The addition of a method with invalid qualifiers to a generic
2086 function does not cause an error to be signalled immediately;
2087 a warning is signalled, and the error is generated only on
2088 calling the generic function.
2089 * changed .fasl file version number, in part to add type codes for
2090 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2091 obscure ANSI requirements
2093 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2094 * fixed bug 148: failure to inline-expand a local function left
2095 garbage, confusing the compiler.
2096 * fixed bugs 3cd: structure slot readers perform type check if the
2097 slot can have an invalid value (i.e. it is either not initialized
2098 or can be written with a less specific slot writer).
2099 * bug fix: the compiler now traps array references to elements off
2100 the end of an array; previously, the bounds checking in some
2101 circumstances could go off-by-one.
2102 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2104 * type declarations for array element types now obey the description
2105 on the CLHS page "Declaration TYPE", as per discussions on
2106 sbcl-help around 2003-05-08. This means that a declaration
2107 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2108 declaration, all references to BAR will be asserted or assumed
2109 (with THE, so dependent on compiler policy) to involve objects of
2110 type FOO. Note that no such declaration is implied in
2111 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2112 * declared types of functions from the "Conditions"
2113 chapter. (reported by Paul Dietz)
2114 * bug fix: CERROR accepts a function as its first argument.
2115 * bug fix: NTH an NTHCDR accept a bignum as index
2116 arguments. (reported by Adam Warner)
2117 * optimization: character compare routines now optimize comparing
2118 against a constant character. (reported by Gilbert Baumann)
2119 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2120 argument now works properly on the MIPS platform.
2121 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2122 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2123 declared to have a functional type, does not cause an error
2125 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2126 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2127 * fixed bug: if last continuation of a deleted block has a
2128 destination, this destination should be deleted too. (reported by
2130 * fixed a bug in the bootstrap process: the host compiler's values
2131 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2132 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2133 test case from Patrik Nordebo)
2134 * improved the ability of the disassembler on the PPC platform to
2135 provide helpful disassembly notes.
2136 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2137 the class in more cases than previously.
2138 * bug fix: FILE-POSITION now understands :START and :END for
2139 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2140 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2141 8) data. (thanks to David Lichteblau)
2142 * bug fix: it is possible to add a method to a generic function
2143 without lambda list.
2144 * bug fix: reader failed to signal END-OF-FILE inside an
2145 object representation. (reported by Nikodemus Siivola)
2146 * fixed some bugs revealed by Paul Dietz' test suite:
2147 ** LAST and [N]BUTLAST should accept a bignum.
2148 ** condition slot accessors are methods.
2149 ** (VECTOR NIL) is a subtype of STRING.
2151 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2152 * minor incompatible change: some nonsensical specialized lambda
2153 lists (used in DEFMETHOD) which were previously ignored now signal
2155 * minor incompatible change: the system is now aware of the types of
2156 variables in the COMMON-LISP package, and will signal errors for
2157 most violations of these type constraints (where previously they
2158 were silently accepted).
2159 * minor incompatible change: COMPILE-FILE now uses the freedom
2160 afforded (ANSI 3.2.2.3) to use derived function types for
2161 functions defined in the same file. This also permits the system
2162 to warn on static type mismatches and function
2163 redefinition. (Currently it does not work with high DEBUG level.)
2164 * minor incompatible change: VALUES declaration is disabled.
2165 * When issuing notes, the compiler now signals a condition of type
2166 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2167 restart for use in user handlers. It is expected that the
2168 COMPILER-NOTE condition will eventually become a condition
2169 supertype to a hierarchy of note types, which will then be
2170 handleable in a similar fashion. However, at the moment, no such
2171 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2172 but it's an implementation detail, not a classification for the
2174 * Changes in type checking closed the following bugs:
2175 ** type checking of unused values (192b, 194d, 203);
2176 ** template selection based on unsafe type assertions (192c, 236);
2177 ** type checking in branches (194bc).
2178 * A short form of VALUES type specifier has ANSI meaning (it has
2179 increased the number of situations when SBCL cannot perform type
2181 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2182 keywords or constants is permissible.
2183 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2184 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2185 classes. (thanks to Antonio Martinez)
2186 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2187 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2188 argument to a pretty-print stream. Instead, it leaves any such
2189 operations to DESCRIBE-OBJECT methods.
2190 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2192 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2193 lambda lists. (thanks to David Lichteblau)
2194 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2195 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2196 (thanks to Gerd Moellmann)
2197 * fixed bug 246: increased compilation speed of long
2198 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2200 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2202 * DEFCONSTANT now throws a condition of type
2203 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2204 constant to a non-EQL value; CONTINUE and ABORT restarts
2205 respectively change and preserve the value.
2206 * fixed bug 63: The code walker, part of the implementation of CLOS,
2207 is now better at handling symbol macros.
2208 * bug fix: There is no longer an internal implementation type named
2209 CL:LENGTH. (reported by Raymond Toy)
2210 * bug fix: In macro-like defining macros/special operators the
2211 implicit block does not enclose the lambda list.
2212 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2213 enclosed in parentheses) are not suitable as type specifiers, and
2214 their use properly signals an error now.
2215 * bug fix: An argument count mismatch for a type specifier in code
2216 being compiled no longer causes an unhandled error at compile
2217 time, but signals a compile-time warning.
2218 * fixed simple vector readable printing
2219 * bug fix: DESCRIBE takes more care over whether the class
2220 precedence list slot of a class is bound before accessing it.
2221 (reported by Markus Krummenacker)
2222 * bug fix: FORMATTER can successfully compile pretty-printer format
2223 strings which use variants of the ~* directive inside.
2224 * bug fix: SEARCH now applies its TEST predicate to the elements of
2225 the arguments in the correct order. (thanks to Wolfhard Buss)
2226 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2227 in type inference. (thanks to Robert E. Brown)
2228 * bug fix: Reading in symbols with an explicit package name of ""
2229 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2230 * fixed some bugs revealed by Paul Dietz' test suite:
2231 ** NIL is now allowed as a structure slot name.
2232 ** Arbitrary numbers, not just REALs, are allowed in certain
2233 circumstances in LOOP for-as-arithmetic clauses.
2234 ** Multiple class redefinitions before slot access no longer
2235 causes a type error.
2236 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2237 association between the name and a class.
2238 ** Generic functions with non-standard method-combination and over
2239 five methods all of which return constants no longer return NIL
2240 after the first few invocations. (thanks to Gerd Moellmann)
2241 ** CALL-NEXT-METHOD with no arguments now passes the original
2242 values of the arguments, even in the presence of assignment.
2243 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2245 ** DISASSEMBLE works with closures and funcallable instances.
2246 ** ADD-METHOD now returns the generic function, not the new method.
2247 ** FIND-METHOD signals an error if the lengths of the specializers
2248 is incompatible with the generic function, even if the ERRORP
2250 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2251 which its argument is a member.
2252 ** DEFCLASS only redefines the class named by its class-name
2253 argument if that name is the proper name of the class;
2254 otherwise, it creates a new class.
2255 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2256 of the UNBOUND-SLOT condition to the name of the slot.
2257 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2258 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2259 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2260 SLOT-MAKUNBOUND in the specified fashion.
2262 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2263 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2264 cross-compilation host. As a consequence, we can now bootstrap our
2265 way up to SBCL starting with a bare gcc toolchain and human-readable
2266 source code (first the source to CLISP, then the source to SBCL).
2267 * A contributed module containing a partial implementation of the
2268 simple-streams interface has been included. (thanks to Rudi
2270 * A contributed module implementing the RFC1321 Message Digest
2271 Algorithm, known as MD5, has been included.
2272 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2273 no longer has any effect, as the code controlled by this feature
2274 has been deleted. (As far as we know, no-one has ever built using
2275 this feature, and its semantics were confused in any case).
2276 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2277 work on conditions (as required by the ANSI specification),
2278 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2279 expected behaviour on conditions. Users should note, however,
2280 that such behaviour is not required by the ANSI specification,
2281 and so use of this behaviour may render their code unportable.
2282 * fixed some bugs revealed by Paul Dietz' test suite:
2283 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2285 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2286 ** SXHASH on condition objects no longer returns NIL.
2287 ** :ALLOCATION :CLASS slots are better treated; their values are
2288 updated on class redefinition, and initforms inherited from
2289 superclasses are applied.
2290 ** REMOVE-METHOD returns its generic function argument even when
2291 no method was removed.
2292 ** SHARED-INITIALIZE now initializes the values of the requested
2293 slots, including those with :ALLOCATION :CLASS.
2294 ** ALLOCATE-INSTANCE now works on structure classes defined via
2295 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2297 ** SLOT-EXISTS-P now works on conditions, as well as structures
2299 ** MAKE-LOAD-FORM now has the required methods on
2300 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2301 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2302 keyword, and now implements the SLOT-NAMES argument.
2303 ** methods with &OPTIONAL arguments no longer allow too many
2304 arguments to be passed in the call without error.
2305 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2306 option is consistent with the required arguments of the generic
2307 function lambda list.
2308 * bug fix: REQUIRE accepts a string designator. (Thanks to
2310 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2311 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2312 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2313 * bug fix: The compiler now checks for duplicated variables in macro
2315 * bug fix: SETQ on globals returns the correct value.
2316 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2317 not a COMPILER-ERROR (followed by some other strange error on
2318 choosing the CONTINUE restart).
2319 * bug fix: make.sh and friends are now more consistent in the way that
2320 they look for GNU "make".
2322 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2323 * experimental native threads support (on x86 Linux >=2.4 only).
2324 This is not compiled in by default: you need to add :SB-THREAD to
2325 the target features. See the "Beyond ANSI" chapter of the manual
2327 * fix for longstanding nonANSIism: The old distinction between
2328 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2329 The return value from CL:FIND-CLASS is now a CLOS class, and
2330 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2331 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2332 * An interface to the MetaObject Protocol, as described in Kiczales,
2333 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2334 MIT Press, 1991, is available from the SB-MOP package.
2335 * incompatible change: the SB-PCL package should now be considered
2336 a private implementation detail, and no longer a semi-private MOP
2338 * minor incompatible change: due to rearrangement for threads, the
2339 control stack and binding stack are now allocated at arbitrary
2340 addresses instead of being hardcoded per-port. Users affected by
2341 this probably have to be doing advanced things with shared
2342 libraries, and will know who they are.
2343 * minor incompatible change: Previously, all --eval forms used were
2344 processed with READ before any of them were processed with EVAL.
2345 Now each --eval form is processed with both READ and EVAL before
2346 the next --eval form is processed. (Thus package operations like
2347 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2348 work as the user might reasonably expect.)
2349 * minor incompatible change: *STANDARD-INPUT* is now only an
2350 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2352 * minor incompatible change: Y-OR-N-P is now character-oriented,
2353 not line oriented. Also, YES-OR-NO-P now works without errors.
2354 (thanks to Antonio Martinez)
2355 * sb-aclrepl module improvements: an integrated inspector, added
2356 repl features, and a bug fix to :trace command.
2357 * Known functions, which cannot be open coded by the backend, are
2358 considered to be able to check types of their arguments. (fixing
2359 a bug report by Nathan J. Froyd)
2360 * fixed a bug in computing method discriminating functions: It is
2361 now possible to define methods specialized on classes which have
2362 forward-referenced superclasses. (thanks to Gerd Moellmann)
2363 * fixed evaluation order in optional entries (reported by Gilbert
2365 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2366 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2367 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2368 required-by-AMOP NAME argument, as well as CLASS and
2369 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2370 * fixed bug 20: DEFMETHOD can define methods using names that are
2371 not the proper names of classes to designate class specializers.
2372 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2373 Unix file descriptors, instead of blowing up. (thanks to Antonio
2375 * Garbage collection refactoring: user-visible change is that a
2376 call to the GC function during WITHOUT-GCING will not do garbage
2377 collection until the end of the WITHOUT-GCING. If you were doing
2378 this you were probably losing anyway.
2379 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2380 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2381 TYPEP the latter but not the former.
2382 * The compiler issues a full WARNING for calls to undefined functions
2383 with names from the CL package.
2384 * MAP-INTO for a vector destination is open coded. (reported by
2385 Brian Downing on c.l.l)
2386 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2387 documentation string.
2388 * fixed some bugs revealed by Paul Dietz' test suite:
2389 ** COPY-ALIST now signals an error if its argument is a dotted
2391 ** Condition slots are now accessed more correctly in the presence
2392 of multiple initargs for a given slot.
2393 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2394 exclude restarts of the same name associated with a different
2396 ** DEFCLASS of forward-referenced classes with another
2397 forward-referenced class in the superclasses list no longer
2399 ** Condition slots are now initialized once each, not multiple
2400 times. (thanks to Gerd Moellmann)
2401 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2403 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2404 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2405 arguments contain duplicated elements.
2406 ** RESTART-CASE understands local macros.
2407 ** RESTART-CASE associates exactly its own restarts with a condition.
2408 ** ENDP in safe mode checks its argument to be of type LIST.
2409 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2410 in question is unbound.
2411 ** Optimized MAKE-INSTANCE functions no longer cause internal
2412 assertion failures in the presence of duplicate initargs.
2413 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2414 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2415 propagated even in the ctor optimized implementation of
2417 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2419 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2420 * a better implementation of SXHASH on (simple) bit vectors,
2421 measured both in execution speed and in distribution of results
2422 over the positive fixnums, has been installed. Likewise, a better
2423 implementation of EQUAL for simple bit vectors is now available.
2424 * fixed CEILING optimization for a divisor of form 2^k.
2425 * fixed bug 240 (emitting extra style warnings "using the lexical
2426 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2427 by Antonio Martinez)
2428 * fixed SXHASH, giving different results for NIL depending on type
2429 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2430 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2431 variable. (found by Rolf Wester)
2432 * SBCL does not ignore type declarations for special
2433 variables. (reported by rif on c.l.l 2003-03-05)
2434 * some bug fixes in contrib/sb-aclrepl/
2435 * fixed some bugs revealed by Paul Dietz' test suite:
2436 ** a bug in the CONS type specifier, whereby the CAR and CDR
2437 types got intertwined, has been fixed;
2438 ** the type system is now able to reason about the interaction
2439 between INTEGER and RATIO types more completely;
2440 ** APPEND, [N]REVERSE and NRECONC check that those their
2441 arguments, which must be proper lists, are really so;
2442 ** An array specialized to be unable to hold elements has been
2443 implemented, as required -- yes, really -- by ANSI;
2444 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2445 on malformed property lists;
2447 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2448 * incompatible packaging change: in line with Unix convention,
2449 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2450 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2451 * REQUIRE and PROVIDE are now optionally capable of doing something
2452 useful. See the documentation string for REQUIRE.
2453 * infrastructure for a managed SBCL contrib system: contributed
2454 modules in this release include:
2455 ** the ASDF system definition facility;
2456 ** an interface to the BSD Sockets API;
2457 ** an ACL-like convenience interface to the repl;
2458 (thanks to Kevin Rosenberg)
2459 ** an implementation of ROTATE-BYTE, with efficient implementation
2461 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2462 gives a closer approximation to the right answer than previously.
2463 (thanks to Raymond Toy)
2464 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2465 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2466 optional environment arguments, as required by ANSI.
2467 * fixed bugs in other functions taking environment objects, allowing
2468 calls with an explicit NIL environment argument to be compiled
2470 * fixed bug 228: primary return values from
2471 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2472 COMPILE or FUNCTION.
2473 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2474 :TYPE will no longer signal errors on innocuous objects.
2475 * fixed bug 231b: SETQ is better at respecting type declarations in
2476 the lexical environment.
2477 * fixed a bug in DEFCLASS: classes named by symbols with no or
2478 unprintable packages can now be defined.
2479 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2480 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2481 * fixed compiler failure related to checking types of functions.
2482 (reported by Robert E. Brown)
2483 * the compiler is now much more consistent in its error-checking
2484 treatment of bounding index arguments to sequence functions: in
2485 (SAFETY 3) code, errors will be signalled in almost all cases if
2486 invalid sequence bounding indices are passed to functions defined
2487 by ANSI to operate on sequences.
2488 * fixed a bug in the build procedure: documentation of SBCL-specific
2489 packages is now preserved and available in the final Lisp image.
2490 * lifted FDEFINITION lookup out of loops in the implementation of
2491 many list operations. (thanks to Robert E. Brown)
2492 * fixed a bug in the reader: the #n# reader macro now works for
2493 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2494 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2495 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2496 E. Brown and Tony Martinez respectively)
2497 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2498 SLOT-BOUNDP: in particular, it is now able to optimize them much
2499 better, and is now not vulnerable to having packages renamed.
2500 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2501 always called when a slot is not present in an instance. (thanks
2503 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2504 longer signal an error on unreading a character following EOF on
2505 the previous constituent stream. (thanks to Tony Martinez)
2506 * fixed some bugs revealed by Paul Dietz' test suite:
2507 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2508 not just nonnegative fixnums;
2509 ** the logical bit-array operators such as BIT-AND now accept an
2510 explicit NIL for their "opt-arg" argument (to indicate a
2511 freshly-consed result bit-array);
2512 ** ELT now signals an error on an invalid sequence index in safe
2514 ** the type system is now cleverer about negations of numeric
2515 types, and consequently understands the BIGNUM and RATIO types
2517 ** the type system is now cleverer about the interaction between
2518 INTEGER and RATIO types: while bugs still remain, many more
2519 cases are accurately computed;
2520 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2521 if it is in the last clause;
2522 ** CONSTANTLY now correctly returns a side-effect-free function in
2524 ** DECLARE is no longer treated as a special-operator; in
2525 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2526 * incremented fasl file version number due to the change in the
2527 DEFSTRUCT-SLOT-DESCRIPTION structure.
2529 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2530 * minor incompatible change: code processed by the "interpreter" or
2531 EVAL now has a compilation optimization policy of (DEBUG 2)
2532 (changed from (DEBUG 1)) to improve debuggability of interactive
2533 development, and to allow the use of the debug RETURN command in
2535 * an experimental implementation of the RETURN command for the
2536 debugger has been included. (thanks to Frederik Kuivinen)
2537 * fixed bug 62: constraints were not propagated into a loop.
2538 * fixed bug in embedded calls of SORT (reported and investigated by
2540 * fixed some bugs revealed by Paul Dietz' test suite:
2541 ** printing and reading of arrays with some dimensions having
2542 length 0 (thanks to Gerd Moellmann);
2543 ** BOA constructor with &AUX argument without a default value does
2544 not cause a type error;
2545 ** CONSTANTP now returns true for all self-evaluating objects.
2547 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2548 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2549 accessors that are related by inheritance, as specified in the
2550 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2551 Valtteri Vuorikoski)
2552 * The compiler is now able to inline functions that were defined in
2553 a complex lexical environment (e.g. inside a MACROLET).
2554 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2556 * Support for the upcoming FreeBSD-5.0 release has been included.
2557 (thanks to Dag-Erling Smorgrav)
2558 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2559 effect when it is not in a toplevel context.
2560 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2561 stand a better chance of being correct. (thanks to Gerd
2563 * fixed bug in COERCE, which now signals an error on coercing a
2564 rational to a bounded real type which excludes the expected
2566 * The compiler is now able to derive types more accurately from the
2567 COERCE and COMPILE functions.
2568 * fixed bug 223: functional binding is considered to be constant
2569 only for symbols in the CL package.
2570 * fixed bug 231: SETQ did not check the type of a variable being set
2571 (reported by Robert E. Brown)
2572 * A new optimization for MAKE-INSTANCE has been included, fixing
2573 various bugs (including relating to :ALLOCATION :CLASS slots and
2574 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2576 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2577 ** As required by ANSI, LOOP now disallows anonymous collection
2578 clauses such as COLLECT I in conjunction with aggregate boolean
2579 clauses such as THEREIS (= I 1);
2580 ** LOOP now signals an error when any variable is reused in the
2581 same loop (including the potentially useful construct analogous
2582 to WITH A = 1 WITH A = (1+ A);
2583 ** IT is only a special loop symbol within the first clause of a
2584 conditional loop clause;
2585 ** LOOP with a typed iteration variable over a hashtable now
2586 signals a type error iff it should.
2587 * fixed some other bugs revealed by Paul Dietz' test suite:
2588 ** FILE-STREAM now names the class previously known as FD-STREAM;
2589 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2590 argument) no longer signals an error;
2591 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2592 of the slot symbol, rather than using the current package
2593 ((:CONC-NAME "") continues to intern the slot's name in the
2595 * incremented fasl file version number, because of the incompatible
2596 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2597 of the new implementation of DEFINE-COMPILER-MACRO.
2599 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2600 * Support for building SBCL for MIPS platforms running in
2601 little-endian mode has now been checked in, and basic
2602 functionality on said platforms verified.
2603 * minor incompatible change: PCL now records the pathname of a file
2604 in which methods and the like are defined, rather than its
2606 * minor incompatible change: TRUENAME now considers the truename of
2607 a file naming a directory to be the pathname with :DIRECTORY
2608 component indicating that directory.
2609 * minor incompatible change: a NAMED clause in the extended form of
2610 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2611 reason for the previous behaviour is unclear.
2612 * more systematization and improvement of CLOS and MOP conformance
2613 in PCL (thanks to Gerd Moellman and Pierre Mai):
2614 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2616 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2617 primary methods with no specializers;
2618 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2620 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2621 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2622 FORWARD-REFERENCED-CLASSes; error reporting on
2623 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2625 ** SXHASH on CLOS instances now uses a slot internal to the
2626 instance to return different numbers on distinct instances,
2627 while preserving the same return value through invocations of
2629 ** DEFMETHOD signals errors when methods with longer incongruent
2630 lambda lists are added to generic functions;
2631 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2632 CLASS, as specified in AMOP;
2633 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2634 on the order returned by the primary method for classes of
2635 class STANDARD-CLASS;
2636 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2637 * fixed some bugs shown by Paul Dietz' test suite:
2638 ** DOLIST puts its body in TAGBODY;
2639 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2641 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2642 value producing form;
2643 ** if more variables are given to PROGV than values, extra
2644 variables are bound and made to have no value;
2645 ** NSUBSTITUTE on list arguments gets the right answer with
2647 ** ELT signals an error of type TYPE-ERROR when the index argument
2648 is not a valid sequence index;
2649 ** LOOP signals (at macroexpansion time) an error of type
2650 PROGRAM-ERROR when duplicate variable names are found;
2651 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2652 ** FUNCALL of special-operators now cause an error of type
2654 ** PSETQ now works as required in the presence of side-effecting
2655 symbol-macro places;
2656 ** NCONC accepts any object as its last argument;
2657 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2659 ** loop-for-as-package does not require a package to be explicitely
2661 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2663 * fixed bug 166: compiler preserves "there is a way to go"
2664 invariant when deleting code.
2665 * fixed bug 172: macro lambda lists with required arguments after
2666 &REST arguments now cause an error to be signalled. (thanks to
2668 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2669 bug. (thanks to Matthew Danish)
2670 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2672 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2673 arguments are lexically rebound. (thanks to Gerd Moellmann and
2675 * fixed bug 194: error messages are now more informative when there
2676 is no primary method applicable in a call to a generic
2677 function. (thanks to Gerd Moellmann)
2678 * fixed bug in command line argument checking (thanks to Julian
2680 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2681 (thanks to Matthew Danish)
2682 * incremented fasl file version number, because of the
2683 SXHASH-related changes in the layout of CLOS data structures
2685 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2686 * minor incompatible change: The runtime (the Unix executable named
2687 "sbcl") is now much pickier about the .core files it will load.
2688 Essentially it now requires .core files to descend from the same
2689 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2690 as the runtime does. (The intent is to prevent the crashes which
2691 can occur, and which can even be reported as mysterious failures,
2692 when people patch the sources or change the build parameters
2693 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2694 sbcl and .core files.)
2695 * fixed bug: VALUES-LIST is no longer optimized away.
2696 * fixed bug 142: The FFI conversion of C string values to Lisp
2697 string values no longer conses excessively. (thanks to Nathan
2698 Froyd porting Raymond Toy's fix to CMU CL)
2699 * began to systematize and improve MOP conformance in PCL (thanks to
2700 Nathan Froyd, Gerd Moellman and Pierre Mai):
2701 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2703 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2704 ** FINALIZE-INHERITANCE is now called on class finalization;
2705 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2706 argument precedence order.
2707 * fixed bug 202: The compiler no longer fails on functions whose
2708 derived types contradict their declared type.
2709 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2710 so it can be non-toplevel.
2711 * The fasl file version number has changed (because of the new
2712 implementation of DEFMACRO).
2713 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2714 safe code, that any length requirement by their type-specifier
2715 argument is valid. The exceptions to this are described in bug
2717 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2718 functions now check the types of their inputs as required by ANSI.
2719 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2720 introduced symbol is DECLAREd to be SPECIAL.
2721 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2722 * fixed bug: :COUNT argument to sequence functions may be negative
2723 * fixed bug: body of DO-SYMBOLS may contain declarations
2724 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2725 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2726 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2727 symbol macro only once
2728 * fixed printing of call frame when argument list is unavailable
2729 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2730 * compiler no longer signals WARNING on unknown keyword
2733 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2734 * A beta-quality port to the mips architecture running Linux,
2735 based on the old CMUCL backend, has been made. It has been tested
2736 on a big-endian kernel, and works sufficiently well to be able to
2737 rebuild itself; it has not been tested in little-endian mode.
2738 * fixed an inconsistency between gencgc.c and purify.c which made
2739 dumping/loading .core files unreliable
2740 * fixed bug 120a: The compiler now deals correctly with IFs where
2741 the consequent is the same as the alternative, instead of
2742 misderiving the return type. (thanks to Alexey Dejneka)
2743 * fixed bug 113: Logical pathnames are now dumpable (the logical
2744 host is resolved at load-time, throwing an error if it is not
2746 * fixed bug 174: FORMAT's error message is slightly clearer when a
2747 non-printing character is used in a format directive.
2748 * fixed several bugs in compiler checking of type declarations, i.e.
2749 violations of the Python "declarations are assertions" principle
2750 (thanks to Alexey Dejneka)
2751 * fixed several bugs in PCL's error checking (thanks to Gerd
2753 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2755 * fixed bug in compilation of functions as first class values
2756 (thanks to Antonio Martinez-Shotton)
2757 * The compiler's handling TYPE-ERRORs which it can prove will
2758 inevitably happen at runtime has been cleaned up and corrected
2759 in several ways. (thanks to Alexey Dejneka)
2760 * improved argument type checking for various basic arithmetic
2761 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2762 lavished on them in the past that they can be compiled in many
2763 ways in different special cases
2764 * fixed bug 181: compiler checks validity of user supplied type
2766 * cleaned up code flushing in optimization: Function calls which
2767 should signal errors for safety purposes (e.g. which ANSI says
2768 should signal errors when their arguments are of incorrect type)
2769 are no longer optimized away.
2770 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2771 * incremented fasl file version number, because changes in the
2772 implementation of sequence functions like COERCE caused
2773 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2774 in old inline expansions) to become undefined. (Actually these
2775 changes were later undone, so we might very well be binary
2776 compatible with 0.7.7 after all, but leaving the version number
2777 incremented seemed like the simplest and most conservative
2780 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2781 * An alpha-quality port to the parisc architecture running Linux,
2782 based on the old CMUCL backend, has been made. This, even more so
2783 than the other backends, should be considered still a work in
2784 progress; known problems include that the Linux kernel in 64-bit
2785 mode does not propagate the correct sigcontext structure to
2786 userspace, and consequently SBCL on a parisc64 kernel will not
2788 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2789 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2790 suppressed.) Also now that the compiler is looking at declarations
2791 in the environment, it checks optimization declarations as well,
2792 and suppresses inlining when (> DEBUG SPEED).
2793 * More fixes have been made to treatment of floating point exception
2794 treatment and other Unix signals. In particular, floating point
2795 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2796 * The detection and handling of control stack exhaustion (infinite
2797 or very deeply nested recursion) has changed. Stack exhaustion
2798 detection is now done by write-protecting pages at the OS level
2799 and applies at all optimization settings; when found, a
2800 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2801 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2802 be caught using IGNORE-ERRORS.
2803 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2804 correctly when transforming recursive LABELS functions to LETs.
2805 (thanks to Alexey Dejneka)
2806 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2807 that are names of constants or global variables.
2808 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2809 alien routines with docstrings.
2810 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2811 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2813 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2814 (thanks to Eric Marsden)
2815 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2816 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2817 * Bug fix: The SPARC backend can now compile functions involving
2818 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2819 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2820 to a functions expecting a PATHNAME-DESIGNATOR.
2821 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2822 lambda lists. (thanks to Alexey Dejneka)
2823 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2824 functions so that EXPORTing the name of the class doesn't cause
2825 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2826 to Antonio Martinez for reporting this)
2827 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2828 constant-folded without failing an assertion. (thanks to Einar
2829 Floystad Dorum for reporting this)
2830 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2831 types are now dealt with more correctly by the compiler.
2832 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2833 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2834 behaviour analogous to RENAME-FILE. This puts its behaviour more
2835 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2837 * The fasl file version number has changed again. (because of the
2838 bug fix involving the names of PCL MAKE-INSTANCE functions)
2840 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2841 * bug fix: Floating point exceptions are treated much more
2842 consistently on the x86/Linux and PPC/Linux platforms.
2843 * Array initialization with :INITIAL-ELEMENT is now much faster for
2844 cases when the compiler cannot open code the array creation, but
2845 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2846 array accesses have also seen a speed increase.
2847 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2848 specified. (This was at the root of some bad interactions between
2849 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2851 * bug fix: Internal error arguments for undefined functions are now
2852 computed correctly on the PPC/Linux platform.
2853 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2854 Raymond Toy's patch for CMU CL)
2855 * Support for the Solaris 9 operating environment has been included
2856 (thanks to Daniel Merritt)
2857 * A very ugly but hopefully complete draft of the missing FFI chapter
2858 of the manual has been created by reformatting the corresponding
2859 CMU CL manual chapter into (currently very ugly and incoherent)
2860 DocBook and bringing it up to date for SBCL behavior. Thus, the
2861 manual is now essentially complete, at least by my extreme
2862 once-and-only-once standards, whereby it's acceptable to refer to
2863 the doc strings of SB-EXT functions as the primary documentation.
2864 * The fasl file version number has changed again, due to cleanup of
2865 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2866 is no longer a static symbol.)
2868 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2869 * SBCL now builds with OpenMCL (version 0.12) as the
2870 cross-compilation host; also, more progress has been made toward
2871 bootstrapping under CLISP.
2872 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2874 * bug 158 fixed: The compiler can now deal with integer loop
2875 increments different from 1; fixing this turned out also to fix
2877 * bug 169 fixed: no more bogus warnings about using lexical bindings
2878 despite the presence of perfectly good SPECIAL declarations (thanks
2879 to David Lichteblau)
2880 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2881 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2882 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2884 * bug 180 fixed: Method combination specifications no longer ignore
2885 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2886 * bug fix: Structure type predicate functions now check their argument
2887 count as they should.
2888 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2889 correctly. (thanks to Pierre Mai)
2890 * minor incompatible change: The --noprogrammer option is deprecated
2891 in favor of the new --disable-debugger option, which is very similar.
2892 (The major difference is that it takes effect at a slightly different
2893 time at startup, causing handling of errors in --sysinit and
2894 --userinit files will be affected differently.) The
2895 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2896 been added to allow this functionality to be controlled from ordinary
2897 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2898 maintainers, who might want to run non-interactive scripts to
2899 build SBCL cores which will later be used interactively.)
2900 * minor incompatible change: The LOAD function no longer, when given
2901 a wild pathname to load, loads all files matching that pathname.
2902 Instead, an error of type FILE-ERROR is signalled.
2904 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2905 * bug 147 fixed: The compiler preserves its block link/count
2906 invariants more correctly now so that it doesn't crash. (thanks
2908 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2910 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2911 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2912 no-value-for-this-character as NIL (as specified by ANSI).
2913 * HOST-NAMESTRING on physical pathnames now returns a string that is
2914 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2915 (thanks to Christophe Rhodes)
2916 * The Alpha port handles icache flushing more correctly. (thanks to
2918 * More progress has been made toward bootstrapping under CLISP. (thanks
2919 to Christophe Rhodes)
2920 * The fasl file format has changed again, because dynamic loading
2921 on OpenBSD (which has non-ELF object files) motivated some cleanups
2922 in the way that foreign symbols are transformed and passed around.
2923 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2924 is no longer treated as whitespace by the reader, but instead as
2925 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2926 |A
\7fB|, instead of A as it used to.
2928 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2929 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2930 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2931 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2932 this as of 0.7.1.45, but was left out of the previous news section
2933 (thanks to Dan Barlow)
2934 * SBCL now runs on the Solaris operating system on SPARC architectures
2935 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2936 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2937 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2938 * SPARC backend cleanups, allowing builds of cores optimized for V8
2939 and V9 SPARCS, and also emission of code targeted to a particular
2940 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2942 * SBCL is closer to bootstrapping under CLISP, thanks to various
2943 fixes by Christophe Rhodes.
2944 * The fasl file format has changed again, to allow the compiler's
2945 INFO database to support symbol macros.
2946 * The user manual (in doc/) is formatted into HTML more nicely.
2947 (thanks to coreythomas)
2948 * The system is smarter about SUBTYPEP relationships, especially
2949 those involving NOT types (including types such as ATOM which are
2950 represented internally using NOT types). Thus SUBTYPEP is less
2951 likely to return (VALUES NIL NIL) in general, and in particular
2952 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2954 * The fasl file format has changed again, because the internal
2955 representation of types now includes a new slot to support the new
2956 SUBTYPEP-of-NOT-types logic.
2957 * (not a change in the main branch of SBCL, but a related prototype
2958 which can hopefully be merged into the main branch of SBCL in the
2959 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2960 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2961 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2962 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2963 dispatch macro character combinations. (thanks to Alexey Dejneka)
2964 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2965 that superficially look like logical namestrings correctly.
2966 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2968 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2969 default supplied-p) for &optional and &key arguments. (thanks to
2972 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2973 * incompatible change: The compiler is now less aggressive about
2974 tail call optimization, doing it only when (> SPACE DEBUG) or
2975 (> SPEED DEBUG). (This is an incompatible change because there are
2976 programs which relied on the old CMU-CL-style behavior to optimize
2977 away their unbounded recursion which will now die of stack overflow.)
2978 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2979 for non-GENCGC systems has been increased to 20M (since that
2980 seems much closer to the likely performance optimum for modern
2981 systems than the old 4M value was)
2982 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2983 and *DEBUG-PRINT-LEVEL*
2984 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2985 of CMU CL's support for SPARC, and various endianness and other
2986 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2987 * new syntactic sugar for the Unix command line: --load foo.bar is now
2988 an alternate notation for --eval '(load "foo.bar")'.
2990 ** The system now detects stack overflow and handles it gracefully,
2991 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2992 optimization settings. (This is a good thing in general, and
2993 its introduction in this version should be particularly timely
2994 for anyone whose code fails because of suppression of tail
2996 ** The system now hunts for the C variable "environ" in a more
2997 devious way, to avoid segfaults when the C library version differs
2998 between compile time and run time. (thanks to Christophe Rhodes)
2999 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3000 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3001 ** The compiler no longer issues bogus style warnings for undefined
3002 classes in the same source file as the DEFCLASSes which defined
3003 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3005 ** fixes in CONDITION class precedence list for undefined function
3006 errors (thanks to Alexei Dejneka)
3007 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3008 correctly. (thanks to Dan Barlow)
3009 ** portability fixes aiming at bootstrapping under CLISP (thanks
3010 to Dave McDonald and Christophe Rhodes)
3011 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3012 ** fixes in type translation and and type inference (thanks to
3014 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3015 ** various fixes in the new ports (thanks to Dan Barlow)
3016 * several changes related to debugging:
3017 ** suppression of tail recursion, as noted above
3018 ** stack overflow detection, as noted above
3019 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3020 is now the default. (For some time encapsulation has been more
3021 reliable than the breakpoint-based :ENCAPSULATE NIL
3022 implementation, at least on X86 systems; and I just noticed that
3023 encapsulation also seems closer to the spirit of the ANSI
3026 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3028 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3029 up properly again. (There was a packaging bug in 0.7.0 which
3030 left their definitions in SB-SYS::LOAD-FOREIGN and
3031 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3032 vital for most things which interface to C-level interfaces,
3033 like extensions working with sockets or databases or
3034 Perl-compatible regexes or whatever, and the need to fix
3035 this bug is the main reason that 0.7.1 was released so
3037 ** DEFGENERIC is now choosier about the methods it redefines, so that
3038 reLOADing a previously-LOADed file containing DEFGENERICs does
3039 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3040 cycle now works as it should. (thanks to Alexey Dejneka)
3041 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3042 fixed by Christophe Rhodes. (He actually submitted this patch
3043 months ago, and I delayed until after 0.7.0.)
3044 ** Bug 111 (internal compiler confusion about runtime checks on
3045 FUNCTION types) was fixed by Alexey Dejneka.
3046 * Some internal cleanups (getting rid of variables which aren't
3047 needed now that the byte interpreter is gone) caused the fasl
3048 file format number to change again.
3050 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3051 * major incompatible change: The default fasl file extension, i.e. the
3052 default extension for files produced by COMPILE-FILE, has changed
3053 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3055 ** There are many changes in the implementation of the compiler.
3056 SBCL is now essentially a compiler-only implementation of ANSI
3057 Common Lisp. EVAL still "interprets" a few special cases, but
3058 almost all the interesting cases are handled by creating
3059 a LAMBDA expression, calling COMPILE on it, then calling
3060 FUNCALL on the result.
3061 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3062 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3063 Since the code is newer, there might still be some new bugs
3064 (though not as many as before Martin Atzmueller's fixes:-). But
3065 the new code is substantially simpler and clearer, and hopefully
3066 any remaining bugs will be simpler, less fundamental, and more
3067 fixable then the bugs in the old code.
3068 ** The revised compiler is still a little unsteady on its feet.
3070 *** The debugging information it produces (particularly the names
3071 of FUNCTION objects) is sometimes much less useful than what
3072 the old compiler produced.
3073 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3074 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3075 within a MACROLET) has been temporarily weakened.
3076 ** There are new compiler optimizations for various functions:
3077 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3078 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3079 *** the math functions TRUNCATE, FLOOR, and CEILING
3080 *** the function-of-all-trades COERCE
3081 Mostly these should be transparent, but there's one
3082 potentially-annoying problem (bug 117): when the compiler
3083 inline-expands a function and does type analysis on the result,
3084 it can create control paths which have type mismatches, and
3085 when it can't prove that those control paths aren't taken,
3086 it will issue WARNINGs about the type mismatches. This is
3087 a particular problem in practice for the new sequence functions.
3088 It's not clear how this should be fixed, and for now, a
3089 workaround is given in the entry for 117 in the BUGS file.
3090 ** (Because of the interaction between the two previous items --
3091 occasional inlining problems and new inline expansions -- some
3092 of the new sequence function optimizations won't really kick in
3093 completely until debugging information, and then inlining, are
3094 straightened out in some future version.)
3095 * minor incompatible changes:
3096 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3097 DIRECTORY no longer implicitly promotes NIL slots of its
3098 pathname argument to :WILD. In particular, when you ask for the
3099 contents of a directory (which you used to be able to do without
3100 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3101 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3102 ** changes in behavior that ANSI explicitly defines to be
3103 implementation dependent:
3104 *** The new compiler-only implementation still conforms with ANSI,
3105 but acts a little different than before. Besides the obvious
3106 changes in performance tradeoffs (that the cost per form passed
3107 to EVAL has gone up, and the cost per form executed by EVAL
3108 has gone down), the behavior of the system changes a little
3109 because there are no longer any interpreted function objects.
3110 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3111 e.g. doing COMPILE on the output of interactive DEFUN is
3113 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3115 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3116 has changed from (:CL) to NIL.
3117 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3118 e.g. #\Newline and #\Tab, don't have names specified in the
3119 ANSI Common Lisp standard, is now based on their ASCII symbolic
3120 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3121 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3122 are no longer used for output.
3123 ** changes in internal implementation constants:
3124 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3125 4 million. (If your application spends a lot of time GCing and
3126 you have a lot of RAM, you might want to experiment with
3127 increasing it even more.)
3128 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3129 However, almost all old code should still continue to work without
3130 immediate update, as SB-C-CALL is now a (deprecated) nickname
3132 ** Old operator names in the style DEF-FOO are now deprecated in
3133 favor of new corresponding names DEFINE-FOO, for consistency with
3134 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3135 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3136 internal symbols, but a few supported extensions like
3137 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3138 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3139 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3140 etc. as you get deeper into recursive calls to the debugger
3141 command loop, instead of the old "5]", "5]]", "5]]]"
3142 sequence. (I was motivated to do this when squabbles between
3143 ILISP and SBCL left me very deeply nested in the debugger. In the
3144 short term, this change will probably provoke more ILISP/SBCL
3145 squabbles, but hopefully it will be an improvement in the long run.)
3146 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3147 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3148 function, since declaiming return types involving aliens is
3149 (1) annoyingly messy to do by hand and (2) vital to efficient
3150 compilation of code which calls such functions.
3151 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3152 longer reexported by the SB-EXT package. They're solely useful
3153 for alien code, so it seems more logical that you should get
3154 them from the SB-ALIEN package, not in SB-EXT.
3155 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3156 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3157 features. Instead, the code that they used to control is always
3158 built into the system.
3159 * many other bug fixes
3160 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3161 advantage of the new EVAL-WHEN stuff and to clean them up in
3162 general, and they are now more ANSI-compliant in a number of
3163 ways. Martin Atzmueller is responsible for a lot of this.
3164 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3166 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3167 *** correct ERROR type for various file operations
3168 *** some fixes for Lisp streams
3169 *** DEFMETHOD syntax checking
3170 *** changing old weird representation of debug information as
3171 strings (which, among their other deficiencies, don't transform
3172 correctly when you rename packages, and don't change their
3173 print representation when you change things like *PACKAGE*
3174 and *PRINT-LENGTH*) to symbols and lists of symbols
3175 He also made several improvements and fixed several bugs in DESCRIBE.
3176 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3178 *** misbehavior of WRITE-STRING/WRITE-LINE
3179 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3180 and several other LOOP problems as well
3181 *** DIRECTORY when similar filenames are present
3182 *** DEFGENERIC with :METHOD options
3183 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3184 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3185 *** argument ordering in FIND with :TEST option
3186 *** mishandled package designator argument in APROPOS-LIST
3187 *** various problems in the backquote readmacro
3188 *** a bug in APROPOS
3189 *** probably some others that I'm not describing very well here,
3190 since the CVS log documents them by reference to sbcl-devel
3191 messages, and the SourceForge archives aren't working well.:-(
3192 ** Dan Barlow improved the Alpha port (and is making progress on the
3193 PPC port, for those of you who think different).
3194 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3195 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3196 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3197 continues to work on the SPARC port (for those of you in a position
3198 to look down upon our little PC-compatible boxes from a great height).
3199 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3200 and uses that copy, rather than the current dynamic value, when
3201 it's trying to decide whether to truncate output. Thus e.g.
3202 (let ((*print-lines* 50))
3203 (pprint-logical-block (stream nil)
3205 (let ((*print-lines* 8))
3206 (print (aref possiblybigthings i) stream)))))
3207 should now truncate the logical block only at 50 lines, instead of
3208 often truncating it at 8 lines, as it did before.
3209 * The doc/cmucl/ directory, containing old CMU CL documentation
3210 from the time of the fork, is no longer part of the base system.
3211 SourceForge has shut down its anonymous FTP service, and with it
3212 my original plan for distributing the old CMU CL documentation
3213 there. For now, if you need these files you can download an old
3214 SBCL source release and extract them from it.
3215 * The fasl file version number changed again, for dozens of reasons,
3216 some of which are apparent above.
3218 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3219 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3220 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3221 patch, so that DISASSEMBLE output is much nicer.
3222 * The code in the SB-PROFILE package now seems reasonably stable.
3223 I still haven't decided what the final interface should look like
3224 (I'd like PROFILE to interact cleanly with TRACE, since both
3225 facilities use function encapsulation) but if you have a need
3226 for profiling now, you can probably use it successfully with
3227 the current CMU-CL-style interface.
3228 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3229 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3230 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3231 link now returns the dangling link itself, and for similar
3232 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3233 link itself. (In these cases the old code signalled an error and
3234 looped endlessly, respectively.) Thus, DIRECTORY now works even
3235 in the presence of dangling and cyclic symbolic links.
3236 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3237 is now a supported extension again, since the consensus on
3238 sbcl-devel was that it can be useful for ordinary development
3239 work, not just for debugging SBCL itself.
3240 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3241 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3242 that currently-defined functions might be redefined later with
3243 different return types.
3244 * Hash tables can be printed readably, as inspired by CMU CL code
3245 of Eric Marsden and SBCL code of Martin Atzmueller.
3246 * better error handling in CLOS method combination, thanks to
3247 Martin Atzmueller porting Pierre Mai's CMU CL patches
3248 * more overflow fixes for >16Mbyte I/O buffers
3249 * A bug in READ has been fixed, so that now a single Ctrl-D
3250 character suffices to cause end-of-file on character streams.
3251 In particular, now you only need one Ctrl-D at the command
3252 line (not two) to exit SBCL.
3253 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3255 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3256 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3257 does the right thing.
3258 * The implementation of some type tests, especially for CONDITION
3259 types, is now tidier and maybe faster, due to CMU CL code
3260 originally by Douglas Crosher, ported by Martin Atzmueller.
3261 * Some math functions have been fixed, and there are new
3262 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3263 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3264 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3265 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3266 likely to be built into sbcl-0.7.0.)
3267 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3268 CL, ported to SBCL by Martin Atzmueller)
3269 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3270 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3271 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3272 * There's a new slam.sh hack to shorten the edit/compile/debug
3273 cycle for low-level changes to SBCL itself, and a new
3274 :SB-AFTER-XC-CORE target feature to control the generation of
3275 the after-xc.core file needed by slam.sh.
3276 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3277 COMPILE-FILE is no longer supported, so that now every function
3278 gets an entry point, so that block compilation looks a little
3279 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3280 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3281 returns the number of bytes consed since the system started,
3282 rather than the number consed since the first time the function
3283 was called. (The new definition parallels ANSI functions like
3284 CL:GET-INTERNAL-RUN-TIME.)
3285 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3286 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3287 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3288 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3289 * Fasl file version numbers are now independent of the target CPU,
3290 since historically most system changes which required version
3291 number changes have affected all CPUs equally. Similarly,
3292 the byte fasl file version is now equal to the ordinary
3295 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3296 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3297 is no longer recognized. I apologize for this, because it was
3298 listed in SB-EXT as a supported extension, but I found that
3299 its existing behavior was poorly specified, as well as incorrectly
3300 specified, and it looked like too much of a mess to straighten it
3301 out. I have enough on my hands trying to get ANSI stuff to work..
3302 * many patches ported from CMU CL by Martin Atzmueller, with
3303 half a dozen bug fixes in pretty-printing and the debugger, and
3304 half a dozen others elsewhere
3305 * fixed bug 13: Floating point infinities are now supported again.
3306 They might still be a little bit flaky, but thanks to bug reports
3307 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3308 as flaky as they were.
3309 * The --noprogrammer command line option is now supported. (Its
3310 behavior is slightly different in detail from what the old man
3311 page claimed it would do, but it's still appropriate under the
3312 same circumstances that the man page talks about.)
3313 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3314 are now supported, and enabled by default. Thus, the compiler can
3315 handle many floating point and complex operations much less
3316 inefficiently. (Thus e.g. you can implement a complex FFT
3318 * The compiler now detects type mismatches between DECLAIM FTYPE
3319 and DEFUN better, and implements CHECK-TYPE more correctly, and
3320 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3321 thanks to patches from Martin Atzmueller.
3322 * various fixes to make the cross-compiler more portable to
3323 ANSI-conforming-but-different cross-compilation hosts (notably
3324 Lispworks for Windows, following bug reports from Arthur Lemmens)
3325 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3326 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3327 fixed thanks to Pierre Mai's CMU CL patches.
3328 * fixes to keep the system from overflowing internal counters when
3329 it tries to use i/o buffers larger than 16M bytes
3330 * fixed bug 45a: Various internal functions required to support
3331 complex special functions have been merged from CMU CL sources.
3332 (When I was first setting up SBCL, I misunderstood a compile-time
3333 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3334 * improved support for type intersection and union, fixing bug 12
3335 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3336 more obscure bugs as well
3337 * some steps toward byte-compiling non-performance-critical
3338 parts of the system, courtesy of patches from Martin Atzmueller
3339 * Christophe Rhodes has made some debian packages of sbcl at
3340 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3341 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3342 stable, but are nonetheless usable. When he's ready, I'd be happy
3343 to add them to the SourceForge "File Releases" section. (And if
3344 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3345 * new fasl file format version number (because of changes in
3346 internal representation of (OR ..) types to accommodate the new
3347 support for (AND ..) types, among other things)
3349 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3350 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3352 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3354 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3355 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3356 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3357 * DESCRIBE now works on structure objects again.
3358 * Most function call argument type mismatches are now handled as
3359 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3360 know whether the function will be redefined before the call is
3361 executed. (The compiler could flag local calls with full WARNINGs,
3362 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3363 it doesn't keep track of enough information to know whether calls
3364 are local in this sense.)
3365 * Compiler output is now more verbose, with messages truncated
3366 later than before. (There should be some supported way for users
3367 to override the default verbosity, but I haven't decided how to
3368 provide it yet, so this behavior is still controlled by the internal
3369 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3370 src/compiler/ir1util.lisp.)
3371 * Fasl file format version numbers have increased again, because
3372 support for the Gray streams extension changes the layout of the
3373 system's STREAM objects.
3374 * The Gray subclassable streams extension now works, thanks to a
3375 patch from Martin Atzmueller.
3376 * The full LOAD-FOREIGN extension (not just the primitive
3377 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3378 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3379 but like most other programs, it defaults to copying the Unix
3380 environment from the original process instead of starting the
3381 new process in an empty environment.
3382 * Extensions which manipulate the Unix environment now support
3383 an :ENVIRONMENT keyword option which doesn't smash case or
3384 do other bad things. The CMU-CL-style :ENV option is retained
3385 for porting convenience.
3386 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3387 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3389 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3391 * A patch from Martin Atzmueller seems to have solved the SIGINT
3392 problem, and as far as we know, signal-handling now works cleanly.
3393 (If you find any new bugs, please report them!)
3394 * The system no longer defaults Lisp source file names to types
3395 ".l", ".cl", or ".lsp", but only to ".lisp".
3396 * The compiler no longer uses special default file extensions for
3397 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3398 seems to expect a single default extension for all compiled code,
3399 and there's no compelling reason to try to stretch the standard
3400 to allow two different extensions.) Instead, byte-compiled files
3401 default to the same extension as native-compiled files.
3402 * Fasl file format version numbers have increased again, because
3403 a rearrangement of internal implementation packages made some
3404 dumped symbols in old fasl files unreadable in new cores.
3405 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3406 many fewer weird special cases.
3407 * Bug #17 (differing COMPILE-FILE behavior between logical and
3408 physical pathnames) has been fixed, and some related misbehavior too,
3409 thanks to a patch from Martin Atzmueller.
3410 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3411 by Tim Moore, ported to SBCL by Martin Atzmueller.
3412 * Martin Atzmueller fixed several filesystem-related problems,
3413 including bug #36, in part by porting CMU CL patches, which were
3414 written in part by Paul Werkowski.
3415 * More compiler warnings in src/runtime/ are gone, thanks to
3416 more patches from Martin Atzmueller.
3417 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3420 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3422 * DESCRIBE now works on CONDITION objects.
3423 * The debugger now handles errors which arise when trying to print
3424 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3426 * The build system now uses an additional file, customize-target-features.lisp,
3427 to allow local modifications to the target *FEATURES* list. (The point of
3428 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3429 debugging features enabled, without having to worry about propagating it
3430 into everyone's system when I do a "cvs update".) When no
3431 customize-target-features.lisp file exists, the target *FEATURES* list
3432 should be constructed the same way as before.
3433 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3434 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3435 warn about behavior which is undefined under the ANSI spec. Note: This
3436 is specified by ANSI, but it's not very popular with programmers.
3437 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3438 macro in the SBCL sources for an example of a workaround which you
3439 might use to make portable ANSI-standard code which does what you want.
3440 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3441 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3442 when it's used as a non-toplevel form. (This is required in order
3443 to implement the DEFCONSTANT-EQX macro.)
3444 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3445 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3446 funny things, due to bugs in the implementation of EVAL-WHEN
3447 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3448 non-toplevel DEFCONSTANTs are uncommon.)
3449 * The core file version number and fasl file version number have been
3450 incremented, because the old noncompliant DEFCONSTANT behavior involved
3451 calling functions which no longer exist, and because I also took the
3452 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3453 * fixed bug 1 (error handling before read-eval-print loop starts), and
3454 redid debugger restarts and related debugger commands somewhat while
3456 ** The QUIT debugger command is gone, since it did something
3457 rather different than the SB-EXT:QUIT command, and since it never
3458 worked properly outside the main toplevel read/eval/print loop.
3459 Invoking the new TOPLEVEL restart provides the same functionality.
3460 ** The GO debugger command is also gone, since you can just invoke
3461 the CONTINUE restart directly instead.
3462 ** The TOP debugger command is also gone, since it's redundant with the
3463 FRAME 0 command, and since it interfered with abbreviations for the
3465 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3467 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3468 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3469 some porting work by Martin Atzmueller)
3470 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3471 with a boatload of patches which he ported to SBCL. Now that those
3473 ** The system tries to make sure that its low-priority messages
3474 are prefixed by semicolons, to help people who like to use
3475 syntax highlighting in their ILISP buffer. (This patch
3476 was originally due to Raymond Toy.)
3477 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3478 patches originally written by Raymond Toy.
3479 ** The compiler understands coercion between single-value and
3480 multiple-VALUES type expressions better, getting rid of some very
3481 weird behavior, thanks to patches originally by Robert MacLachlan
3482 and Douglas Crosher.
3483 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3484 lambda lists, thanks to a patch originally by Pierre Mai.
3485 ** The system no longer bogusly warns about "abbreviated type
3487 ** The compiler gets less confused by inlining and RETURN-FROM,
3488 thanks to some patches originally by Tim Moore.
3489 ** The system no longer hangs when dumping circular lists to fasl
3490 files, thanks to a patch originally from Douglas Crosher.
3491 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3492 undefined function error.
3493 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3494 in structure slot initforms) for now, documented workaround instead:-|
3495 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3496 * fixed bug 5: added stubs for various Gray stream functions called
3497 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3498 installed, at least appropriate type errors are generated
3499 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3500 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3501 consistently in DEFMETHOD forms.
3502 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3503 it doesn't seem to affect SBCL after all
3504 * The C runtime system now builds with better optimization and many
3505 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3507 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3509 * The system is now under CVS at SourceForge (instead of the
3510 CVS repository on my home machine).
3511 * The new signal handling code has been tweaked to treat register
3512 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3513 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3514 downstream, e.g. in debug-int.lisp, has implicit dependencies
3515 on the unsignedness of integer representation of machine words,
3516 and that caused the system to bomb out with infinite regress
3517 when trying to recover from type errors involving signed values,
3518 e.g. (BUTLAST '(1 2 3) -1).)
3519 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3520 (This was one of the bugs Peter Van Eynde reported back in July.)
3521 * The system now uses code inspired by Colin Walters' O(N)
3522 implementation of MAP (from the cmucl-imp@cons.org mailing
3523 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3524 inline the MAP operation, and there is more than one
3525 sequence argument to the MAP call (so that it can't just
3526 do ETYPECASE once and for all based on the type of the
3527 single sequence argument). (The old non-inline implementation
3528 of the general M-argument sequence-of-length-N case required
3529 O(M*N*N) time when any of the sequence arguments were LISTs.)
3530 * The QUIT :UNIX-CODE keyword argument has been renamed to
3531 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3532 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3534 * Raymond Wiker's patches to port dynamic loading from Linux to
3535 FreeBSD have been added.
3536 * The BUGS file is now more nearly up to date, thanks in large part
3537 to Martin Atzmueller's review of it.
3538 * The debugger now flushes standard output streams before it begins
3539 its output ("debugger invoked" and so forth).
3540 * The core version number and fasl file version number have both
3541 been incremented, because of incompatible changes in the layout
3543 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3544 used to not be called for a saved Lisp image.)
3545 * Martin Atzmueller's version of a patch to fix a compiler crash,
3546 as posted on sbcl-devel 13 September 2000, has been installed.
3547 * Instead of installing Martin Atzmueller's patch for the
3548 compiler transform for SUBSEQ, I deleted the compiler transform,
3549 and transforms for some similar consing operations.
3550 * A bug in signal handling which kept TRACE from working on OpenBSD
3552 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3553 away by constant folding
3554 * The system now defines its address space constants in one place
3555 (in the Lisp sources), and propagates them automatically elsewhere
3556 (through GENESIS and the sbcl.h file). Therefore, patching the
3557 address map is less unnecessarily tedious and error-prone. The
3558 Lisp names of address space constants have also been systematized.
3559 * CVS tags like dollar-Header-dollar have been removed from
3560 the sources, because they have never saved me trouble and
3561 they've been source of trouble working with patches and other
3562 diff-related operations.
3563 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3564 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3566 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3568 * The system has been ported to OpenBSD.
3569 * The system now compiles with a simple "sh make.sh" on the systems
3570 that it's supported on. I.e., now you no longer need to tweak
3571 text in the target-features.lisp-expr and symlinks in src/runtime/
3572 by hand, the make.sh takes care of it for you.
3573 * The system is no longer so grossly inefficient when compiling code
3574 involving vectors implemented as general (not simple) vectors (VECTOR T),
3575 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3576 which dares to use the various sequence functions on non-simple
3577 vectors, takes less of a performance hit.
3578 * There is now a primitive type predicate VECTOR-T-P
3579 to test for the (VECTOR T) type, so that e.g.
3580 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3581 can now be compiled with some semblance of efficiency. (The old code
3582 turned the type declaration into a full call to %TYPEP at runtime!)
3583 * AREF on (VECTOR T) is still not fast, since it's still compiled
3584 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3585 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3586 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3587 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3588 without making full calls to SUBTYPEP at runtime.
3589 (Some analogous efficiency issues for non-simple vectors specialized to
3590 element types other than T, or for non-simple multidimensional arrays,
3591 have not been addressed. They could almost certainly be handled the
3592 same way if anyone is motivated to do so.)
3593 * The changes in array handling break binary compatibility, so
3594 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3595 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3596 instead of (VALUES T T).
3597 * By following the instructions that Dan Barlow posted to sbcl-devel
3598 on 2 July 2000, I was able to enable primitive dynamic object
3599 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3600 functionality is not implemented (since it calls ld to resolve
3601 library references automatically, requiring RUN-PROGRAM for its
3602 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3603 try to resolve library references) is now supported.
3604 * The system now flushes the standard output streams when it terminates,
3605 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3606 them at several other probably-convenient times, e.g. in each pass of
3607 the toplevel read-eval-print loop, and after evaluating a form given
3608 as an "--eval" command-line option. (These changes were motivated by a
3609 discussion of stream flushing issues on cmucl-imp in August 2000.)
3610 * The source transform for TYPEP of array types no longer assumes
3611 that an array whose element type is a not-yet-defined type
3612 is implemented as an array of T, but instead punts, so that the
3613 type will be interpreted at runtime.
3614 * There is now some support for cross-compiling in make.sh: each of
3615 the phases of make.sh has its own script. (This should be transparent
3616 to people doing ordinary, non-cross-compile builds.)
3617 * Since my laptop doesn't have hundreds of megabytes of memory like
3618 my desktop machine, I became more motivated to do some items on
3619 my to-do list in order to reduce the size of the system a little:
3620 ** Arrange for various needed-only-at-cold-init things to be
3621 uninterned after cold init. To support this, those things have
3622 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3623 symbols with such names are now uninterned after cold init).
3624 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3625 fasl files for cold load.
3626 ** Remove the old compiler structure pooling code (which used to
3627 be conditional on the target feature :SB-ALLOC) completely.
3628 ** Redo the representation of some data in cold init to be more compact.
3629 (I also looked into supporting byte compiled code at bootstrap time,
3630 which would probably reduce the size of the system a lot, but that
3631 looked too complicated, so I punted for now.)
3632 * The maximum signal nesting depth in the src/runtime/ support code has
3633 been reduced from 4096 to 256. (I don't know any reason for the very
3634 large old value. If the new smaller value turns out to break something,
3635 I'll probably just bump it back up.)
3636 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3638 * Many, many bugs reported by Peter Van Eynde have been added to
3639 the BUGS list; some have even been fixed.
3640 * While enabling dynamic object file loading, I tried to make the
3641 code easier to understand, renaming various functions and variables
3642 with less ambiguous names, and changing some function calling
3643 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3644 * While trying to figure out how to do the OpenBSD port, I tried to
3645 clean up some of the code in src/runtime/. In particular, I dropped
3646 support for non-POSIX signal handling, added various comments,
3647 tweaked the code to reduce the number of compilation warnings, and
3648 renamed some files to increase consistency.
3649 * To support the new automatic configuration functionality in make.sh,
3650 the source file target-features.lisp-expr has been replaced with the
3651 source file base-target-features.lisp-expr and the machine-generated
3652 file local-target-features.lisp-expr.
3653 * fixed a stupid quoting error in make.sh so that using CMU CL
3654 "lisp -batch" as cross-compilation host works again
3656 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3658 * DESCRIBE no longer tries to call itself recursively to describe
3659 bound/fbound values, so that it no longer fails on symbols which are
3660 bound to themselves (like keywords, T, and NIL).
3661 * DESCRIBE now works on generic functions.
3662 * The printer now prints less-screwed-up representations of closures
3663 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3664 * A private symbol is used instead of the :EMPTY keyword previously
3665 used to mark empty slots in hash tables. Thus
3666 (DEFVAR *HT* (MAKE-HASH-TABLE))
3667 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3668 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3669 now does what ANSI says that it should. (You can still get
3670 similar noncompliant behavior if bang on the hash table
3671 implementation with all the symbols you get back from
3672 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3673 This breaks binary compatibility, since tests for equality to
3674 :EMPTY are wired into things like the macroexpansion of
3675 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3677 * There's now a minimal placeholder implementation for CL:STEP,
3678 as required by ANSI.
3679 * An obscure bug in the interaction of the normal compiler, the byte
3680 compiler, inlining, and structure predicates has been patched
3681 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3682 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3684 * Missing ordinary arguments in a macro call are now detected even
3685 when the macro lambda list contains &KEY or &REST.
3686 * The debugger no longer complains about encountering the top of the
3687 stack when you type "FRAME 0" to explicitly instruct it to go to
3688 the top of the stack. And it now prints the frame you request even
3689 if it's the current frame (instead of saying "You are here.").
3690 * As specified by ANSI, the system now always prints keywords
3691 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3693 * The default initial SIZE of HASH-TABLEs is now smaller.
3694 * Type information from CLOS class dispatch is now propagated
3695 into DEFMETHOD bodies, so that e.g.
3696 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3698 is now basically equivalent to
3699 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3700 (DECLARE (TYPE SINGLE-FLOAT X))
3702 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3703 operation, without having to do run-time type dispatch.
3704 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3705 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3706 * Since I don't seem to be making big file reorganizations very often
3707 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3708 is overflowing my ability to conveniently back them up), I've finally
3709 checked the system into CVS. (The CVS repository is on my home system,
3710 not at SourceForge -- putting it on SourceForge might come later.)
3711 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3712 high-level GC-NOTIFY-FOO functions send their output. (There's
3713 still very little control of where low-level verbose GC functions
3714 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3715 less than it used to -- the GC-NOTIFY-FOO functions are now under
3716 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3717 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3718 in only one place in the source code, and propagates it automatically
3719 everywhere that it's needed. Thus e.g. when I bump the version from
3720 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3721 * The C source files now include boilerplate legalese and documentation
3722 at the head of each file (just as the Lisp source files already did).
3723 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3724 to his page will be replaced with a link to his new CLiki service.
3726 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3728 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3729 * The build process now looks for GNU make under the default name "gmake",
3730 instead of "make" as it used to. If GNU make is not available as "gmake"
3731 on your system, you can change this default behavior by setting the
3732 GNUMAKE environment variable.
3733 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3734 can build without error under CMU CL.
3736 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3738 * There is now a partial SBCL user manual (with some new text and some
3739 text cribbed from the CMU CL manual).
3740 * The beginnings of a profiler have been added (starting with the
3741 CMU CL profiler and simplifying and cleaning up). Eventually the
3742 main interface should be through the TRACE macro, but for now,
3743 it's still accessed through vaguely CMU-CL-style functions and macros
3744 exported from the package SB-PROFILE.
3745 * Some problems left over from porting CMU CL to the new
3746 cross-compilation bootstrap process have been cleaned up:
3747 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3748 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3749 part when getting rid of the compiler *BACKEND* stuff.)
3750 ** The value of *NULL-TYPE* was screwed up, because it was
3751 being initialized before the type system knew the final
3752 definition of the 'NULL type. This screwed up several key
3753 optimizations in the compiler, causing inefficiency in all sorts
3754 of places. (I found it because I wanted to understand why
3755 GET-INTERNAL-RUN-TIME was consing.)
3756 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3757 PROCLAIM FTYPE information. Unfortunately this broke binary
3758 compatibility again, since now the forms output by DEFGENERIC
3759 to refer to functions which didn't exist in 0.6.3.
3760 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3761 can use the new (as of 0.6.3) transform for SOME into MAP into
3763 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3764 values to (INTEGER 0 1000000), so that the time code will no longer
3765 occasionally get blown up by Linux returning 1000000 microseconds
3766 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3767 its output conform to the ANSI spec. (Alas, this makes its output
3768 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3769 * A full call to MAP NIL with a single sequence argument no longer conses.
3770 * fixes to problems pointed out by Martin Atzmueller:
3771 * The manual page no longer talks about multiprocessing as though
3772 it were currently supported.
3773 * The ILISP support patches have been removed from the distribution,
3774 because as of version 5.10.1, ILISP now supports SBCL without us
3775 having to maintain patches.
3776 * added a modified version of Raymond Toy's recent CMU CL patch for
3777 EQUALP comparison of HASH-TABLE
3779 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3781 * The system still can't cross-compile itself with
3782 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3783 entails), but at least it can compile more of itself that way
3784 than it used to be able to, and various buglets which were uncovered
3785 by trying to cross-compile itself that way have now been fixed.
3786 * This release breaks binary compatibility again. This time
3787 at least I've incremented the FASL file format version to 2, so that the
3788 problem can be detected reliably instead of just causing weird errors.
3789 * various new style warnings:
3790 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3791 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3792 ** using the lexical binding of a variable named in the *FOO* style
3793 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3794 as specified by ANSI.
3795 * *RANDOM-STATE* is no longer automatically initialized from
3796 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3797 default behavior of the system is to repeat its behavior every time
3798 it's run. If you'd like to change this behavior, you can always
3799 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3800 old convention there was no comparably easy way to get the system to
3801 repeat its behavior every time it was run.
3802 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3803 FTYPE declarations has been removed, in favor of their ANSI
3804 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3805 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3806 the types of their sequence arguments can be determined at compile time.
3807 This is done through a new open code expansion for MAP which eliminates
3808 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3809 argument types can be determined at compile time.
3810 * The optimizer now transforms COERCE into an identity operation when it
3811 can prove that the coerced object is already of the correct type. (This
3812 can be a win for machine generated code, including the output of other
3813 optimization transforms, such as the MAP transform above.)
3814 * Credit information has been moved from source file headers into CREDITS.
3815 * Source file headers have been made more standard.
3816 * The CASE macro now compiles without complaining even when it has
3819 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3821 * (Note that the way that the PCL macroexpansions were rewritten
3822 to accommodate the change in DEFGENERIC below breaks binary
3823 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3824 not run under sbcl-0.6.2. Once we get out of alpha releases,
3825 i.e. hit release 1.0.0, we'll probably try to maintain binary
3826 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3827 and sbcl-1.4.4. Until then, however, it might be fairly common
3828 for maintenance releases to break binary compatibility.)
3829 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3831 * The reporting of unhandled signals has been changed to print some
3832 explanatory text as well as the report form. (Previously only
3833 the report form was printed.)
3834 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3835 it defines, so that the compiler no longer issues undefined function
3836 warnings for compiled-but-not-yet-loaded generic functions.
3837 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3838 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3839 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3840 section of the ANSI spec are supported.
3841 * The "" nickname for the "KEYWORD" package has been removed.
3842 The reader still handles symbol tokens which begin with a package marker
3843 as keywords, but it doesn't expose its mechanism for doing so in the
3844 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3845 * The system now issues STYLE-WARNINGs for contradictory TYPE
3846 proclamations. (Warnings for contradictory FTYPE proclamations would
3847 be nice too, but those can't be done usefully unless the type system
3848 is made smarter about FUNCTION types.)
3849 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3850 systematized, so that "*target-*.lisp is supposed to exist only on the
3851 target and imply that there's a related file which exists on the
3852 host, and *host-*.lisp is supposed to exist only on the host and imply
3853 that there's a related file which exists on the target. This involves a
3854 lot of renaming. Hopefully the acute confusion caused by the renaming
3855 will be justified by the reduction in chronic confusion..
3856 ** runtime-type.lisp -> early-target-type.lisp
3857 ** target-type.lisp -> late-target-type.lisp
3858 ** early-host-format.lisp -> early-format.lisp
3859 ** late-host-format.lisp -> late-format.lisp
3860 ** host-error.lisp -> misc-error.lisp
3861 ** early-error.lisp -> early-target-error.lisp
3862 ** late-error.lisp -> late-target-error.lisp
3863 ** host-defboot.lisp -> early-defboot.lisp
3864 ** code/misc.lisp -> code/target-misc.lisp
3865 ** code/host-misc.lisp -> code/misc.lisp
3866 ** code/numbers.lisp -> code/target-numbers.lisp
3867 ** code/early-numbers.lisp -> numbers.lisp
3868 ** early-host-type.lisp -> early-type.lisp
3869 ** late-host-type.lisp -> late-type.lisp
3870 ** host-typep.lisp -> typep.lisp
3871 ** load.lisp -> target-load.lisp
3872 ** host-load.lisp -> load.lisp
3873 ** host-disassem.lisp -> disassem.lisp
3874 ** host-insts.lisp -> insts.lisp
3875 ** byte-comp.lisp -> target-byte-comp.lisp
3876 ** host-byte-comp.lisp -> byte-comp.lisp
3877 ** host-signal.lisp -> signal.lisp
3878 ** host-defstruct.lisp -> defstruct.lisp
3879 ** late-target-type.lisp -> deftypes-for-target.lisp
3880 Furthermore, several other previously target-only files foo.lisp (e.g.
3881 hash-table.lisp and random.lisp) have been split into a target-and-host
3882 foo.lisp file and a target-only target-foo.lisp file, with their key type
3883 definitions in the target-and-host part, so that the cross-compiler will
3884 know more about target types.
3885 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3886 gone away. In their place are various *BACKEND-FOO* variables
3887 corresponding to the slots of the old structure.
3888 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3889 into the target SBCL has been fixed.
3890 * The chill.lisp system for loading cold code into a running SBCL
3892 * Support for the CMU CL "scavenger hook" extension has been removed.
3893 (It was undocumented and unused in the CMU CL sources that SBCL was
3894 derived from, and stale in sbcl-0.6.1.)
3895 * Various errors in the cross-compiler type system were detected
3896 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3897 (enabling various consistency checks). Many of them were fixed,
3898 but some hard problems remain, so the compiler is back to
3899 running without *TYPE-SYSTEM-INITIALIZED* for now.
3900 * As part of the cross-compiler type system cleanup, I implemented
3901 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3902 * I have started adding UNCROSS calls throughout the type system
3903 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3904 on cross-compiler input files will be able to go away, and various
3906 * CONSTANTP now returns true for quoted forms (as explicitly required
3909 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3911 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3912 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3913 (MAKE-LIST -1). In the longer term, it ought to become true
3914 that declarations are assertions even at SAFETY 1. For now, it's not
3915 quite true even at SAFETY 3, but it's at least more nearly true..
3916 (Note that this change seems to increases the size of the system by
3917 O(5%) and to decrease the speed of the compiler by 20% or more.)
3918 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3919 for the problem of printing dozens of lines of distracting information
3920 about low-level system machinery as part of the top stack frame
3921 on entry to the debugger when an undefined function was called.
3922 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3923 is not reset to COMMON-LISP-USER.
3924 * Compilation of stuff related to dyncount.lisp has been made conditional
3925 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3926 smaller. The various dyncount-related symbols have been moved into
3927 a new "SB-DYNCOUNT" package.
3928 * tty-inspect.lisp has been renamed to inspect.lisp.
3929 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3930 feature has gone away. (When we eventually port to other flavors of
3931 libc and/or Unix, we'll try to make the differences between flavors
3932 invisible at the user level.)
3933 * Various other *FEATURES* tags, and/or their associated conditionals,
3934 have been removed if obsolescent, or given better documentation, or
3935 sometimes given more-mnemonic names.
3937 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3939 * tidied up "make.sh" script
3940 * tidied up system directory structure
3941 * better "clean.sh" behavior
3942 * added doc/FOR-CMUCL-DEVELOPERS
3943 * many many small tweaks to output format, e.g. removing possibly-confusing
3944 trailing #\. character in DESCRIBE-INSTANCE
3945 * (EQUALP #\A 'A) no longer signals an error.
3946 * new hashing code, including EQUALP hashing
3947 * tidied up Lisp initialization and toplevel
3948 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3949 * command line argument processing
3950 * added POSIX-GETENV function to deal with Unix-ish environment variables
3951 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3952 terminating SBCL on EOF
3953 * non-verbose GC by default
3954 * There is no more "sbcl" shell script; the sbcl file is now the C
3955 runtime executable (just like CMU CL).
3956 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3958 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3959 symbol and package information as Lisp native symbol and package objects
3960 instead of strings naming symbols and strings naming packages. This way,
3961 whenever packages are renamed (as in warm init), debug information is
3962 transformed along with everything else.
3963 * tweaked the optimization policy declarations which control the building
3964 of SBCL itself. Now, among other things, the system no longer saves
3965 source location debugging information. (This helps two problems at once
3966 by reducing SBCL size and by keeping SBCL from trying to look for its
3967 sources -- which may not exist -- when reporting errors.)
3968 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3969 debugging and testing purposes
3970 * cleaned up printing, making the printer call PRINT-OBJECT for
3971 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3972 methods, giving nearly-ANSI behavior
3973 * converted almost all special variables to use *FOO* naming convention
3974 * deleted PARSE-TIME functionality, since it can be done portably
3975 * moved some files out of cold init into warm init
3976 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3978 * regularized formatting of source files
3979 * added an install.sh script
3980 * fixed ridiculous memory usage of cross-compiler by making
3981 compiler/alloc.lisp not try to do pooling unless it can hook
3982 itself into the GC of the cross-compilation host. Now the system
3983 builds nicely on my old laptop.
3984 * added :SB-ALLOC in target-features.lisp-expr
3985 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3986 was not implemented)
3987 * re-did condition handling and note reporting in the compiler. Notes
3988 are no longer handled by signalling conditions. Style warnings
3989 and warnings are handled more correctly and reported in such a way
3990 that it's easy to find one or the other in your output (so that you
3991 can e.g. figure out which of many problems caused COMPILE-FILE to
3993 * changed the severity of several compiler warnings from full WARNING
3994 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3995 compiler note reporting so that it doesn't use the condition system
3996 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3997 COMPILE-FILE command)
3998 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3999 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4000 no longer supported, since their implementation was deeply intertwingled
4001 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4002 * removed bogus "support" for compiler macros named (SETF FOO), and
4003 removed the compiler macro for SETF INFO (but only after making a fool
4004 of myself on the cmucl-imp mailing list by posting a bogus patch for
4005 DEFINE-COMPILER-MACRO..)
4006 * Compiled files containing forms which have side effects on the Lisp
4007 reader (such as DEFPACKAGE forms) are now handled more correctly.
4008 (Compiler queuing of top level lambdas has been suppressed by setting
4009 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4010 * deleted various currently-unused source files, e.g. gengc.lisp. They
4011 may be added back at some point e.g. when porting to other architectures,
4012 but until they are it's distracting to distribute them and to try to
4014 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4015 there were so many of them they're just distractions, and UNCROSS is
4016 known to be able to handle the current sources
4017 * moved PROFILE functionality into TRACE, so that it will be clear
4018 how the wrapping and unwrapping of functions when you profile them
4019 interacts with the wrapping and unwrapping of functions when you
4020 trace them. (Actually, the functionality isn't there yet, but at least
4021 the interface specification is there. Hopefully, the functionality will
4022 arrive with some future maintenance release.)
4023 * removed host-oops.lisp
4024 * changed signature of QUIT function to allow UNIX-CODE argument
4025 * fixed READ-SEQUENCE bug
4026 * tweaked verbose GC output so that it looks more like the progress
4027 output that ANSI specifies for functions like LOAD
4028 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4029 * added <http://sbcl.sourceforge.com> to the banner information printed by