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 * bug fix: compiler bug triggered by a (non-standard) VALUES
8 declaration in a LET* was fixed. (reported by Kaersten Poeck)
9 * bug fix: file compiler no longer confuses validated and already
10 dumped structurres. (reported by Kaersten Poeck)
11 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
12 fill-pointer signals a type-error as required. (thanks to
14 * improvements to the Windows port:
15 ** floating point exceptions are now reported correctly.
16 ** stack exhaustion detection works partially.
17 ** more accurate GET-INTERNAL-REAL-TIME.
18 ** better RUN-PROGRAM behaviour in case of errors in the child
20 ** PROBE-FILE now simplifies pathnames correctly.
21 ** DIRECTORY now works correctly with :WILD-INFERIORS.
22 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
23 1st 1970 (although time timezone data will be unreliable).
25 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
26 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
27 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
29 * bug fix: two potential GC deadlocks affecting threaded builds.
30 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
31 returns T (reported by Anton Kazennikov)
32 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
33 with non-variable places
34 * bug fix: remove a race condition in the setting of
35 funcallable-instance functions, this should make threaded CLOS
36 code more stable against memory faults.
37 * bug fix: corruption of specials when unbinding is interrupted by an
38 asynchronous unwind (reported by Hannu Koivisto)
39 * improvement: the debugger will now also display local variables that
40 are only used once, for code compiled with a DEBUG optimization quality
43 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
44 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
45 * incompatible change: External-format support for FFI calls. The
46 SB-ALIEN:C-STRING no longer implies an ASCII
47 external-format. Instead the string is subject to external-format
48 conversion. Additionally return values of type SB-ALIEN:C-STRING
49 are by default of type (SIMPLE-ARRAY CHARACTER), not
50 SIMPLE-BASE-STRING. For an alien type that behaves like the old
51 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
52 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
53 * incompatible change: SB-EXT package no longer contains the
54 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
55 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
57 * incompatible change: the single-stepper is no longer available
58 on Alpha, Mips and Sparc platforms.
59 * minor incompatible change: the direct superclasses of
60 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
61 not (STANDARD-OBJECT FUNCTION). This makes the
62 class-precedence-lists of GENERIC-FUNCTION and
63 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
65 * minor incompatible change: the default stream external format on
66 non-unicode SBCL is no longer detected from the locale, but is always
68 * new feature: Add a version of evaluator that uses an interpreter instead
69 of the compiler. EVAL still uses the compiler by default, to switch it
70 to use the interpreter, set the value of the variable
71 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
72 * minor incompatible change: the single-stepper REPL has been merged
73 with the normal debugger (see the "Stepping" heading of the debugger help
74 for more details). The debugger command STEP will no longer switch
75 to the single-stepper REPL.
76 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
78 * bug fix: loading of default sysinit file works. (thanks to Leonid
80 * bug fix: better detection of circularities in the file-compiler.
81 (reported by Marco Monteiro)
82 * bug fix: the CL pathname functions now work with files that have
83 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
84 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
85 systems with Unix98 pty semantics.
86 * bug fix: ASDF-INSTALL will now work with bsd tar.
87 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
89 * bug fix: timers expiring in dead threads no longer cause a
90 type-error (reported by Paul "Nonny Mouse").
91 * bug fix: thanks to more lightweight single-stepper instrumentation,
92 code compiled with (DEBUG 3) will compile and execute significantly faster,
93 and will have more accurate type-inferencing than before
94 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
95 whose bindings are modified
96 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
97 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
98 ** replace dummy implementations of CL:MACHINE-INSTANCE and
99 CL:SOFTWARE-VERSION with proper version
101 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
102 * feature: implemented the READER-METHOD-CLASS and
103 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
104 as specified by AMOP.
105 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
107 * optimization: faster LOGCOUNT implementation on x86 and x86-64
108 (thanks to Lutz Euler)
109 * optimization: hashing of general arrays and vectors has been
110 improved. (reported by Any Fingerhut)
111 * enhancement: SB-INTROSPECT is now able to find definitions of
112 profiled functions. (thanks to Troels Henriksen)
113 * enhancement: compiler-macro expansion applies now to FUNCALL forms
114 as well, allowing compiler-macros for SETF-functions to expand.
115 * enhancement: step-instrumentation no longer wraps known
116 single-value functions with multiple-value context, allowing
117 better type inference.
118 * fixed bug #337: use of MAKE-METHOD in method combination now works
119 even in the presence of user-defined method classes. (reported by
120 Bruno Haible and Pascal Costanza)
121 * fixed bug #339(c): if there are applicable methods not part of any
122 long-form method-combination group, call INVALID-METHOD-ERROR.
123 (reported by Bruno Haible)
124 * fixed bug #361: the :FUNCTION initarg in the protocol for
125 initialization of methods can now be used to override
126 internally-produced optimized functions. (reported by Bruno
128 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
129 system-provided lambda expression no longer cause warnings about
130 unbound #:|pv-table| symbols.
131 * bug fix: improved the handling of type declarations and the
132 detection of violations for keyword arguments with non-constant
134 * bug fix: potentially erronous calls to PATHNAME and
135 MERGE-PATHNAMES were being flushed in some cases.
136 (reported by Richard Kreuter)
137 * bug fix: compiled calls to TYPEP were mishandling obsolete
138 instances. (reported by James Bielman and Attila Lendvai)
139 * bug fix: format strings with ~> without matching ~< no longer
140 trigger an AVER, but signal an understandable error instead.
141 (reported by Antonio Martinez)
142 * bug fix: specifying an output-file without a pathname-type for
143 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
144 of input-file instead of "fasl". (reported by Robert Dodier)
145 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
146 (reported by James Y Knight).
147 * bug fix: compiler-macros-function did not consider the environment
148 argument for shadowing by local functions.
149 * bug fix: compiler-macros expansion was inhibited by local INLINE
151 * bug fix: inline expansions of known functions were subject to
152 step-instrumentation in high debug policies, leading to problems
154 * bug fix: compiler failed to differentiate between different CONS
156 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
157 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
158 * bug fix: #368: incorrect use of expressed vs. upgraded array
160 * bug fix: #306a: more precise unions of array types.
161 * thread-safety improvements:
162 ** CONDITION-WAIT could return early on Linux, if the thread was
163 interrupted and subsequently continued with SIGCONT.
164 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
166 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
167 * added support for the ucs-2 external format. (contributed by Ivan
169 * minor incompatible change: pretty printing of objects of type
170 (cons symbol) is, in the default pprint-dispatch-table, now
171 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
173 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
174 called later in a class's lifetime, possibly as late as when the
175 first instance of the class is created. Previously,
176 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
177 class became finalizeable.
178 * fixed bug: FILE-POSITION sometimes returned inconsistent results
179 for multibyte external-format streams. (thanks to "vbzoli")
180 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
181 with :ALLOCATION :CLASS inherited from superclasses of the
183 * fixed bug: anonymous classes can now be created using the :NAME
184 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
185 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
186 * fixed bug: core-files saved with :EXECUTABLE T can again be
187 executed when SBCL_HOME isn't set. (reported by James Knight)
188 * fixed bug: toplevel LOCALLY forms with declarations could
189 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
190 * fixed bug: printing from several different threads using different
191 values of *print-case* could cause invalid output, due to
192 some internal special variables of the printer not being bound
193 thread-locally (reported by Max Mikhanosha)
194 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
195 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
196 (reported by Pascal Costanza)
197 * fixed bug: FINALIZE-INHERITANCE is called from
198 REINITIALIZE-INSTANCE on classes when the class has previously
199 been finalized, as required by AMOP.
200 * minor code generation optimizations:
201 ** better register allocation in CLOS dispatching functions
202 ** overflow detection when coercing signed bytes to fixnums on x86-64
203 ** is now implemented with one IMUL instruction instead of three shifts
204 ** more efficient bit-vector access on x86 and x86-64
205 ** more efficient access to raw structure slots on x86 and x86-64
206 * fixed some bugs revealed by Paul Dietz' test suite:
207 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
210 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
211 * feature: thread support on Solaris/x86, and experimental thread support
213 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
215 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
216 default method for SLOT-UNBOUND.
217 * minor incompatible change: prevent the user from specializing the
218 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
219 somewhat counter to the intent of the protocol, I (CSR) think, and
220 additionally it just doesn't work in SBCL as currently
221 implemented, thanks to optimizations (that are always valid for
222 the other three SLOT-VALUEish functions, but not for the setter).
223 * minor incompatible change: the :SB-LDB feature is now enabled by
224 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
225 the low-level debugger.
226 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
227 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
228 down to an SB-SYS:FD-STREAM, allowing direct communication in
229 more cases. Temporary files and pipes are still used as a fallback
231 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
232 on the non-reentrant C functions localtime(3) and gmtime(3).
233 * bug fix: global optimization policy was not visible in LOCALLY and
235 * bug fix: class objects can be used as specializers in methods.
236 (reported by Pascal Costanza)
237 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
238 with a type error. (reported by blitz_ on #lisp)
239 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
240 failed with a type error. (reported by Pascal Bourguignon)
241 * bug fix: merging pathnames against defaults with :DIRECTORY
242 starting with '(:RELATIVE :BACK) should preserve the :BACK.
243 (reported by James Y Knight)
244 * bug fix: saving large (>2GB) cores on x86-64 now works
245 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
246 constant index and a (simple-array (signed-byte 32)) array
247 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
248 error (patch by Robert J. Macomber)
249 * feature: TIME now displays the amount of run-time spent in GC
250 * bug fix: The debugger now does a better job of respecting
251 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
252 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
253 workaround for bug 403.)
254 * bug fix: floating point exception handling now works on FreeBSD
255 (thanks to NIIMI Satoshi)
256 * fixed some bugs revealed by Paul Dietz' test suite:
257 ** MISC.641: LET-conversion were not supposed to work in late
259 * improvements to the Win32/x86 port:
260 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
261 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
263 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
264 * new feature: source path information is generated for macro-expansion
265 errors for use in IDE's like Slime (thanks to Helmut Eller)
266 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
267 * bug fix: compiler does not loop forever on an invalid type in
269 * improvement: compilation of most CLOS applications is significantly
271 * optimization: added a limited bytecode compiler for simple toplevel
272 forms, speeding up compilation and FASL loading
273 * bug fix: the statistical profiler now properly distinguishes anonymous
276 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
277 * minor incompatible change: in sbcl-0.9.11 (but not earlier
278 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
279 into subclasses of the finalized class. Now user calls to
280 FINALIZE-INHERITANCE finalize just the one class, and calls by the
281 system return before any subclasses are finalized.
282 * minor incompatible change: The reader no longer ignores errors
283 regarding non-existent packages in #+ and #- feature tests.
284 * new feature: command line options --no-sysinit, --no-userinit to
285 inhibit loading the corresponding init files
286 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
287 for sending data through UDP sockets (thanks to François-René Rideau)
288 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
289 error is signalled instead (thanks to François-René Rideau)
290 * minor incompatible change: Error signalling behaviour of lexical
291 operations violating package locks has changed slightly. Refer to
292 documentation on package locks for details.
293 * enhancement: EVAL can process IF-expressions without resorting to the
295 * bug fix: Some operations involving SIN, COS, or TAN weren't being
296 constant-folded properly due to missing out-of-line functions.
297 (reported by Mika Pihlajamäki)
298 * bug fix: LISTEN sometimes returned T even in cases where no data was
299 immediately available from the stream
300 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
301 were reversed. (reported by Levente Mészáros)
302 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
303 when given a vector displaced to an adjustable array. (reported
305 * bug fix: derivation of float boundaries from numbers outside the
306 appropriate float range (reported by John Wiseman)
307 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
309 * bug fix: precomputing cache entries for generic functions with
310 some subclasses of specializers as yet invalid does not attempt to
311 fill a cache line with a negative offset. (reported by Levente
313 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
314 allow condition class objects as arguments to DOCUMENTATION and
315 (SETF DOCUMENTATION); only find and set documentation for
316 structure names for the STRUCTURE doc-type. (suggested by Gary
318 * improvements to the Win32/x86 port:
319 ** (user-homedir-pathname) and default initialization file
320 locations now know about the user's "Documents and Settings"
321 directory (thanks to Yaroslav Kavenchuk)
322 ** run-program is implemented (thanks to Mike Thomas)
323 ** sockets support (thanks to Timothy Ritchey)
324 ** better backtrace support (thanks to Alastair Bridgewater)
325 ** sb-grovel supported
326 ** asdf-install and sb-posix work somewhat
327 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
328 * improvements to the Solaris/x86 port:
329 ** works on Solaris 11/Solaris Express
330 ** floating-point exception handling support
331 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
332 * fixed some bugs revealed by Paul Dietz' test suite:
333 ** REMOVE-METHOD returns its generic function argument even when
334 the method is not one of the generic functions' methods.
335 ** objects declared as MEMBER types can be admissible arguments to
337 ** printing characters should simply be printed by the FORMAT ~:C
339 ** compiler failure when compiling functions with hairy constant
340 defaults for optional parameters.
341 ** compiler produces wrong code when MAYBE-INLINE-expanding a
342 function, which is already optimized.
344 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
345 * new platform: experimental support for SBCL x86/Darwin, including
346 MacOS X 10.4.5 on Intel.
347 * new feature: Unicode character names are now known to the system
348 (through CHAR-NAME and NAME-CHAR).
349 * new feature: the filesystem location of SBCL's core file is
350 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
351 * minor incompatible change: the contrib modules SB-POSIX and
352 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
353 this change is to make it easier to distribute
354 location-independent binaries.
355 * bug fix: as implied by AMOP, standardized classes no longer have
356 slots named by external symbols of public packages. (reported by
358 * optimization: calling functions via constant symbols -- as in
359 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
360 via the function object as in (FUNCALL #'FOO).
361 * optimization: on x86 and x86-64, the calling convention for the
362 case of dealing with an unknown number of values has been altered
363 to be friendlier to the prediction heuristics implemented,
364 particularly with reference to CALL and RET pairing. (thanks to
365 Alastair Bridgewater)
366 * optimization: on x86, the code for access to thread-local symbol
367 values and binding/unbinding of thread-local symbols is smaller.
368 (thanks to Alastair Bridgewater)
369 * enhancement: CONSTANTP is now able to determine constantness of
370 more complex forms, including calls to constant-foldable standardized
371 functions and some special forms beyond QUOTE.
372 * fixed bug: occasional GC crashes on Solaris/x86
373 * optimization: x86-64 supports stack allocation of results of simple
374 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
375 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
376 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
377 (thanks to James Knight)
378 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
379 by the standard (thanks to Stephen Compall)
381 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
382 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
383 be used for bundling the runtime and the core file into one
384 executable binary. This feature is not currently supported on all SBCL
385 platforms. (thanks to James Bielman and NIIMI Satoshi)
386 * new feature: a generational or ephemeral garbage collector is now
387 the default on the PowerPC platform (both Linux and Darwin). The
388 old Cheney (stop and copy) collector is a build-time option.
389 (thanks to Cyrus Harmon, after Raymond Toy)
390 * minor incompatible change: the method by which SBCL finds its
391 contributed modules has changed; it no longer relies on symbolic
392 links from an $SBCL_HOME/systems directory, but searches directly
393 in the subdirectories of $SBCL_HOME.
394 * enhancement: the dynamic heap size on the Linux/PPC platform is
395 markedly larger, even using the older Cheney garbage collector.
396 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
397 certain complicated string types. (reported by Gary King)
398 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
399 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
400 * fixed bug: DOCUMENTATION on structures no longer signals an
401 error if no structure type of the right name exists. (reported by
403 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
404 applicable non-standard primary or :AROUND method on
405 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
406 no longer cause unbound variable errors. (reported by Kevin Reid)
407 * fixed bug: metacircle resolution in cases where methods have slots
408 added before the slots from STANDARD-METHOD. (reported by Jean
410 * fixed bug: the Power PC assembly code for calling into lisp
411 sometimes computed its return address wrongly (depending on the
412 15th bit of the address, fixed at link time). This used to
413 manifest itself in a segmentation violation while building PCL.
414 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
415 many others over the years)
416 * fixed bug: ROOM no longer reports silly numbers for stack usage if
417 the stack pointer (treated as a signed integer) is negative.
418 (thanks to Peter van Eynde)
420 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
421 * new platform: experimental support for the Windows operating
422 system has been added. (thanks to Alastair Bridgewater)
423 * fixed several bugs in and robustified the PPC FFI (including
424 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
425 * bug fix: allow non-simple string symbol names (reported by Paul
427 * bug fix: interrupt handling on NetBSD (thanks to Richard M
429 * bug fix: saving a core corrupted callbacks on x86/x86-64
430 * bug fix: closed a loophole in metacircularity detection and
431 grounding in the PCL implementation of CLOS.
432 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
433 (reported by tomppa on #lisp)
434 * bug fix: clear the direction flag on Lisp -> C transitions, as
435 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
436 (reported by Andrej Grozin and Hendrik Maryns)
437 * optimization: major improvements to GC efficiency on GENCGC platforms
438 * optimization: faster implementation of EQUAL
439 * optimization: emit more efficient opcodes for some common
440 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
441 * optimization: several other minor code-generation improvements on x86-64
442 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
444 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
445 * minor incompatible change: (SETF CLASS-NAME) and (SETF
446 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
447 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
448 by ANSI as a generic function, and for consistency (SETF
449 GENERIC-FUNCTION-NAME) is treated likewise.
450 * fixed bug #233.b: make constraint propagation notice when a variable
451 value is changed after it is referenced but before it is used
452 * fixed bug #296: no more arbitrary behaviour from filenames with
453 odd characters as --load arguments. (reported by Adam Warner)
454 * fixed bug #390: :CHARACTER-SET pathname components now work as
455 expected. (reported by Tim Daly Jr)
456 * fixed bug #391: complicated :TYPE intersections in slot
457 definitions no longer cause an error in PCL internals.
458 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
459 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
460 returning the number of octets which would be written to the
461 file-stream. (thanks to Robert J. Macomber)
462 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
463 arguments to the internal stream functions. (thanks to David
465 * bug fix: stack exhaustion is now handled on the x86/SunOS
466 platform. (thanks to Daisuke Homma)
467 * bug fix: more accurate ROOM results on GENCGC platforms
468 * optimization: improved type inference for arithmetic-for
469 index variables in LOOP
470 * optimization: faster floating-point SQRT on x86-64
471 * fixed some bugs revealed by Paul Dietz' test suite:
472 ** DOCUMENTATION returns NIL instead of "" for method combinations
473 that don't have a docstring
475 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
476 * minor incompatible change: (SETF CLASS-NAME) and (SETF
477 GENERIC-FUNCTION-NAME) are no longer generic functions, and
478 therefore are not customizeable by user code (as seems to be at
479 least permitted and maybe required by AMOP). As a consolation,
480 however, the SBCL implementation of these functions now calls
481 REINITIALIZE-INSTANCE as specified by AMOP.
482 * bug fix: slot-definition documentation is propagated as per ANSI
483 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
484 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
485 argument, as do eslotd initialization methods. (from Pascal
486 Costanza's "Closer" project)
487 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
488 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
490 * bug fix: it is now possible to have more than one subclass of
491 STANDARD-GENERIC-FUNCTION without causing stack overflow.
492 (reported by Bruno Haible, Pascal Costanza and others)
493 * bug fix: the dependent update protocol now works for generic
494 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
496 * bug fix: condition-class instances corresponding to
497 DEFINE-CONDITION forms are now created eagerly. (reported by
498 Kalle Olavi Niemitalo on comp.lang.lisp)
499 * bug fix: floating point printing is more accurate in some
500 circumstances. (thanks to Simon Alexander)
501 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
502 merged with *DEFAULT-PATHNAME-DEFAULTS*.
503 * bug fix: callbacks on OS X now preserve stack-alignment.
504 (thanks to Cyrus Harmon)
505 * enhancement: the x86-64 disassembler is much better at
506 disassembling SSE instructions. (thanks to Lutz Euler)
507 * enhancement: CHECK-TYPE on a variable now allows type-inference
508 (thanks to Vincent Arkesteijn)
509 * optimization: improved performance of EUC-JP external format.
510 (thanks to NIIMI Satoshi)
511 * optimization: performance improvements to IO on file streams of
512 :ELEMENT-TYPE CHARACTER
513 * optimization: much faster memory allocation on x86-64
514 * optimization: faster garbage collections (latency and throughput)
516 * optimization: faster bignum subtraction and division on x86 and x86-64
517 (thanks to Lutz Euler)
518 * optimization: more accurate type-inference for loops that use a
519 floating point index variable or a negative step.
521 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
522 * bug fix: add a workaround to SBCL looping infinitely at startup on
523 Linux kernels with apparently buggy implementations of personality().
524 (thanks to Svein Ove Aas)
525 * bug fix: Unicode symbols are correctly printed in LDB backtraces
526 (thanks to David Lichteblau)
527 * bug fix: local bindings shadow global symbol macros for
528 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
529 * optimization: non-open coded uses of numeric comparison operators
530 (e.g. >) no longer cons when called with more than one parameter
531 on platforms supporting dynamic-extent allocation.
532 * enhancement: saving cores with foreign code loaded is now supported
533 on MIPS/Linux in addition to the previously supported platforms.
534 * bug fix: FIND-RESTART now tests for activity, not applicability when given
535 a restart object as identifier. (reported by Helmut Eller for CMUCL)
536 * bug fix: division by zero in sb-sprof when no samples were collected
537 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
538 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
539 (thanks to Alexander Kjeldaas)
540 * new feature: ignoring the return values of destructive functions
541 like NREVERSE signals a compile-time style-warning.
542 (thanks to Kevin Reid)
543 * new feature: the alignment of alien structure fields can be
544 explicitly specified. (thanks to Cyrus Harmon)
545 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
546 and dump core on SIGQUIT
548 ** incompatible change: threads do not inherit values of specials
549 from their parents (see manual)
550 ** bug fix: threads stacks belonging to dead threads are freed by the
551 next exiting thread, no need to gc to collect thread stacks anymore
552 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
553 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
554 ** bug fix: dynamic variable and thread start related gc lossage
555 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
557 * fixed some bugs revealed by Paul Dietz' test suite:
558 ** SUBTYPEP is slightly more accurate on heinously complicated
559 CONS types where some of the members have uncertain (in the
560 NIL, NIL sense) type relationships to each other.
562 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
563 no longer purified unless :PURIFY T is explicitly specified.
564 ** Non-purified cores are significantly smaller than before
566 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
567 * new feature: timers based on Zach Beane's excellent timer package
568 * added support for the following external formats: koi8-u,
569 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
570 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
571 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
572 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
573 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
574 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
575 * incompatible change: a threaded SBCL will no longer revert to
576 non-threaded mode on non-NPTL systems, but refuse to start entirely.
577 * bug fix: interrupts are disabled until startup is complete; no
578 more sigsegvs when receiving a signal to soon
579 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
581 * bug fix: add a workaround for the memory randomization features in
582 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
583 workaround will only be in effect on systems with the proc filesystem
585 * bug fix: printing objects of type HASH-TABLE signals a
586 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
588 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
589 have been running for over 50 days. (reported by Gilbert Baumann)
590 * bug fix: the logic for getting names of functions gets less
591 confused when confronded with alternate-metaclass
592 funcallable-instances. (reported by Cyrus Harmon)
593 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
594 consistent, even on internal alternate-metaclass objects.
595 * bug fix: flush closure information collected by physical
596 environment analysis prepass before the main pass. (bug reported
598 * bug fix: compiler pack phase does not modify a hashtable, which is
599 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
600 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
601 may be expanded. (reported by Alexander <malishev> on c.l.l)
602 * bug fix: no more style-warnings when compiling code using LAST
603 under high-SPEED low-SPACE settings. (reported by David Wallin
605 * fixed some incompatibilities between SBCL's MOP and the MOP
607 ** the METAOBJECT class is now implemented;
608 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
609 STANDARD-OBJECT, as required;
610 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
611 now compatible (as required by VALIDATE-SUPERCLASS); there
612 remains a consistency requirement over the presence or absence
613 of the FUNCTION class in the superclass list of finalized
614 classes; see the manual for more details;
615 ** the :AROUND method for COMPUTE-SLOTS on
616 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
617 requested slot ordering.
619 ** bug fix: parent thread now can be gc'ed even with a live
621 ** bug fix: binding a special with PROGV to no value is not
622 the same as that symbol not having been bound (thanks to
624 ** bug fix: binding specials is thread safe (thanks to
626 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
627 and SB-SPROF do not die with 'no handler for signal XX in
628 interrupt_handle_now(..)' anymore
629 ** bug fix: WITH-TIMEOUT works with multiple running threads
630 * fixed some bugs revealed by Paul Dietz' test suite:
631 ** ENSURE-GENERIC-FUNCTION should take a method class object for
632 the :method-class keyword argument.
634 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
635 * new port: the Solaris operating system on x86 processors is now
636 mostly supported, though some rough edges in the environment
637 remain. (thanks to Daisuke Homma)
638 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
639 space for its heap. (thanks to Thiemo Seufer)
640 * optimization: SBCL on MIPS platforms now supports dynamic-extent
641 closures. (thanks to Thiemo Seufer)
642 * minor incompatible change: eof selects abort in the debugger.
643 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
644 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
646 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
647 deprecated, and will go away in a future revision of SBCL.
648 * minor incompatible change: GC-ON and GC-OFF are no longer
649 implemented with a counter, it does not matter how many times gc
650 is switched on or off
651 * bug fix: discriminating functions for generic function classes
652 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
653 make invalid assumptions about method precedence order. (reported
655 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
656 platforms. (thanks to Thiemo Seufer)
657 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
658 * bug fix: correct pathname printing with printer escaping is on.
659 (thanks to Kevin Reid)
660 * bug fix: complex VOP definitions in "user-space" no longer trigger
661 package locks. (reported by Zach Beane)
662 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
663 for funcallable-instance functions, and is no different from
665 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
666 characters in the current readtable. (reported by Nicholas Neuss)
667 * bug fix: SUBTYPEP on various CONS types returns more a more
668 accurate acknowledgment of its certainty.
669 * optimizations: REMOVE-DUPLICATES now runs in linear time on
670 lists in some cases. This partially fixes bug 384.
671 * flush all standard streams before prompting in the REPL and the
673 * bug fix: signal handling and triggering gc do not conflict
674 directly anymore, in particular a high frequency sb-sprof does
675 not prevent gc from running
676 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
677 approximation for timezone and DST information between the
678 universal time epoch and the smallest negative 32-bit time_t.
679 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
680 year 1899 so long as the encoded time is non-negative.
681 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
682 the correct answer for non-constant rotate amounts when performing
683 an inline 32-bit rotation.
685 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
686 there is only one thread in the session
687 ** bug fix: memory leak for streams created in one thread and
688 written to in another
689 ** bug fix: lockup when compiled with gcc4
690 ** bug fix: race that allows the gc to be triggered when gc is
692 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
693 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
695 ** bug fix: finished threads are gc'ed properly
696 * fixed some bugs revealed by Paul Dietz' test suite:
697 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
698 has arguments with a different set of applicable methods from
699 the orignal arguments.
700 ** The type error thrown by MAP now has a correct expected-type
702 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
703 name a compiled function.
704 ** Three MISC tests where a large bignum was improperly coerced to
705 a float that couldn't represent that bignum during type
706 derivation were fixed.
707 ** SUBTYPEP can now handle types involving the negation of a
708 list-form FUNCTION type.
709 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
710 as (COMPLEX (AND RATIO (NOT FIXNUM))).
711 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
713 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
714 * New feature: Experimental support for bivalent streams: streams
715 opened with :element-type :default now allow character and binary
716 (unsigned-byte 8) I/O
717 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
718 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
720 * bug fix: no more highly sporadic "couldn't check whether ~S is
721 readable" when reading a stream and an interrupt hits in the middle
722 of a select system call
723 * compiler better recognizes complex arrays (reported by Eduardo
725 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
726 for variables on SBCL built with linkage-tables. (reported by Luis
728 * various error reporting improvements.
729 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
730 (Thanks to Thiemo Seufer)
731 * bug fix: nested reader invokations work correctly
732 * bug fix: it is possible to have simultaneous references to foreign
733 code and foreign data with the same name.
735 ** added x86-64 support
736 ** incompatible change: the threading api now works with thread
737 objects instead of thread ids
738 ** bug fix: threads are protected from signals and interruption when
739 starting up or going down
740 ** bug fix: a race where an exiting thread could lose its stack to gc
741 ** bug fix: don't halt on infinite error in threads if possible
742 ** fixed numerous gc deadlocks introduced in the pthread merge
743 ** bug fix: fixed thread safety issues in read and print
744 ** bug fix: debugger doesn't hang on session lock if interrupted at
745 an inappropriate moment
746 ** bug fix: run-program is now thread safe(r)
747 ** bug fix: inner with-recursive-lock no longer releases the mutex
748 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
749 * fixed some bugs revealed by Paul Dietz' test suite:
750 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
751 EXPECTED-TYPE slots filled.
752 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
754 ** MAKE-CONDITION accepts classes as type-designators.
755 ** COMPILE may never return NIL.
756 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
757 range before calling Unix time functions
759 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
760 * numerous signal handling fixes to increase stability
761 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
762 * minor incompatible change: we now correctly canonize default
763 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
764 as per the MOP, rather than the historical (INITARG INITFUNCTION
765 INITFORM). (reported by Bruno Haible)
766 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
767 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
768 for more information.
769 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
770 pathname is a directory pathname.
771 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
772 * dynamic space size on PPC has been increased to 768Mb. (thanks to
774 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
775 :METACLASS argument in addition to a class name. (reported by
776 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
777 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
778 (reported by Bruno Haible; patch from Gerd Moellmann)
779 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
781 * bug fix: more cleanups to the floating point exception handling on
782 x86-64 (thanks to James Knight)
783 * bug fix: the compiler does not try to constant fold calls of
784 COERCE and BIT-* functions when they return freshly constructed
785 objects. (reported by Nikodemus Siivola and Paul Dietz)
786 * optimization: improved the MIPS versions of generic (in the
787 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
788 * optimization: direct conversion of (unsigned-byte 32) to floats on
789 the PowerPC platform.
790 * optimization: structure instances with raw slots now use less
791 memory, and probably show better memory locality. (thanks to
793 * optimization: DYNAMIC-EXTENT declarations for lists and closures
794 are treated as requests for stack allocation on the x86-64,
795 Alpha-32, PPC and SPARC platforms.
796 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
797 the wrong file descriptor; implementation of SOCKET-OPEN-P.
798 (thanks to Tony Martinez)
800 ** gcing a dead thread can no longer lead to lockups
801 ** threads block signals until they are set up properly
802 ** errno is no longer shared by threads
803 ** interrupt-thread restores the eflags register on x86
804 ** fixed some lockups due to gc/thread interaction
805 * fixed some bugs revealed by Paul Dietz' test suite:
806 ** invalid dotted lists no longer raise a read error when
808 ** signal an error if a symbol that names a declaration is used
809 as the name of a type, or vice versa
810 ** allow using the (declare (typespec var*)) abbreviation for
811 (declare (type typespec var*)) with all type specifiers
812 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
813 ** always return NIL from PPRINT-POP when OBJECT is NIL
814 ** don't signal errors when pretty-printing malformed LABELS,
815 FLET or MACROLET forms
816 ** declarations in a DOLIST body are also in scope for the
818 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
820 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
823 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
824 * fixed cross-compiler leakages that prevented building a 32-bit
825 target with a 64-bit host compiler.
826 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
827 opened with :IF-EXISTS :OVERWRITE.
828 * fixed bug 281: error for an invalid qualifier in a short-form method
829 combination method is not signalled until the faulty method is called.
830 * bug fix: iteration variable type inferrer failed to deal with open
831 intervals. (reported by Alan Shields)
832 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
833 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
834 or more runtime options were provided to the sbcl binary.
835 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
836 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
837 * optimization: there's now a fast-path for fixnum arguments in the
838 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
839 * optimization: the code generated on x86-64 is more compact thanks
840 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
841 * bug fix: floating-point exception handling now partly works on
842 x86-64. (Thanks to James Knight)
843 * improvement to the MIPS backend from Thiemo Seufer: C-style
844 64-bit long long arguments and return values to alien functions
846 * fixed some bugs revealed by Paul Dietz' test suite:
847 ** the type-error signalled from WARN has a filled-in DATUM slot.
848 ** the type-error required when a stream is not associated with
849 a file has the stream as its datum.
850 ** type-errors on single-floats on x86-64 no longer have
851 :INVALID-OBJECT as the datum
852 ** the type-errors signalled for invalid function names now have
853 a correct expected type
854 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
855 for structures defined with a :TYPE. Documentation strings for
856 typed structures are no longer immediately discarded
857 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
858 broadcast streams and synonym streams. FILE-LENGTH now also works
859 on broadcast streams.
861 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
862 * incompatible change: the --noprogrammer option, deprecated since
863 version 0.7.5, has been removed. Please use the equivalent
864 --disable-debugger option instead.
865 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
867 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
868 inoperational for a while now) has been completely removed.
869 * null lexical environments are now printed as #<NULL-LEXENV>,
870 significantly reducing the amount of clutter in typical backtraces.
871 * documentation on weak pointers, finalization, and after GC hooks
872 has been added to the manual.
873 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
874 as other specialized array types, is much faster. SUBSEQ and
875 COPY-SEQ on such arrays have also been sped up.
876 * optimization: EQL is now more efficient when at least other argument
877 is known to be of type (OR FIXNUM (NOT NUMBER)).
878 * fixed bug: the runtime is now less vulnerable to changes in the
879 size of the SBCL object on OS X, and virtual memory is reserved for
880 all spaces that need to be at a fixed address.
881 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
882 * fixed bug: finalizers and after GC hooks that cause consing are now
884 * fixed bug: compiler error messages and summaries are now printed to
885 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
886 * fixed inference of the upper bound of an iteration variable.
887 (reported by Rajat Datta).
888 * fixed bug 211e: calling local functions with duplicated constant
889 keyword argument no longer causes a bogus style warning about an
891 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
892 ftype declaration to be disregarded. (reported by Dave Roberts)
893 * fixed bug 373: caused by erronous compilation of references to alien
894 variables in the runtime on ppc/darwin.
895 * fixed bug 376: CONJUGATE type deriver.
896 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
897 (reported by Baughn on #lisp)
898 * a cross-compiler bug on non-x86 platforms has been identified and
899 fixed. (thanks to Bruno Haible)
900 * improvements to the MIPS runtime code for increased stability.
901 (thanks to Thiemo Seufer)
902 * increased the maximimum compact environment size to allow
903 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
904 * partial workaround for bug 135: don't name the function we're
905 calling for hairy cases of EVAL, so as not to accumulate one environment
906 entry per call to EVAL. (reported by Kevin Reid)
907 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
908 * optimization: type testing for non-vector arrays should be faster.
909 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
911 * bug fix: setting 31st element of a bit vector to zero did not work
913 * bug fix: redefining a class definition which failed due to a
914 previous accessor / function clash now works (but see BUGS entry
915 #380 for more problems in this area). (thanks to Zach Beane)
916 * the long form of DEFINE-METHOD-COMBINATION disables method group
917 checking when given a single method group with pattern *. (thanks
919 * on x86 compiler supports stack allocation of results of simple
920 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
921 * support for building and running on Mac OS X 10.4 "Tiger" has been added
922 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
923 * fixed some bugs related to Unicode integration:
924 ** the restarts for recovering from input and output encoding
925 errors only appear when there is in fact such an error to
927 * fixed some bugs revealed by Paul Dietz' test suite:
928 ** MISC.549 and similar: late transformation of unsafe type
929 assertions into derived types caused unexpected code
931 ** SCALE-FLOAT type deriver is less wrong.
932 ** type derivers for EXP, LOG and similar functions compute result
933 types for complex arguments better.
934 ** (MISC.563) CONJUGATE type deriver works for very restricted
936 ** out-of-line type testers for character strings are available.
937 ** EQUAL compiler transform understands specialness of objects
939 ** accessing double-floats stored on the stack now works on x86-64.
940 ** debugger internals could sometimes create invalid lispobjs,
941 resulting in GC crashes.
942 ** MISC.548: type check weakening can convert required type into
944 ** initialization forms for bindings are not in scope of free special
947 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
948 * incompatible change: thread support for non-NPTL systems has
949 been removed - locking is buggy and unreliable. A threaded
950 SBCL build will now warn at startup and refuse to create
951 new threads, unless futex support is detected in the kernel
952 * incompatible change: the top level REPL now has only an ABORT
953 restart associated with it, not TOPLEVEL and ABORT as it used to.
954 TOP and TOPLEVEL are now available as debugger commands for
955 returning to the top level.
956 * incompatible change: forms evaluated in the REPL now use the
957 global optimization policy.
958 * incompatible change: user- and system-initialization files are
959 no longer processed with LOAD, but by READ and EVAL; hence the
960 global optimization policy, startup package, readtable, etc,
962 * The .fasl file format number has been incremented because of
963 various incompatible changes.
964 * internal entry point details and argument counts no longer appear
965 in backtraces unless explicitly requested by setting
966 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
967 * built-in and standard functions no longer have names like "top
968 level local call to FOO".
969 * fixed bug 32: functions defined in non-null lexical environments
970 now have more legible printed representation
971 * fixed bug 33: functions defined in non-null lexical environemnts
972 are now more amenable to inspection by INSPECT.
973 * workaround for bug 354: XEPs no longer appear in backtraces unless
974 explicitly requested.
975 * fixed bug: receiving the signal which results from attempting to
976 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
977 SIGBUS on other *BSDs) on architectures where the C stack is also
978 the Lisp stack (x86 and x86-64) from foreign code no longer leads
979 to debugger confusion or wild instability. (reported by Cheuksan
981 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
982 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
983 (reported by Lutz Euler)
984 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
985 compiled code. (reported by Paul Dietz)
986 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
987 specialized parameter to a method to something that is not TYPEP
988 the specializer is now possible.
989 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
990 face of package deletion.
991 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
992 pathname. (reported by David Tolpin on c.l.l)
993 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
994 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
995 STRING-STREAM and FILE-STREAM, respectively.
996 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
997 than previously; in addition, its test suite is now run on build.
998 (thanks to Luke Gorrie)
999 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1000 * optimization: added a immediate representation for single-floats
1002 * fixed some bugs revealed by Paul Dietz' test suite:
1003 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1004 ** attempting to create a package with a colliding nickname causes
1005 correctable errors to be signalled.
1006 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1007 ** several x86-64 backend bugs related to sign-extension of immediate
1010 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1011 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1012 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1014 * fixed loading of multiply forward-referenced layouts.
1015 (thanks to Cheuksan Wang)
1016 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1017 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1018 (see COMPILE-FILE documentation for details of :PRINT option)
1019 * fixed bugs 19 and 317: fixed-format floating point printing is
1020 more accurate. This also fixes a bug reported by Adam Warner
1021 related to the ~@F format directive.
1022 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1024 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1025 dispatch table if the from-char is a dispatch macro character.
1026 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1027 by the wordsize no longer ignore the last word. (reported by Lutz
1029 * fixed bug in type checking of dynamic-extent variables. (reported
1031 * optimization: sequence traversal functions use their freedom to
1032 coerce function designators to functions.
1033 * optimization: code with many calls to CLOS methods specialized on
1034 CLOS classes has had redundant type checks removed.
1035 * optimization: added declarations to speed up operations that access
1036 the internal character database (for example STRING-UPCASE)
1037 * optimization: comparison operations between floats and sufficiently small
1038 fixnums no longer create extra rationals
1039 * fixed some bugs related to Unicode integration:
1040 ** portions of multibyte characters at the end of buffers for
1041 character-based file input are correctly transferred to the
1042 start of the buffer at the next read.
1043 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1044 passing it through to OPEN.
1045 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1046 argument given to internal calls to OPEN.
1047 ** the built-in comment readers (introduced by character sequences
1048 ";" and "#|") are more forgiving to encoding errors; they will
1049 STYLE-WARN and then attempt to resync the stream at a character
1050 boundary. (thanks to Teemu Kalvas)
1051 * fixed some bugs in the x86-64 port:
1052 ** Negative short int return values from c-calls are sign-extended
1054 ** The stack is aligned to 16-bytes for c-calls, as required by
1055 the ABI. (thanks to Cheuksan Wang)
1056 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1057 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1058 for enums. (thanks to Vincent Arkesteijn)
1059 ** Multiple small optimizations and bugfixes for floating point
1061 * fixed some bugs revealed by Paul Dietz' test suite:
1062 ** Space, Tab, Linefeed, Return and Page have the invalid
1063 secondary constituent character trait.
1064 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1066 ** WITH-INPUT-FROM-STRING should only update the index place on
1068 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1069 works more reliably.
1070 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1071 with one-dimensional arrays with fill pointers.
1072 ** TYPE-OF failed on a complex with an integer realpart and a
1074 ** compiler failure during type inference for the code of form
1075 (IF (EQL X (THE ...)) ...) (MISC.535).
1077 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1078 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1079 platform. The port supports 61-bit fixnums, large memory spaces
1080 and reloading shared object files.
1081 * enhancement: saving cores with foreign code loaded is now
1082 supported on ppc/Darwin in addition to the previously supported
1084 * enhancement: the statistical profiler now walks deeper into the
1085 call stack for more meaningful call-graphs and accrued time
1086 reports (x86/x86-64 only). It also now reports time spent in
1088 * enhancement: it is now possible to trace most individual methods
1089 of a generic function in addition to tracing the generic function
1091 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1093 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1094 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1095 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1096 produces an error. (thanks to Vincent Arkesteijn)
1097 * bug fix: NAMESTRING on pathnames with :WILD components in their
1098 directories works correctly. (thanks to Artem V. Andreev)
1099 * fixed bug 125: compiler preserves identity of closures. (reported
1101 * bug fixed: functions with &REST arguments sometimes failed with
1102 "Undefined function" when compiled with (DEBUG 3). (reported by
1104 * bug fix: overflow during compiling of setting element of a bit
1105 vector with constant index and value. (reported by Timmy Douglas)
1106 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1107 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1108 * on x86 compiler partially supports stack allocation of dynamic-extent
1110 * GO and RETURN-FROM do not check the extent of their exit points
1111 when compiled with SAFETY 0.
1112 * fixed some bugs related to Unicode integration:
1113 ** encoding and decoding errors are now much more robustly
1114 handled; it should now be possible to recover even from invalid
1115 input or output to the terminal. (thanks to Teemu Kalvas)
1116 ** provided a first cut at implementing STRING-TO-OCTETS and
1117 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1118 ** altered the SB-MD5 contributed module slightly, changing the
1119 interface just enough for it to be supportable for builds where
1120 lisp characters are not eight bits.
1121 * fixed some bugs revealed by Paul Dietz' test suite:
1122 ** the FORMATTER-generated functions for ~V[ conditionals require
1123 the correct number of arguments.
1124 ** READ-FROM-STRING returns the mandated second value when applied
1125 to displaced strings.
1126 ** the #\Rubout and #\Backspace characters are treated as invalid
1127 constituent characters by the tokenizer.
1129 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1130 * new feature: reloading changed shared object files with
1131 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1132 * new feature: references to foreign variables and functions
1133 can now be compiled and loaded before the corresponding shared
1134 object file is loaded, as long as the foreign definitions are
1135 available at runtime.
1136 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1137 supported. (thanks to Dan Debertin)
1138 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1139 just keywords. (thanks to Vincent Arkesteijn)
1140 * fixed bug #331: structure-class instances corresponding to
1141 DEFSTRUCT forms are now created eagerly.
1142 * fixed bug #345: backtraces from calls to undefined functions work
1143 on x86 as well. Related bug #61 is now also partially fixed on x86
1144 (backtraces from throws to unknown catch tags.)
1145 * bug fix: lambda-list parsing is now stricter vrt. order and number
1146 of lambda-list keywords.
1147 * bug fix: as specified by AMOP, an error is signalled if a
1148 class-option appears multiple times in a DEFCLASS form. (reported
1150 * bug fix: RANDOM can be compiled when the compiler derives the type
1151 of its numeric argument as a disjoint set of small integers.
1152 (reported by Paul Dietz)
1153 * bug fix: starting a core saved with shared objects loaded when
1154 those objects are not available no longer causes threaded SBCL to
1155 hang. (reported by Sean Ross)
1156 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1158 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1159 argument, which must always be NIL. (reported by Kalle Niemitalo)
1160 * bug fix: printing 1.0d+23 no longer results in an error.
1161 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1162 * bug fix: structure slot setters preserve evaluation order. (thanks
1164 * fixed some bugs related to Unicode integration:
1165 ** RUN-PROGRAM can allow its child to take input from a Lisp
1166 stream. (reported by Stefan Scholl)
1167 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1168 locales. (reported by Ken Causey)
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** INCF, DECF and REMF evaluate their place form as specified in
1172 ** FORMATTER expands ~{ ~} iteration directives with V or #
1173 parameters correctly.
1174 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1175 consequent uses no arguments correctly.
1176 ** the system has a partial understanding of the (COMPLEX RATIO)
1178 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1181 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1182 * new feature: a build-time option (controlled by the :SB-UNICODE
1183 keyword feature, enabled by default) for building the system with
1184 support for the entire 21-bit character space defined by the
1186 * new feature: the system now has rudimentary external-format
1187 support; the primary user-visible change at this time is that
1188 characters with the high bit set (such as lower-case-e-acute) will
1189 print correctly to a terminal in a UTF-8 environment.
1190 * minor incompatible change: BASE-CHAR no longer names a class;
1191 however, CHARACTER continues to do so, as required by ANSI.
1192 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1193 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1194 has been moved to the SB-EXT package (temporarily re-exported from
1196 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1197 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1199 * fixed bug #351: better error-handlind and reporting for malformed
1201 * fixed bug #350: bignum-printing is now more memory-efficient,
1202 allowing printing of very large bignums, eg. (expt 2 10000000).
1203 (reported by Bruno Haible)
1204 * fixed bug #302: better primitive-type selection for intersection
1206 * fixed bug #308: non-graphic characters now all have names, as
1207 required. (reported by Bruno Haible)
1208 * bug fix: redefining a class with different superclasses now correctly
1209 removes it from the direct-subclasses of its previous superclasses.
1210 (reported by David Morse)
1211 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1212 new value now works. (reported by Bruno Haible)
1213 * bug fix: correct canonicalization of multiple non-standard slot
1214 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1215 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1216 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1217 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1218 now exists, an signals an error.
1219 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1220 during macroexpansion for non-iteration variables as well. (reported
1221 by Bruno Haible for CMUCL)
1222 * bug fix: Cyclic structures and unprintable objects in compiler
1223 messages no longer cause errors. (reported by Bruno Haible)
1224 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1225 and READ-FROM-STRING all now return a primary value of NIL if
1226 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1227 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1228 (reported by Bruno Haible for CMUCL)
1229 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1230 do not designate a condition. (reported by Bruno Haible for
1232 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1233 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1234 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1235 permitting resolution in favour of any of the conflicting symbols.
1236 (reported by Bruno Haible for CMUCL)
1237 * bug fix: EQUAL compiler optimizations is less aggressive on
1238 strings which can potentially compare true despite having distinct
1239 specialized array element types.
1240 * bug fix: unit enumerations can be defined without dividing by
1241 zero. (thanks to Vincent Arkesteijn)
1242 * FORMAT compile-time argument count checking has been enhanced.
1243 (report from Bruno Haible for CMUCL)
1244 * a partial workaround for the bug 262: the compiler does not try to
1245 inline-expand a local function doing RETURN-FROM from a deleted
1246 BLOCK. (thanks to Peter Denno for the bug report and to David
1247 Wragg for the simple test case)
1248 * fixed some bugs revealed by Paul Dietz' test suite:
1249 ** RENAME-PACKAGE allows all package designators as new package
1251 ** constraint propagation and lambda variable substitution are
1252 more cautious in dealing with partially deleted code.
1253 ** compiler sometimes lost reoptimization passes.
1254 ** CERROR, given a condition as condition designator, passes its
1255 remaining arguments to the continue format control without
1257 ** Case-altering FORMAT directives work correctly on non-ASCII
1259 ** The REAL type specifier handles bounds outside the floating
1260 point ranges without signalling FLOATING-POINT-OVERFLOW.
1261 ** Functions with IR1-transformations can create intercomponent
1262 references to global functions.
1263 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1265 ** FORMAT ~R treats a nil value for its first parameter correctly.
1267 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1268 * enhancement: saving cores with foreign code loaded is now
1269 supported on x86/NetBSD and sparc/Linux in addition to the previously
1270 supported platforms.
1271 * bug fix: on some platforms repeated installations caused multiple
1272 copies of HTML documentation to be installed -- should not happen
1273 any more. (reported by Stefan Scholl)
1274 * bug fix: parsing self-recursive alien record types multiple times
1275 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1276 original patch by Helmut Eller for CMUCL)
1277 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1278 (thanks to Richard Kreuter)
1279 * bug fix: defining classes whose accessors are methods on existing
1280 generic functions in other (locked) packages no longer signals
1281 bogus package lock violations. (reported by François-René Rideau)
1282 * bug fix: special variables as DEFMETHOD parameters no longer have
1283 associated bogus type declarations. (reported by David Wragg and
1285 * bug fix: read-write consistency on streams of element-type
1286 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1287 * bug fix: redefiniton of the only method of a generic function with
1288 no DEFGENERIC no longer emits a full WARNING. In addition,
1289 redefinition of generic functions with no DEFGENERIC to an
1290 incompatible lambda list now signals an error. (thanks to Zach
1292 * bug fix: DEFGENERIC now works even when there's a function of the
1293 same name in an enclosing lexical environment. (thanks to Zach
1295 * fixed compiler failure, caused by instrumenting code during
1296 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1297 * optimization: added loop analysis and improved register allocation
1298 * fixed some bugs revealed by Paul Dietz' test suite:
1299 ** POSITION on displaced vectors with non-zero displacement
1300 returns the right answer.
1301 ** (SIMPLE-STRING) is a valid type specifier for sequence
1303 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1305 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1306 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1308 ** The justification version of the FORMAT ~< directive treats
1309 non-zero minpad parameter correctly.
1311 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1312 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1313 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1314 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1315 the supported interface.
1316 * new feature: Single-stepping of code compiled with DEBUG 2 or
1317 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1318 * new feature: saving cores with foreign code loaded is now
1319 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1320 Timothy Moore's work for CMUCL)
1321 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1322 parameters to * instead of NIL if no initform is supplied.
1323 (reported by Johan Bockgaard)
1324 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1325 correctly when FUNCALL appears as the car of the form. Note:
1326 despite this FUNCALL forms are not currently subject to
1327 compiler-macro expansion. (port of Raymond Toy's fix for the
1328 same from CMUCL, reported by Johan Bockgaard)
1329 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1330 (thanks for Teemu Kalvas)
1331 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1332 case instead of terminating the iteration (thanks for Julian
1333 Squires, Sean Champ and Raymond Toy)
1334 * bug fix: incorrect expansion of defgeneric that caused a style
1335 warning. (thanks for Zach Beane)
1336 * bug fix: slot accessor effective method computation works properly
1337 for classes with multiple non-standard applicable methods on
1338 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1339 * on x86 compiler supports stack allocation of results of LIST and
1340 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1341 CMUCL implementation by Gerd Moellmann)
1342 * fixed some bugs revealed by Paul Dietz' test suite:
1343 ** FORMAT strings with both the ~<~:;~> form of the justification
1344 directive and pretty-printing directives cause an error.
1346 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1347 * incompatible change: the internal functions
1348 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1349 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1350 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1351 instead of the old functions.
1352 * new feature: on platforms where "dladdr" is available foreign
1353 function names now appear in backtraces. (based on Helmut Eller's
1355 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1356 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1358 * documentation: ASDF manual is now distributed with SBCL.
1359 * bug fix: SBCL can now load its contributed modules with REQUIRE
1360 even if the system-provided entries have been removed from
1361 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1363 * bug fix: backtraces involving undefined functions or assembly
1364 routines are more informative. (thanks to Brian Downing)
1365 * bug fix: mutually referent alien structures now work correctly.
1366 (reported by Rick Taube)
1367 * bug fix: structures defined by WITH-ALIEN can be referred to
1368 within other definitions in the same WITH-ALIEN.
1369 * bug fix: division operators (MOD, TRUNCATE and the like) with
1370 constant zero divisors and integer dividends no longer generate
1372 * bug fix: provide default methods for INPUT-STREAM-P and
1373 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1374 * bug fix: improve the detection and resolution of MOP metacycles
1375 (where effective method computation depends on the generic function
1376 having its effective method computed). (reported by Bruno Haible)
1377 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1378 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1379 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1380 * bug fix: pathnames with relative directory components can be
1381 represented relative to default pathnames.
1382 * optimization: in taking the GCD of bignums, reduce the two bignums
1383 to approximately the same size (using Euclid's algorithm) before
1384 applying the more sophisticated binary GCD. (thanks to Juho
1386 * optimization: COUNT on bitvectors now operates word-at-a-time.
1387 * optimization: ASH with a positive, but not necessarily constant,
1388 (leftwards) shift, when in a modular context, is compiled to a
1390 * fixed some bugs revealed by Paul Dietz' test suite:
1391 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1392 if the corresponding argument is NIL.
1393 ** FORMAT directives accepting a minpad parameter treat negative
1395 ** literal commas as character parameters to FORMAT directives are
1397 ** literal spaces directly after ~<Newline> directives within a
1398 format-logical-block (~:< ~@:>) do not induce :FILL-style
1399 conditional newlines.
1400 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1401 as specified: it no longer includes conditional newlines).
1402 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1404 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1405 :PER-LINE-PREFIX argument does not evaluate to a string.
1407 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1408 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1409 the manual for details; for now, package locks can be disabled by
1410 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1411 if no major problems are found then it is likely that they will be
1412 compiled in unconditionally.
1413 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1414 unsupported operators on all platforms. To load a shared library
1415 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1416 object file, link it into a shared library outside of SBCL and
1417 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1419 * bug workaround: ROOM T and the GC were not working together
1420 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1421 reporting facility aren't preserved. That reporting has been
1422 disabled, so now until and unless someone figures out how to make it
1423 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1424 * minor incompatible change: as threatened around sbcl-0.8.0, the
1425 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1426 The symbols are also exported from SB-PCL for backwards
1427 compatibility, but more so than before SB-PCL should be treated as
1428 an implementation-internal package.
1429 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1431 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1432 It is known as of this release not to work on the Alpha, however.
1433 * fixed bug #167: errors signalled due to illegal syntax in method
1434 bodies are now more legible.
1435 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1436 designators and can hence be used with TYPEP.
1437 * fixed bug #333: CHECK-TYPE now ensures that the type error
1438 signalled, if any, has the right object to be accessed by
1439 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1440 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1441 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1443 * fixed bug #334: programmatic addition of slots using specialized
1444 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1445 and :ALLOCATION :CLASS slots.
1446 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1447 float exponents. (rereported by Peter Seibel)
1448 * fixed a bug: #\Space (and other whitespace characters) are no
1449 longer considered to be macro characters in standard syntax by
1450 GET-MACRO-CHARACTER.
1451 * fixed bug: initialization of condition class metaobjects no longer
1452 causes an instance of the condition to be created. (reported by
1454 * fixed bug: it is now possible to have slots such that
1455 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1456 description is neither :INSTANCE nor :CLASS.
1457 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1458 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1459 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1460 non-standard SLOT-DEFINITION-ALLOCATION values.
1461 * fixed bugs: various race conditions handling exiting threads.
1462 CL-PPCRE's thread torture test now passes.
1463 * fixed arguably-a-bug: GC time proportional to number of threads in
1464 system even when most of them are idle
1465 * optimization: improved performance of BIT and SBIT on bit-vectors.
1466 * .fasl file incompatibility: The fasl file version number has
1467 been incremented because of changes associated with package locks.
1469 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1470 * minor incompatible change: the system no longer provides
1471 optimization hints (conditions which are TYPEP
1472 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1473 mode; these hints are still emitted when higher SPEED optimization
1475 * new contrib module: a sampling profiler (profiling by statistical
1476 sampling, rather than by instrumenting functions) is available as
1477 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1478 * the behaviour of the standard function ED is now customizeable by
1479 third parties through a hook variable: see ED's documentation
1480 string for information on the protocol.
1481 * the compiler no longer emits efficiency notes for (FUNCALL X)
1482 when the type of X is uncertain under default optimization
1484 * fixed bug 276: mutating a binding of a specialized parameter to a
1485 method to something that is not TYPEP the specializer is now
1487 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1488 exhibit the required behaviour on the x86 platform. (thanks to
1489 Peter van Eynde, Eric Marsden and Bruno Haible)
1490 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1491 even for difficult arguments. (reported by Peter Graves)
1492 * fixed bug 141a: the backquote printer now descends quoted
1494 * fixed another bug in backquote printing: no more destructive
1495 modification of the form's list structure. (reported by Brian
1497 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1498 move between its address being taken and the call to
1499 interrupt_thread, fixing a crashing race condition.
1500 * the SB-POSIX contrib implementation has been adjusted so that it
1501 no longer exhibits ridiculously poor performance when constructing
1502 instances corresponding to C structs.
1504 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1505 * minor incompatible change: the sb-grovel contrib now treats C
1506 structures as alien (in the sense of SB-ALIEN) objects rather than
1507 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1508 has implications for memory management of client code
1509 (sb-grovel-returned objects must now be manually managed) and for
1510 type safety (alien objects now have full types).
1511 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1512 used to control emission of compiler diagnostics, rather than the
1513 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1514 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1515 quality should be considered deprecated.
1516 * install.sh now installs the user manual as well
1517 * (not quite a new documentable feature, but worth considering in
1518 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1519 beginnings of a semantically meaningful condition hierarchy is
1520 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1522 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1523 designator as the defaults argument.
1524 * fixed bug: Displaced arrays whose displaced-to array has become
1525 too small now cause ARRAY-DIMENSION to signal an error, providing
1526 for safer bounds-checking. (reported by Bruno Haible)
1527 * fixed bug: DEFCLASS slot definitions with identical :READER and
1528 :WRITER names now signal a reasonable error. (reported by Thomas
1530 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1532 * fixed bug: Invalid slot specification errors now print correctly.
1533 (thanks to Zach Beane)
1534 * fixed bug 320: Shared to local slot value transfers in class
1535 redefinitions now happen corrently from superclasses as
1536 well. (reported by Bruno Haible)
1537 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1539 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1540 improper lists correctly. (reported by Bruno Haible)
1541 * fixed bug 313: source-transform for <fun-name> was erroneously
1542 applied to a call of a value of a variable with name <fun-name>.
1543 (reported by Antonio Menezes Leitao)
1544 * fixed bug 307: The obsolete instance protocol ensures that
1545 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1546 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1547 remove dead unknown-values globs from the middle of the stack.
1548 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1549 easier to resume long computations after using *BREAK-ON-SIGNALS*
1550 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1551 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1552 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1553 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1554 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1555 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1557 * fixed a (fairly theoretical) bug in string printing: if
1558 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1559 does not have array-element-type equal to the most general string
1561 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1562 function name. (thanks to Nikodemus Siivola)
1563 * fixed bug: compile-time format string checker failed on
1564 non-closed ~{. (reported by Thomas F Burdick)
1565 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1566 don't warn on legal loop code involving more than one aggregate
1567 boolean. (thanks to Nikodemus Siivola)
1568 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1569 in keyword position to MAKE-PACKAGE was not being reported as
1571 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1572 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1574 * fixed bugs 280 and 312: the checking for multiple definitions in a
1575 file is less likely to become confused by uses of inline
1577 * fixed bug: the #S reader macro performs the keyword coercion
1578 specified for slot names. (reported by Kalle Niemitalo)
1579 * fixed bug: lambda lists may contain symbols whose names start with
1580 & but are not lambda-list-keywords; their occurrence triggers a
1582 * fixed bug 321: define-method-combination argument lambda lists do
1583 not cause invalid code to be generated when &optional or &aux
1584 variables have default values. (reported by Bruno Haible)
1585 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1586 have CLOS classes; this appears also to have fixed reported
1587 instability in stack exhaustion detection.
1588 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1589 MUFFLE-WARNING when no associated restart is present is now
1591 * optimization: rearranged the expansion of various defining macros
1592 so that each expands into only one top-level form in a
1593 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1595 * optimization: used a previously little-used slot in symbols to
1596 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1598 * fixed some bugs revealed by Paul Dietz' test suite:
1599 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1601 ** Fixed an optimization bug related to inheritance of initforms
1602 from local to shared slots.
1603 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1604 ** CAST optimizer forgot to flush argument derived type.
1605 ** print/read consistency on floats is now orders of magnitude
1606 more likely. (thanks also to Bruno Haible for a similar report
1608 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1609 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1610 for objects of type REAL. Make it so.
1611 ** SXHASH is specified (infelicitously) to respect similarity,
1612 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1613 it so. (thanks to Markus Ziegler)
1614 ** on the Alpha, the compiler succeeds in compiling functions
1615 returning a known number of arguments greater than 63.
1616 ** fixed handling of invalid NIL arguments in keyword position
1618 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1620 ** provided a readably-printable representation for RANDOM-STATE
1622 ** ensured that pathnames, where they have a namestring, always
1623 print using #P"..." syntax.
1625 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1626 * Support for the forthcoming 2.0 version of the NetBSD kernel
1627 running on x86 hardware has been added. (thanks to Perry
1628 E. Metzger most immediately, and others for their past work)
1629 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1630 current release version; SBCL's OpenBSD support had been broken
1631 since about the time of OpenBSD's switch to ELF binary format.)
1632 * [placeholder for DX summary]
1633 ** user code with &REST lists declared dynamic-extent, under high
1634 speed or space and low safety and debug optimization policy.
1635 * The manual has been converted to Texinfo format and the debugger
1636 chapter from the cmucl manual has been added.
1637 * A facility has been added to extract documentation strings from
1638 sbcl and store them as Texinfo-formatted snippets for inclusion in
1639 the manual (via Texinfo's @include directive)
1640 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1641 the test case to Dave Roberts)
1642 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1643 pointers. (reported by Sean Ross)
1644 * bug fix: PROFILE output is printed nicely even for large numerical
1645 values. (thanks to Zach Beane)
1646 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1647 greater than 32 handle EOF correctly.
1648 * bug fix: on X86 an immediate argument of the IMUL instruction is
1649 correctly printed in disassembly. (thanks to Lutz Euler)
1650 * bug fix: class slots in redefined classes preserve their old
1651 values. (thanks to Bruno Haible and Nikodemus Siivola)
1652 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1653 succeeds. (reported by Marco Baringer)
1654 * fixed some bugs revealed by Paul Dietz' test suite:
1655 ** READ-SEQUENCE now works on ECHO-STREAMs.
1656 ** RATIONALIZE works more according to its specification. (thanks
1659 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1660 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1661 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1662 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1663 should matter to you only if you rebind the printer control
1664 variables and then find you want different bindings in the
1665 debugger than in the ordinary execution of your program.)
1666 * The runtime build system has been tweaked to support building
1667 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1668 assembler and linker. (thanks to Nikodemus Siivola)
1669 * Unbound, undefined, undeclared variables now trigger full
1670 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1671 more often programmer error than deliberate exploitation of undefined
1673 * optimization: the hash algorithm for strings has changed to one
1674 that is less vulnerable to spurious collisions. (thanks to Juho
1676 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1677 needless bounds checking. (thanks to Juho Snellman)
1678 * optimization: implemented multiplication as a modular
1679 (UNSIGNED-BYTE 32) operation on the PPC backend.
1680 * fixed some bugs revealed by Paul Dietz' test suite:
1681 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1683 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1684 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1685 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1687 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1689 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1691 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1692 their output stream on EOF from read.
1693 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1694 have been read to end-of-file.
1695 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1697 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1698 description of determination of which consecutive characters
1700 ** Printing the "Space" character with escaping on now yields "#\\ ",
1701 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1702 ** Reading floating-point numbers with *READ-BASE* set to a number
1703 less than 10 works correctly.
1704 ** Reading floating-point numbers with *READ-BASE* set to a number
1705 more than 10 works correctly.
1706 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1707 the readtable currently in effect.
1709 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1710 * minor incompatible change: parsing of namestrings on a physical
1711 (Unix) host has changed; numbers after the final #\. in a
1712 namestring are no longer interpreted as a version field. This is
1713 intented to be largely invisible to the user, except that the
1714 meaning of the namestring "*.*.*" has changed: it now refers to a
1715 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1716 should usually be replaced by
1717 (make-pathname :name :wild :type :wild :version :wild)
1718 with the added benefit that this is more likely to be portable.
1719 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1720 option to OPEN now signals an error if the file being opened
1721 exists; this may have an impact on existing code.
1722 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1723 signals received, on the PowerPC platforms (both Linux and
1724 Darwin). (thanks to Pierre Mai for pointing out the location of
1726 * several fixes on OS X: The system now builds and runs cleanly on
1727 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1728 (thanks to Brian Mastenbrook)
1729 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1730 second-resolution: integer multiples of 1/3600 between -24 and 24.
1731 (thanks to Vincent Arkesteijn)
1732 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1733 type when called with 1 argument; PEEK-CHAR checked type of
1734 PEEK-TYPE only after having read first character from a
1735 stream. (reported by Peter Graves)
1736 * bug fix: the garbage collector now has much better locality
1737 behaviour, and in particular no longer treats all memory as being
1738 exhausted when many small objects point to each other in a deeply
1740 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1741 recognized as being TYPEP their class.
1742 * bug fix: the PUSHNEW documentation string has been corrected.
1743 (thanks to Vincent Arkesteijn)
1744 * bug fix: defaulting of the value for the last of an atypically
1745 large number of multiple values being bound was not being
1746 performed correctly on the Alpha or PPC platforms
1747 * optimization: implemented multiplication as a modular
1748 (UNSIGNED-BYTE 32) operation on the x86 backend.
1749 * optimization: SEARCH on simple-base-strings can now be open-coded.
1750 (see also contrib/compiler-extras.lisp for inspiration for
1751 teaching the compiler about the Boyer-Moore algorithm).
1752 * value, returned by MAX (and MIN) called with several EQUALP, but
1753 not EQL, arguments now does not depend on compiler settings.
1754 * fixed some bugs revealed by Paul Dietz' test suite:
1755 ** in stack analysis liveness information is propagated from
1756 non-local entry points.
1757 ** pathwise CAST removing failed when the CAST node did not start
1759 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1760 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1762 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1763 host is already defined.
1764 ** RENAME-FILE works on streams instead of signalling an internal
1766 ** PEEK-CHAR uses the current readtable when determining whether
1767 or not a character is whitespace.
1768 ** MERGE-PATHNAMES handles the case when the pathname does not
1769 specify a name while the default-pathname specifies a version.
1770 ** Pathnames now stand a better chance of respecting print/read
1772 ** Attempting to use standardized file system operators with a
1773 pathname with invalid :DIRECTORY components signals a
1775 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1777 ** DIRECTORY on logical pathnames is more correct.
1778 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1779 signal a TYPE-ERROR if their argument is not a stream.
1780 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1781 designator argument does not designate a stream.
1782 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1783 examining the synonym.
1784 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1786 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1787 element-type larger than ([UN]SIGNED-BYTE 32).
1789 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1790 * When built with the :SB-FUTEX feature, threaded builds now take
1791 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1792 for faster/more reliable mutex and condition variable support.
1793 * Incompatible change (but one you probably shouldn't have been using
1794 anyway): the interface and code for arbitrating between multiple
1795 threads in the same user session has been redesigned.
1796 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1797 error when called without an explicit environment argument.
1798 (thanks to Nikodemus Siivola)
1799 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1800 confused about their position. (thanks to Adam Warner and Gerd
1802 * bug fix: Pretty printing backquoted forms with unquotations in the
1803 argument list position of various code constructs such as LAMBDA
1804 now works correctly. (reported by Paul Dietz)
1805 * bug fix: Pretty printing unquotations no longer loses all
1806 stream position information.
1807 * optimization: performance of string output streams is now less
1808 poor for multiple small sequence writes.
1809 * optimization: performance of CSUBTYPEP in the presence of complex
1810 expressions involving CONS and NOT many times has been improved.
1811 (reported by Paul Dietz)
1812 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1814 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1815 streams for the REPL, for the debugger, and so on. (thanks to
1817 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1818 * fixed some bugs revealed by Paul Dietz' test suite:
1819 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1820 is ignored if it is too small, rather than propagating through
1821 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1822 ** extremely complex negations of CONS types were not being
1823 sufficiently canonicalized, leading to inconsistencies in
1825 ** VALUES tranformer lost derived type.
1827 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1828 * fixed a bootstrapping bug: the build process no longer assumes
1829 that the various BOOLE-related constants have the same value in
1830 host and target lisps. (noted by Paul Dietz' test suite on an
1831 SBCL binary built from CLISP)
1832 * The system can now be dynamically linked on the MIPS platform,
1833 which enables dynamic loading of foreign code from Lisp. (thanks
1834 to Ralf Baechle for discussions on the MIPS ABI)
1835 * The system now records debugging information for its own source
1836 files in a filesystem-position-independent manner, relative to
1837 the "SYS" logical host.
1838 * fixed a compiler bug: MV-LET convertion did not check references
1839 to the "max args" entry point. (reported by Brian Downing)
1840 * tweaked disassembly notes to be less confident about proclaiming
1841 some instruction as an LRA. (thanks to Brian Downing)
1842 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1843 now each have their own history, command character, and other
1844 characteristics. (thanks to David Lichteblau)
1845 * fixed some bugs revealed by Paul Dietz' test suite:
1846 ** compiler failure in compiling LOGAND expressions including a
1848 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1849 shift greater than 32.
1850 ** FLUSH-DEST did not mark blocks for type check regeneration.
1851 ** HANDLER-CASE failed to accept declarations in handler clauses
1852 in some circumstances.
1854 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1855 * New code in contrib/sb-introspect (still probably not entirely
1856 stable yet) provides some support for smart Lisp development
1857 environments like SLIME.
1858 * The conditions signalled for errors occurring when loading .fasl
1859 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1860 in a way which should help ASDF recover gracefully.
1861 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1862 stuff has been cleaned up. If you code contrib/ stuff, this might
1863 affect you, and you can look at contrib/README, contrib/STANDARDS,
1864 and/or the 0.8.4.27 diff to check.
1865 * In full calls the compiler now does not generate checks for declared
1866 argument types for all arguments.
1867 * various threading fixes
1868 ** and some experimental patches which didn't make it into
1869 the main tree for this release, but which are shipped in
1870 contrib/experimental-thread.patch as a possible fix for some
1871 failures (deadlock, spinning...) in GC-intensive multithreaded
1873 * fixed PPC build problem (source code incompatibility of different
1874 library versions): added offsetof() hackery which attempts to divine
1875 where glibc maintainers put uc_mcontext today
1876 * fixed bug 282: compiler does not trust type assertions while passing
1877 arguments to a full call.
1878 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1879 &OPTIONAL VALUES type parameter.
1880 * fix bug 214: algorithm for noting rejected templates is now more
1881 similar to that of template seletion. (also reported by rydis on
1883 * fixed bug 141b: printing backquoted information readably and prettily
1884 inserts a space where necessary.
1885 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1886 (CHARACTER 10) are now reported as errors, rather than propagated
1887 as unknown types. (reported by piso on #lisp)
1888 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1889 with values NIL and :ERROR. (thanks to Milan Zamazal)
1890 * fixed bug 191c: CLOS now does proper keyword argument checking as
1891 described in CLHS 7.6.5 and 7.6.5.1.
1892 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1893 longer raise an error; further, using a list as a for-as-arithmetic
1894 counter now raises a meaningful error.
1895 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1896 now understood by sequence creation functions such as MAKE-SEQUENCE
1898 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1899 a STRING-INPUT-STREAM.
1900 * compiler enhancement: SIGNUM is now better able to derive the type
1902 * type declarations inside WITH-SLOTS are checked. (reported by
1904 * fixed some bugs revealed by Paul Dietz' test suite:
1905 ** incorrect optimization of TRUNCATE for a positive first
1906 argument and negative second.
1907 ** compiler failure in let-convertion during flushing dead code.
1908 ** compiler failure while deriving type of TRUNCATE on an
1909 interval, containing 0.
1910 ** ASH of a negative bignum by a negative bignum count now returns
1912 ** intersection of CONS types now canonicalizes properly, fixing
1913 inconsistencies in SUBTYPEP.
1915 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1916 * incompatible change: The --disable-debugger command line
1917 option now clobbers the debugger at a more fundamental
1918 level, by redefining #'INVOKE-DEBUGGER instead of by
1919 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1920 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1921 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1922 enter the debugger REPL and then suffer recursive errors
1923 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1924 while under the new system, BREAK in --disable-debugger mode
1925 terminates the system just as an unhandled error would.
1926 * fixed compiler performance when processing loops with a step >1;
1927 * bug fix: DOCUMENTATION now retrieves generic function
1928 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1929 support has been systematized, and now supports the methods
1930 specified by ANSI, along with a default method and a method for
1931 slot documentation. (reported by Nathan Froyd)
1932 * bug fix: effective methods associated with a generic function are
1933 no longer cached over a change of that generic function's method
1934 combination. (reported by Andreas Fuchs)
1935 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1936 names a non-existent directory. (thanks to Andreas Fuchs)
1937 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1938 the PPC platform, lead to stack corruption; this has been fixed.
1939 (reported by Rainer Joswig)
1940 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1941 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1942 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1943 arithmetic in general, is now much more reliable on the PPC
1945 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1946 platform now returns the right answer.
1947 * optimization: restored some effective method precomputation in
1948 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1949 precomputation is now tunable.
1950 * optimization: compiler-internal data structure use has been
1951 reviewed, and changes have been made that should improve the
1952 performance of the compiler by about 20%.
1953 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1954 simple-base-strings and simple-bit-vectors is improved.
1955 * optimization: the optimization of 32-bit logical and arithmetic
1956 functions introduced in version 0.8.3 on the x86 has been
1957 implemented on the mips, ppc and sparc platforms; an
1958 implementation of the same facility, but for 64-bit arithmetic,
1959 has been added for the alpha.
1960 * microoptimization: the compiler is better able to make use of the
1961 x86 LEA instruction for multiplication by constants.
1962 * bug fix: in some situations compiler did not report usage of
1963 generic arithmetic in (SPEED 3) policy.
1964 * bug 145b fix: compiler used wrong type specifier while converting
1965 MEMBER-types to numeric.
1966 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1968 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1969 index argument now works properly on the Alpha platform.
1970 * bug fix: floating point exception treatment on the Alpha platform
1972 * bug fix: FILE-POSITION works much better on string input and
1973 output streams. (thanks to Nikodemus Siivola)
1974 * bug fix: many threading/garbage collection symptoms sorted.
1975 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1976 might be pseudo-atomic.
1977 * internal change: Stopping for GC is now done with signals not
1978 ptrace. GC is now done in whichever thread wanted it, instead of
1980 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1982 * bug fix: result form in DO is not contained in the implicit
1984 * incompatible change: ICR structure is changed; the value part of
1985 CONTINUATION is now called LVAR; corresponding functions are
1986 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1987 * added type deriver for ISQRT (thanks to Robert E. Brown).
1988 * bug fix: better support for loading from the command line when an
1989 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1991 * fixed some bugs revealed by Paul Dietz' test suite:
1992 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1993 ** ROUND and FROUND now give the right answer when given very
1994 small float arguments.
1995 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1997 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1998 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1999 ** LCM should return a non-negative integer.
2000 ** PARSE-INTEGER returned the index of a terminator instead of the
2001 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2002 ** PARSE-INTEGER returned an incorrect index being applied to a
2004 ** LCM with two arguments of 0 returns 0 rather than signalling
2006 ** unsigned addition of a 32-bit constant with the high bit set no
2007 longer causes an internal compiler error on the x86.
2008 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2009 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2010 with negative last argument.
2011 ** byte specifiers with zero size and position no longer cause
2012 an error during type derivation.
2013 ** bignum multiplication on the Alpha platform now returns the
2015 * porting: The system now builds on SuSE AMD64, although it still
2016 generates a 32-bit binary.
2017 * .fasl file incompatibility: The fasl file version number has
2018 been incremented (because of the changes to internal compiler
2019 data structures referred to above).
2021 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2022 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2023 more accurately, on the Darwin kernel running on PowerPC hardware.
2024 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2025 * Compiler code deletion notes now signal a condition of type
2026 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2027 an associated MUFFLE-WARNING restart.
2028 * The compiler now performs limited argument count validation of
2029 constant format strings in FORMAT, and where appropriate in ERROR,
2030 CERROR and WARN. (thanks to Gerd Moellmann)
2031 * New ASDF-INSTALL contrib can be used for automatic download and
2032 installation of third-party Lisp code from CCLAN or other sites
2034 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2035 which forces another thread to execute a function supplied by the
2037 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2038 now accept and act upon their :ELEMENT-TYPE keyword argument.
2039 (reported by Martin Atzmueller, Edi Weitz)
2040 * bug fix: FILE-POSITION now accepts position designators up to
2041 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2042 is the greater. (thanks to Patrik Nordebo)
2043 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2044 Kalle Olavi Niemitalo)
2045 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2046 printable. (reported by Eric Marsden)
2047 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2048 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2049 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2050 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2051 (thanks to Stig Sandoe)
2052 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2053 UNREAD-CHAR now correctly includes the unread character in the
2054 target sequence. (thanks to Gerd Moellmann)
2055 * bug fix in threaded builds: the system can now be suspended and
2056 resumed by shell job control with minimal disruption.
2057 * bug fixes in times and timezones >2038AD
2058 * better handling of "where is GNU make?" problem in build scripts
2059 (thanks to Nikodemus Siivola)
2060 * new optimization: inside a named function any reference to a
2061 function with the same name is considered to be a self-reference;
2062 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2063 optimization quality.
2064 * new optimization on x86: logical functions and + now have
2065 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2066 used when the result is truncated to 32 bits.
2067 * VALUES declaration is partially enabled.
2068 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2069 * fixed some bugs revealed by Paul Dietz' test suite:
2070 ** The system now obeys the constraint imposed by
2071 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2072 types form a lattice under type intersection.
2073 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2074 ** ASSOC now ignores NIL elements in an alist.
2075 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2076 and (1+ MOST-POSITIVE-FIXNUM) answers.
2077 ** The addition of a method with invalid qualifiers to a generic
2078 function does not cause an error to be signalled immediately;
2079 a warning is signalled, and the error is generated only on
2080 calling the generic function.
2081 * changed .fasl file version number, in part to add type codes for
2082 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2083 obscure ANSI requirements
2085 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2086 * fixed bug 148: failure to inline-expand a local function left
2087 garbage, confusing the compiler.
2088 * fixed bugs 3cd: structure slot readers perform type check if the
2089 slot can have an invalid value (i.e. it is either not initialized
2090 or can be written with a less specific slot writer).
2091 * bug fix: the compiler now traps array references to elements off
2092 the end of an array; previously, the bounds checking in some
2093 circumstances could go off-by-one.
2094 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2096 * type declarations for array element types now obey the description
2097 on the CLHS page "Declaration TYPE", as per discussions on
2098 sbcl-help around 2003-05-08. This means that a declaration
2099 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2100 declaration, all references to BAR will be asserted or assumed
2101 (with THE, so dependent on compiler policy) to involve objects of
2102 type FOO. Note that no such declaration is implied in
2103 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2104 * declared types of functions from the "Conditions"
2105 chapter. (reported by Paul Dietz)
2106 * bug fix: CERROR accepts a function as its first argument.
2107 * bug fix: NTH an NTHCDR accept a bignum as index
2108 arguments. (reported by Adam Warner)
2109 * optimization: character compare routines now optimize comparing
2110 against a constant character. (reported by Gilbert Baumann)
2111 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2112 argument now works properly on the MIPS platform.
2113 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2114 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2115 declared to have a functional type, does not cause an error
2117 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2118 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2119 * fixed bug: if last continuation of a deleted block has a
2120 destination, this destination should be deleted too. (reported by
2122 * fixed a bug in the bootstrap process: the host compiler's values
2123 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2124 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2125 test case from Patrik Nordebo)
2126 * improved the ability of the disassembler on the PPC platform to
2127 provide helpful disassembly notes.
2128 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2129 the class in more cases than previously.
2130 * bug fix: FILE-POSITION now understands :START and :END for
2131 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2132 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2133 8) data. (thanks to David Lichteblau)
2134 * bug fix: it is possible to add a method to a generic function
2135 without lambda list.
2136 * bug fix: reader failed to signal END-OF-FILE inside an
2137 object representation. (reported by Nikodemus Siivola)
2138 * fixed some bugs revealed by Paul Dietz' test suite:
2139 ** LAST and [N]BUTLAST should accept a bignum.
2140 ** condition slot accessors are methods.
2141 ** (VECTOR NIL) is a subtype of STRING.
2143 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2144 * minor incompatible change: some nonsensical specialized lambda
2145 lists (used in DEFMETHOD) which were previously ignored now signal
2147 * minor incompatible change: the system is now aware of the types of
2148 variables in the COMMON-LISP package, and will signal errors for
2149 most violations of these type constraints (where previously they
2150 were silently accepted).
2151 * minor incompatible change: COMPILE-FILE now uses the freedom
2152 afforded (ANSI 3.2.2.3) to use derived function types for
2153 functions defined in the same file. This also permits the system
2154 to warn on static type mismatches and function
2155 redefinition. (Currently it does not work with high DEBUG level.)
2156 * minor incompatible change: VALUES declaration is disabled.
2157 * When issuing notes, the compiler now signals a condition of type
2158 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2159 restart for use in user handlers. It is expected that the
2160 COMPILER-NOTE condition will eventually become a condition
2161 supertype to a hierarchy of note types, which will then be
2162 handleable in a similar fashion. However, at the moment, no such
2163 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2164 but it's an implementation detail, not a classification for the
2166 * Changes in type checking closed the following bugs:
2167 ** type checking of unused values (192b, 194d, 203);
2168 ** template selection based on unsafe type assertions (192c, 236);
2169 ** type checking in branches (194bc).
2170 * A short form of VALUES type specifier has ANSI meaning (it has
2171 increased the number of situations when SBCL cannot perform type
2173 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2174 keywords or constants is permissible.
2175 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2176 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2177 classes. (thanks to Antonio Martinez)
2178 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2179 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2180 argument to a pretty-print stream. Instead, it leaves any such
2181 operations to DESCRIBE-OBJECT methods.
2182 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2184 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2185 lambda lists. (thanks to David Lichteblau)
2186 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2187 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2188 (thanks to Gerd Moellmann)
2189 * fixed bug 246: increased compilation speed of long
2190 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2192 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2194 * DEFCONSTANT now throws a condition of type
2195 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2196 constant to a non-EQL value; CONTINUE and ABORT restarts
2197 respectively change and preserve the value.
2198 * fixed bug 63: The code walker, part of the implementation of CLOS,
2199 is now better at handling symbol macros.
2200 * bug fix: There is no longer an internal implementation type named
2201 CL:LENGTH. (reported by Raymond Toy)
2202 * bug fix: In macro-like defining macros/special operators the
2203 implicit block does not enclose the lambda list.
2204 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2205 enclosed in parentheses) are not suitable as type specifiers, and
2206 their use properly signals an error now.
2207 * bug fix: An argument count mismatch for a type specifier in code
2208 being compiled no longer causes an unhandled error at compile
2209 time, but signals a compile-time warning.
2210 * fixed simple vector readable printing
2211 * bug fix: DESCRIBE takes more care over whether the class
2212 precedence list slot of a class is bound before accessing it.
2213 (reported by Markus Krummenacker)
2214 * bug fix: FORMATTER can successfully compile pretty-printer format
2215 strings which use variants of the ~* directive inside.
2216 * bug fix: SEARCH now applies its TEST predicate to the elements of
2217 the arguments in the correct order. (thanks to Wolfhard Buss)
2218 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2219 in type inference. (thanks to Robert E. Brown)
2220 * bug fix: Reading in symbols with an explicit package name of ""
2221 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2222 * fixed some bugs revealed by Paul Dietz' test suite:
2223 ** NIL is now allowed as a structure slot name.
2224 ** Arbitrary numbers, not just REALs, are allowed in certain
2225 circumstances in LOOP for-as-arithmetic clauses.
2226 ** Multiple class redefinitions before slot access no longer
2227 causes a type error.
2228 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2229 association between the name and a class.
2230 ** Generic functions with non-standard method-combination and over
2231 five methods all of which return constants no longer return NIL
2232 after the first few invocations. (thanks to Gerd Moellmann)
2233 ** CALL-NEXT-METHOD with no arguments now passes the original
2234 values of the arguments, even in the presence of assignment.
2235 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2237 ** DISASSEMBLE works with closures and funcallable instances.
2238 ** ADD-METHOD now returns the generic function, not the new method.
2239 ** FIND-METHOD signals an error if the lengths of the specializers
2240 is incompatible with the generic function, even if the ERRORP
2242 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2243 which its argument is a member.
2244 ** DEFCLASS only redefines the class named by its class-name
2245 argument if that name is the proper name of the class;
2246 otherwise, it creates a new class.
2247 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2248 of the UNBOUND-SLOT condition to the name of the slot.
2249 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2250 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2251 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2252 SLOT-MAKUNBOUND in the specified fashion.
2254 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2255 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2256 cross-compilation host. As a consequence, we can now bootstrap our
2257 way up to SBCL starting with a bare gcc toolchain and human-readable
2258 source code (first the source to CLISP, then the source to SBCL).
2259 * A contributed module containing a partial implementation of the
2260 simple-streams interface has been included. (thanks to Rudi
2262 * A contributed module implementing the RFC1321 Message Digest
2263 Algorithm, known as MD5, has been included.
2264 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2265 no longer has any effect, as the code controlled by this feature
2266 has been deleted. (As far as we know, no-one has ever built using
2267 this feature, and its semantics were confused in any case).
2268 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2269 work on conditions (as required by the ANSI specification),
2270 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2271 expected behaviour on conditions. Users should note, however,
2272 that such behaviour is not required by the ANSI specification,
2273 and so use of this behaviour may render their code unportable.
2274 * fixed some bugs revealed by Paul Dietz' test suite:
2275 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2277 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2278 ** SXHASH on condition objects no longer returns NIL.
2279 ** :ALLOCATION :CLASS slots are better treated; their values are
2280 updated on class redefinition, and initforms inherited from
2281 superclasses are applied.
2282 ** REMOVE-METHOD returns its generic function argument even when
2283 no method was removed.
2284 ** SHARED-INITIALIZE now initializes the values of the requested
2285 slots, including those with :ALLOCATION :CLASS.
2286 ** ALLOCATE-INSTANCE now works on structure classes defined via
2287 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2289 ** SLOT-EXISTS-P now works on conditions, as well as structures
2291 ** MAKE-LOAD-FORM now has the required methods on
2292 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2293 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2294 keyword, and now implements the SLOT-NAMES argument.
2295 ** methods with &OPTIONAL arguments no longer allow too many
2296 arguments to be passed in the call without error.
2297 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2298 option is consistent with the required arguments of the generic
2299 function lambda list.
2300 * bug fix: REQUIRE accepts a string designator. (Thanks to
2302 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2303 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2304 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2305 * bug fix: The compiler now checks for duplicated variables in macro
2307 * bug fix: SETQ on globals returns the correct value.
2308 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2309 not a COMPILER-ERROR (followed by some other strange error on
2310 choosing the CONTINUE restart).
2311 * bug fix: make.sh and friends are now more consistent in the way that
2312 they look for GNU "make".
2314 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2315 * experimental native threads support (on x86 Linux >=2.4 only).
2316 This is not compiled in by default: you need to add :SB-THREAD to
2317 the target features. See the "Beyond ANSI" chapter of the manual
2319 * fix for longstanding nonANSIism: The old distinction between
2320 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2321 The return value from CL:FIND-CLASS is now a CLOS class, and
2322 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2323 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2324 * An interface to the MetaObject Protocol, as described in Kiczales,
2325 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2326 MIT Press, 1991, is available from the SB-MOP package.
2327 * incompatible change: the SB-PCL package should now be considered
2328 a private implementation detail, and no longer a semi-private MOP
2330 * minor incompatible change: due to rearrangement for threads, the
2331 control stack and binding stack are now allocated at arbitrary
2332 addresses instead of being hardcoded per-port. Users affected by
2333 this probably have to be doing advanced things with shared
2334 libraries, and will know who they are.
2335 * minor incompatible change: Previously, all --eval forms used were
2336 processed with READ before any of them were processed with EVAL.
2337 Now each --eval form is processed with both READ and EVAL before
2338 the next --eval form is processed. (Thus package operations like
2339 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2340 work as the user might reasonably expect.)
2341 * minor incompatible change: *STANDARD-INPUT* is now only an
2342 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2344 * minor incompatible change: Y-OR-N-P is now character-oriented,
2345 not line oriented. Also, YES-OR-NO-P now works without errors.
2346 (thanks to Antonio Martinez)
2347 * sb-aclrepl module improvements: an integrated inspector, added
2348 repl features, and a bug fix to :trace command.
2349 * Known functions, which cannot be open coded by the backend, are
2350 considered to be able to check types of their arguments. (fixing
2351 a bug report by Nathan J. Froyd)
2352 * fixed a bug in computing method discriminating functions: It is
2353 now possible to define methods specialized on classes which have
2354 forward-referenced superclasses. (thanks to Gerd Moellmann)
2355 * fixed evaluation order in optional entries (reported by Gilbert
2357 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2358 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2359 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2360 required-by-AMOP NAME argument, as well as CLASS and
2361 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2362 * fixed bug 20: DEFMETHOD can define methods using names that are
2363 not the proper names of classes to designate class specializers.
2364 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2365 Unix file descriptors, instead of blowing up. (thanks to Antonio
2367 * Garbage collection refactoring: user-visible change is that a
2368 call to the GC function during WITHOUT-GCING will not do garbage
2369 collection until the end of the WITHOUT-GCING. If you were doing
2370 this you were probably losing anyway.
2371 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2372 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2373 TYPEP the latter but not the former.
2374 * The compiler issues a full WARNING for calls to undefined functions
2375 with names from the CL package.
2376 * MAP-INTO for a vector destination is open coded. (reported by
2377 Brian Downing on c.l.l)
2378 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2379 documentation string.
2380 * fixed some bugs revealed by Paul Dietz' test suite:
2381 ** COPY-ALIST now signals an error if its argument is a dotted
2383 ** Condition slots are now accessed more correctly in the presence
2384 of multiple initargs for a given slot.
2385 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2386 exclude restarts of the same name associated with a different
2388 ** DEFCLASS of forward-referenced classes with another
2389 forward-referenced class in the superclasses list no longer
2391 ** Condition slots are now initialized once each, not multiple
2392 times. (thanks to Gerd Moellmann)
2393 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2395 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2396 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2397 arguments contain duplicated elements.
2398 ** RESTART-CASE understands local macros.
2399 ** RESTART-CASE associates exactly its own restarts with a condition.
2400 ** ENDP in safe mode checks its argument to be of type LIST.
2401 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2402 in question is unbound.
2403 ** Optimized MAKE-INSTANCE functions no longer cause internal
2404 assertion failures in the presence of duplicate initargs.
2405 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2406 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2407 propagated even in the ctor optimized implementation of
2409 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2411 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2412 * a better implementation of SXHASH on (simple) bit vectors,
2413 measured both in execution speed and in distribution of results
2414 over the positive fixnums, has been installed. Likewise, a better
2415 implementation of EQUAL for simple bit vectors is now available.
2416 * fixed CEILING optimization for a divisor of form 2^k.
2417 * fixed bug 240 (emitting extra style warnings "using the lexical
2418 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2419 by Antonio Martinez)
2420 * fixed SXHASH, giving different results for NIL depending on type
2421 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2422 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2423 variable. (found by Rolf Wester)
2424 * SBCL does not ignore type declarations for special
2425 variables. (reported by rif on c.l.l 2003-03-05)
2426 * some bug fixes in contrib/sb-aclrepl/
2427 * fixed some bugs revealed by Paul Dietz' test suite:
2428 ** a bug in the CONS type specifier, whereby the CAR and CDR
2429 types got intertwined, has been fixed;
2430 ** the type system is now able to reason about the interaction
2431 between INTEGER and RATIO types more completely;
2432 ** APPEND, [N]REVERSE and NRECONC check that those their
2433 arguments, which must be proper lists, are really so;
2434 ** An array specialized to be unable to hold elements has been
2435 implemented, as required -- yes, really -- by ANSI;
2436 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2437 on malformed property lists;
2439 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2440 * incompatible packaging change: in line with Unix convention,
2441 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2442 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2443 * REQUIRE and PROVIDE are now optionally capable of doing something
2444 useful. See the documentation string for REQUIRE.
2445 * infrastructure for a managed SBCL contrib system: contributed
2446 modules in this release include:
2447 ** the ASDF system definition facility;
2448 ** an interface to the BSD Sockets API;
2449 ** an ACL-like convenience interface to the repl;
2450 (thanks to Kevin Rosenberg)
2451 ** an implementation of ROTATE-BYTE, with efficient implementation
2453 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2454 gives a closer approximation to the right answer than previously.
2455 (thanks to Raymond Toy)
2456 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2457 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2458 optional environment arguments, as required by ANSI.
2459 * fixed bugs in other functions taking environment objects, allowing
2460 calls with an explicit NIL environment argument to be compiled
2462 * fixed bug 228: primary return values from
2463 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2464 COMPILE or FUNCTION.
2465 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2466 :TYPE will no longer signal errors on innocuous objects.
2467 * fixed bug 231b: SETQ is better at respecting type declarations in
2468 the lexical environment.
2469 * fixed a bug in DEFCLASS: classes named by symbols with no or
2470 unprintable packages can now be defined.
2471 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2472 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2473 * fixed compiler failure related to checking types of functions.
2474 (reported by Robert E. Brown)
2475 * the compiler is now much more consistent in its error-checking
2476 treatment of bounding index arguments to sequence functions: in
2477 (SAFETY 3) code, errors will be signalled in almost all cases if
2478 invalid sequence bounding indices are passed to functions defined
2479 by ANSI to operate on sequences.
2480 * fixed a bug in the build procedure: documentation of SBCL-specific
2481 packages is now preserved and available in the final Lisp image.
2482 * lifted FDEFINITION lookup out of loops in the implementation of
2483 many list operations. (thanks to Robert E. Brown)
2484 * fixed a bug in the reader: the #n# reader macro now works for
2485 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2486 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2487 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2488 E. Brown and Tony Martinez respectively)
2489 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2490 SLOT-BOUNDP: in particular, it is now able to optimize them much
2491 better, and is now not vulnerable to having packages renamed.
2492 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2493 always called when a slot is not present in an instance. (thanks
2495 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2496 longer signal an error on unreading a character following EOF on
2497 the previous constituent stream. (thanks to Tony Martinez)
2498 * fixed some bugs revealed by Paul Dietz' test suite:
2499 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2500 not just nonnegative fixnums;
2501 ** the logical bit-array operators such as BIT-AND now accept an
2502 explicit NIL for their "opt-arg" argument (to indicate a
2503 freshly-consed result bit-array);
2504 ** ELT now signals an error on an invalid sequence index in safe
2506 ** the type system is now cleverer about negations of numeric
2507 types, and consequently understands the BIGNUM and RATIO types
2509 ** the type system is now cleverer about the interaction between
2510 INTEGER and RATIO types: while bugs still remain, many more
2511 cases are accurately computed;
2512 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2513 if it is in the last clause;
2514 ** CONSTANTLY now correctly returns a side-effect-free function in
2516 ** DECLARE is no longer treated as a special-operator; in
2517 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2518 * incremented fasl file version number due to the change in the
2519 DEFSTRUCT-SLOT-DESCRIPTION structure.
2521 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2522 * minor incompatible change: code processed by the "interpreter" or
2523 EVAL now has a compilation optimization policy of (DEBUG 2)
2524 (changed from (DEBUG 1)) to improve debuggability of interactive
2525 development, and to allow the use of the debug RETURN command in
2527 * an experimental implementation of the RETURN command for the
2528 debugger has been included. (thanks to Frederik Kuivinen)
2529 * fixed bug 62: constraints were not propagated into a loop.
2530 * fixed bug in embedded calls of SORT (reported and investigated by
2532 * fixed some bugs revealed by Paul Dietz' test suite:
2533 ** printing and reading of arrays with some dimensions having
2534 length 0 (thanks to Gerd Moellmann);
2535 ** BOA constructor with &AUX argument without a default value does
2536 not cause a type error;
2537 ** CONSTANTP now returns true for all self-evaluating objects.
2539 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2540 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2541 accessors that are related by inheritance, as specified in the
2542 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2543 Valtteri Vuorikoski)
2544 * The compiler is now able to inline functions that were defined in
2545 a complex lexical environment (e.g. inside a MACROLET).
2546 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2548 * Support for the upcoming FreeBSD-5.0 release has been included.
2549 (thanks to Dag-Erling Smorgrav)
2550 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2551 effect when it is not in a toplevel context.
2552 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2553 stand a better chance of being correct. (thanks to Gerd
2555 * fixed bug in COERCE, which now signals an error on coercing a
2556 rational to a bounded real type which excludes the expected
2558 * The compiler is now able to derive types more accurately from the
2559 COERCE and COMPILE functions.
2560 * fixed bug 223: functional binding is considered to be constant
2561 only for symbols in the CL package.
2562 * fixed bug 231: SETQ did not check the type of a variable being set
2563 (reported by Robert E. Brown)
2564 * A new optimization for MAKE-INSTANCE has been included, fixing
2565 various bugs (including relating to :ALLOCATION :CLASS slots and
2566 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2568 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2569 ** As required by ANSI, LOOP now disallows anonymous collection
2570 clauses such as COLLECT I in conjunction with aggregate boolean
2571 clauses such as THEREIS (= I 1);
2572 ** LOOP now signals an error when any variable is reused in the
2573 same loop (including the potentially useful construct analogous
2574 to WITH A = 1 WITH A = (1+ A);
2575 ** IT is only a special loop symbol within the first clause of a
2576 conditional loop clause;
2577 ** LOOP with a typed iteration variable over a hashtable now
2578 signals a type error iff it should.
2579 * fixed some other bugs revealed by Paul Dietz' test suite:
2580 ** FILE-STREAM now names the class previously known as FD-STREAM;
2581 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2582 argument) no longer signals an error;
2583 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2584 of the slot symbol, rather than using the current package
2585 ((:CONC-NAME "") continues to intern the slot's name in the
2587 * incremented fasl file version number, because of the incompatible
2588 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2589 of the new implementation of DEFINE-COMPILER-MACRO.
2591 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2592 * Support for building SBCL for MIPS platforms running in
2593 little-endian mode has now been checked in, and basic
2594 functionality on said platforms verified.
2595 * minor incompatible change: PCL now records the pathname of a file
2596 in which methods and the like are defined, rather than its
2598 * minor incompatible change: TRUENAME now considers the truename of
2599 a file naming a directory to be the pathname with :DIRECTORY
2600 component indicating that directory.
2601 * minor incompatible change: a NAMED clause in the extended form of
2602 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2603 reason for the previous behaviour is unclear.
2604 * more systematization and improvement of CLOS and MOP conformance
2605 in PCL (thanks to Gerd Moellman and Pierre Mai):
2606 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2608 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2609 primary methods with no specializers;
2610 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2612 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2613 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2614 FORWARD-REFERENCED-CLASSes; error reporting on
2615 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2617 ** SXHASH on CLOS instances now uses a slot internal to the
2618 instance to return different numbers on distinct instances,
2619 while preserving the same return value through invocations of
2621 ** DEFMETHOD signals errors when methods with longer incongruent
2622 lambda lists are added to generic functions;
2623 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2624 CLASS, as specified in AMOP;
2625 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2626 on the order returned by the primary method for classes of
2627 class STANDARD-CLASS;
2628 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2629 * fixed some bugs shown by Paul Dietz' test suite:
2630 ** DOLIST puts its body in TAGBODY;
2631 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2633 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2634 value producing form;
2635 ** if more variables are given to PROGV than values, extra
2636 variables are bound and made to have no value;
2637 ** NSUBSTITUTE on list arguments gets the right answer with
2639 ** ELT signals an error of type TYPE-ERROR when the index argument
2640 is not a valid sequence index;
2641 ** LOOP signals (at macroexpansion time) an error of type
2642 PROGRAM-ERROR when duplicate variable names are found;
2643 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2644 ** FUNCALL of special-operators now cause an error of type
2646 ** PSETQ now works as required in the presence of side-effecting
2647 symbol-macro places;
2648 ** NCONC accepts any object as its last argument;
2649 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2651 ** loop-for-as-package does not require a package to be explicitely
2653 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2655 * fixed bug 166: compiler preserves "there is a way to go"
2656 invariant when deleting code.
2657 * fixed bug 172: macro lambda lists with required arguments after
2658 &REST arguments now cause an error to be signalled. (thanks to
2660 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2661 bug. (thanks to Matthew Danish)
2662 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2664 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2665 arguments are lexically rebound. (thanks to Gerd Moellmann and
2667 * fixed bug 194: error messages are now more informative when there
2668 is no primary method applicable in a call to a generic
2669 function. (thanks to Gerd Moellmann)
2670 * fixed bug in command line argument checking (thanks to Julian
2672 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2673 (thanks to Matthew Danish)
2674 * incremented fasl file version number, because of the
2675 SXHASH-related changes in the layout of CLOS data structures
2677 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2678 * minor incompatible change: The runtime (the Unix executable named
2679 "sbcl") is now much pickier about the .core files it will load.
2680 Essentially it now requires .core files to descend from the same
2681 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2682 as the runtime does. (The intent is to prevent the crashes which
2683 can occur, and which can even be reported as mysterious failures,
2684 when people patch the sources or change the build parameters
2685 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2686 sbcl and .core files.)
2687 * fixed bug: VALUES-LIST is no longer optimized away.
2688 * fixed bug 142: The FFI conversion of C string values to Lisp
2689 string values no longer conses excessively. (thanks to Nathan
2690 Froyd porting Raymond Toy's fix to CMU CL)
2691 * began to systematize and improve MOP conformance in PCL (thanks to
2692 Nathan Froyd, Gerd Moellman and Pierre Mai):
2693 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2695 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2696 ** FINALIZE-INHERITANCE is now called on class finalization;
2697 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2698 argument precedence order.
2699 * fixed bug 202: The compiler no longer fails on functions whose
2700 derived types contradict their declared type.
2701 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2702 so it can be non-toplevel.
2703 * The fasl file version number has changed (because of the new
2704 implementation of DEFMACRO).
2705 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2706 safe code, that any length requirement by their type-specifier
2707 argument is valid. The exceptions to this are described in bug
2709 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2710 functions now check the types of their inputs as required by ANSI.
2711 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2712 introduced symbol is DECLAREd to be SPECIAL.
2713 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2714 * fixed bug: :COUNT argument to sequence functions may be negative
2715 * fixed bug: body of DO-SYMBOLS may contain declarations
2716 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2717 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2718 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2719 symbol macro only once
2720 * fixed printing of call frame when argument list is unavailable
2721 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2722 * compiler no longer signals WARNING on unknown keyword
2725 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2726 * A beta-quality port to the mips architecture running Linux,
2727 based on the old CMUCL backend, has been made. It has been tested
2728 on a big-endian kernel, and works sufficiently well to be able to
2729 rebuild itself; it has not been tested in little-endian mode.
2730 * fixed an inconsistency between gencgc.c and purify.c which made
2731 dumping/loading .core files unreliable
2732 * fixed bug 120a: The compiler now deals correctly with IFs where
2733 the consequent is the same as the alternative, instead of
2734 misderiving the return type. (thanks to Alexey Dejneka)
2735 * fixed bug 113: Logical pathnames are now dumpable (the logical
2736 host is resolved at load-time, throwing an error if it is not
2738 * fixed bug 174: FORMAT's error message is slightly clearer when a
2739 non-printing character is used in a format directive.
2740 * fixed several bugs in compiler checking of type declarations, i.e.
2741 violations of the Python "declarations are assertions" principle
2742 (thanks to Alexey Dejneka)
2743 * fixed several bugs in PCL's error checking (thanks to Gerd
2745 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2747 * fixed bug in compilation of functions as first class values
2748 (thanks to Antonio Martinez-Shotton)
2749 * The compiler's handling TYPE-ERRORs which it can prove will
2750 inevitably happen at runtime has been cleaned up and corrected
2751 in several ways. (thanks to Alexey Dejneka)
2752 * improved argument type checking for various basic arithmetic
2753 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2754 lavished on them in the past that they can be compiled in many
2755 ways in different special cases
2756 * fixed bug 181: compiler checks validity of user supplied type
2758 * cleaned up code flushing in optimization: Function calls which
2759 should signal errors for safety purposes (e.g. which ANSI says
2760 should signal errors when their arguments are of incorrect type)
2761 are no longer optimized away.
2762 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2763 * incremented fasl file version number, because changes in the
2764 implementation of sequence functions like COERCE caused
2765 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2766 in old inline expansions) to become undefined. (Actually these
2767 changes were later undone, so we might very well be binary
2768 compatible with 0.7.7 after all, but leaving the version number
2769 incremented seemed like the simplest and most conservative
2772 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2773 * An alpha-quality port to the parisc architecture running Linux,
2774 based on the old CMUCL backend, has been made. This, even more so
2775 than the other backends, should be considered still a work in
2776 progress; known problems include that the Linux kernel in 64-bit
2777 mode does not propagate the correct sigcontext structure to
2778 userspace, and consequently SBCL on a parisc64 kernel will not
2780 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2781 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2782 suppressed.) Also now that the compiler is looking at declarations
2783 in the environment, it checks optimization declarations as well,
2784 and suppresses inlining when (> DEBUG SPEED).
2785 * More fixes have been made to treatment of floating point exception
2786 treatment and other Unix signals. In particular, floating point
2787 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2788 * The detection and handling of control stack exhaustion (infinite
2789 or very deeply nested recursion) has changed. Stack exhaustion
2790 detection is now done by write-protecting pages at the OS level
2791 and applies at all optimization settings; when found, a
2792 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2793 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2794 be caught using IGNORE-ERRORS.
2795 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2796 correctly when transforming recursive LABELS functions to LETs.
2797 (thanks to Alexey Dejneka)
2798 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2799 that are names of constants or global variables.
2800 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2801 alien routines with docstrings.
2802 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2803 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2805 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2806 (thanks to Eric Marsden)
2807 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2808 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2809 * Bug fix: The SPARC backend can now compile functions involving
2810 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2811 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2812 to a functions expecting a PATHNAME-DESIGNATOR.
2813 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2814 lambda lists. (thanks to Alexey Dejneka)
2815 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2816 functions so that EXPORTing the name of the class doesn't cause
2817 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2818 to Antonio Martinez for reporting this)
2819 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2820 constant-folded without failing an assertion. (thanks to Einar
2821 Floystad Dorum for reporting this)
2822 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2823 types are now dealt with more correctly by the compiler.
2824 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2825 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2826 behaviour analogous to RENAME-FILE. This puts its behaviour more
2827 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2829 * The fasl file version number has changed again. (because of the
2830 bug fix involving the names of PCL MAKE-INSTANCE functions)
2832 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2833 * bug fix: Floating point exceptions are treated much more
2834 consistently on the x86/Linux and PPC/Linux platforms.
2835 * Array initialization with :INITIAL-ELEMENT is now much faster for
2836 cases when the compiler cannot open code the array creation, but
2837 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2838 array accesses have also seen a speed increase.
2839 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2840 specified. (This was at the root of some bad interactions between
2841 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2843 * bug fix: Internal error arguments for undefined functions are now
2844 computed correctly on the PPC/Linux platform.
2845 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2846 Raymond Toy's patch for CMU CL)
2847 * Support for the Solaris 9 operating environment has been included
2848 (thanks to Daniel Merritt)
2849 * A very ugly but hopefully complete draft of the missing FFI chapter
2850 of the manual has been created by reformatting the corresponding
2851 CMU CL manual chapter into (currently very ugly and incoherent)
2852 DocBook and bringing it up to date for SBCL behavior. Thus, the
2853 manual is now essentially complete, at least by my extreme
2854 once-and-only-once standards, whereby it's acceptable to refer to
2855 the doc strings of SB-EXT functions as the primary documentation.
2856 * The fasl file version number has changed again, due to cleanup of
2857 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2858 is no longer a static symbol.)
2860 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2861 * SBCL now builds with OpenMCL (version 0.12) as the
2862 cross-compilation host; also, more progress has been made toward
2863 bootstrapping under CLISP.
2864 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2866 * bug 158 fixed: The compiler can now deal with integer loop
2867 increments different from 1; fixing this turned out also to fix
2869 * bug 169 fixed: no more bogus warnings about using lexical bindings
2870 despite the presence of perfectly good SPECIAL declarations (thanks
2871 to David Lichteblau)
2872 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2873 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2874 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2876 * bug 180 fixed: Method combination specifications no longer ignore
2877 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2878 * bug fix: Structure type predicate functions now check their argument
2879 count as they should.
2880 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2881 correctly. (thanks to Pierre Mai)
2882 * minor incompatible change: The --noprogrammer option is deprecated
2883 in favor of the new --disable-debugger option, which is very similar.
2884 (The major difference is that it takes effect at a slightly different
2885 time at startup, causing handling of errors in --sysinit and
2886 --userinit files will be affected differently.) The
2887 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2888 been added to allow this functionality to be controlled from ordinary
2889 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2890 maintainers, who might want to run non-interactive scripts to
2891 build SBCL cores which will later be used interactively.)
2892 * minor incompatible change: The LOAD function no longer, when given
2893 a wild pathname to load, loads all files matching that pathname.
2894 Instead, an error of type FILE-ERROR is signalled.
2896 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2897 * bug 147 fixed: The compiler preserves its block link/count
2898 invariants more correctly now so that it doesn't crash. (thanks
2900 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2902 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2903 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2904 no-value-for-this-character as NIL (as specified by ANSI).
2905 * HOST-NAMESTRING on physical pathnames now returns a string that is
2906 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2907 (thanks to Christophe Rhodes)
2908 * The Alpha port handles icache flushing more correctly. (thanks to
2910 * More progress has been made toward bootstrapping under CLISP. (thanks
2911 to Christophe Rhodes)
2912 * The fasl file format has changed again, because dynamic loading
2913 on OpenBSD (which has non-ELF object files) motivated some cleanups
2914 in the way that foreign symbols are transformed and passed around.
2915 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2916 is no longer treated as whitespace by the reader, but instead as
2917 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2918 |A
\7fB|, instead of A as it used to.
2920 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2921 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2922 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2923 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2924 this as of 0.7.1.45, but was left out of the previous news section
2925 (thanks to Dan Barlow)
2926 * SBCL now runs on the Solaris operating system on SPARC architectures
2927 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2928 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2929 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2930 * SPARC backend cleanups, allowing builds of cores optimized for V8
2931 and V9 SPARCS, and also emission of code targeted to a particular
2932 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2934 * SBCL is closer to bootstrapping under CLISP, thanks to various
2935 fixes by Christophe Rhodes.
2936 * The fasl file format has changed again, to allow the compiler's
2937 INFO database to support symbol macros.
2938 * The user manual (in doc/) is formatted into HTML more nicely.
2939 (thanks to coreythomas)
2940 * The system is smarter about SUBTYPEP relationships, especially
2941 those involving NOT types (including types such as ATOM which are
2942 represented internally using NOT types). Thus SUBTYPEP is less
2943 likely to return (VALUES NIL NIL) in general, and in particular
2944 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2946 * The fasl file format has changed again, because the internal
2947 representation of types now includes a new slot to support the new
2948 SUBTYPEP-of-NOT-types logic.
2949 * (not a change in the main branch of SBCL, but a related prototype
2950 which can hopefully be merged into the main branch of SBCL in the
2951 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2952 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2953 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2954 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2955 dispatch macro character combinations. (thanks to Alexey Dejneka)
2956 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2957 that superficially look like logical namestrings correctly.
2958 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2960 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2961 default supplied-p) for &optional and &key arguments. (thanks to
2964 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2965 * incompatible change: The compiler is now less aggressive about
2966 tail call optimization, doing it only when (> SPACE DEBUG) or
2967 (> SPEED DEBUG). (This is an incompatible change because there are
2968 programs which relied on the old CMU-CL-style behavior to optimize
2969 away their unbounded recursion which will now die of stack overflow.)
2970 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2971 for non-GENCGC systems has been increased to 20M (since that
2972 seems much closer to the likely performance optimum for modern
2973 systems than the old 4M value was)
2974 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2975 and *DEBUG-PRINT-LEVEL*
2976 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2977 of CMU CL's support for SPARC, and various endianness and other
2978 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2979 * new syntactic sugar for the Unix command line: --load foo.bar is now
2980 an alternate notation for --eval '(load "foo.bar")'.
2982 ** The system now detects stack overflow and handles it gracefully,
2983 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2984 optimization settings. (This is a good thing in general, and
2985 its introduction in this version should be particularly timely
2986 for anyone whose code fails because of suppression of tail
2988 ** The system now hunts for the C variable "environ" in a more
2989 devious way, to avoid segfaults when the C library version differs
2990 between compile time and run time. (thanks to Christophe Rhodes)
2991 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2992 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2993 ** The compiler no longer issues bogus style warnings for undefined
2994 classes in the same source file as the DEFCLASSes which defined
2995 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2997 ** fixes in CONDITION class precedence list for undefined function
2998 errors (thanks to Alexei Dejneka)
2999 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3000 correctly. (thanks to Dan Barlow)
3001 ** portability fixes aiming at bootstrapping under CLISP (thanks
3002 to Dave McDonald and Christophe Rhodes)
3003 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3004 ** fixes in type translation and and type inference (thanks to
3006 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3007 ** various fixes in the new ports (thanks to Dan Barlow)
3008 * several changes related to debugging:
3009 ** suppression of tail recursion, as noted above
3010 ** stack overflow detection, as noted above
3011 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3012 is now the default. (For some time encapsulation has been more
3013 reliable than the breakpoint-based :ENCAPSULATE NIL
3014 implementation, at least on X86 systems; and I just noticed that
3015 encapsulation also seems closer to the spirit of the ANSI
3018 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3020 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3021 up properly again. (There was a packaging bug in 0.7.0 which
3022 left their definitions in SB-SYS::LOAD-FOREIGN and
3023 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3024 vital for most things which interface to C-level interfaces,
3025 like extensions working with sockets or databases or
3026 Perl-compatible regexes or whatever, and the need to fix
3027 this bug is the main reason that 0.7.1 was released so
3029 ** DEFGENERIC is now choosier about the methods it redefines, so that
3030 reLOADing a previously-LOADed file containing DEFGENERICs does
3031 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3032 cycle now works as it should. (thanks to Alexey Dejneka)
3033 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3034 fixed by Christophe Rhodes. (He actually submitted this patch
3035 months ago, and I delayed until after 0.7.0.)
3036 ** Bug 111 (internal compiler confusion about runtime checks on
3037 FUNCTION types) was fixed by Alexey Dejneka.
3038 * Some internal cleanups (getting rid of variables which aren't
3039 needed now that the byte interpreter is gone) caused the fasl
3040 file format number to change again.
3042 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3043 * major incompatible change: The default fasl file extension, i.e. the
3044 default extension for files produced by COMPILE-FILE, has changed
3045 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3047 ** There are many changes in the implementation of the compiler.
3048 SBCL is now essentially a compiler-only implementation of ANSI
3049 Common Lisp. EVAL still "interprets" a few special cases, but
3050 almost all the interesting cases are handled by creating
3051 a LAMBDA expression, calling COMPILE on it, then calling
3052 FUNCALL on the result.
3053 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3054 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3055 Since the code is newer, there might still be some new bugs
3056 (though not as many as before Martin Atzmueller's fixes:-). But
3057 the new code is substantially simpler and clearer, and hopefully
3058 any remaining bugs will be simpler, less fundamental, and more
3059 fixable then the bugs in the old code.
3060 ** The revised compiler is still a little unsteady on its feet.
3062 *** The debugging information it produces (particularly the names
3063 of FUNCTION objects) is sometimes much less useful than what
3064 the old compiler produced.
3065 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3066 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3067 within a MACROLET) has been temporarily weakened.
3068 ** There are new compiler optimizations for various functions:
3069 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3070 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3071 *** the math functions TRUNCATE, FLOOR, and CEILING
3072 *** the function-of-all-trades COERCE
3073 Mostly these should be transparent, but there's one
3074 potentially-annoying problem (bug 117): when the compiler
3075 inline-expands a function and does type analysis on the result,
3076 it can create control paths which have type mismatches, and
3077 when it can't prove that those control paths aren't taken,
3078 it will issue WARNINGs about the type mismatches. This is
3079 a particular problem in practice for the new sequence functions.
3080 It's not clear how this should be fixed, and for now, a
3081 workaround is given in the entry for 117 in the BUGS file.
3082 ** (Because of the interaction between the two previous items --
3083 occasional inlining problems and new inline expansions -- some
3084 of the new sequence function optimizations won't really kick in
3085 completely until debugging information, and then inlining, are
3086 straightened out in some future version.)
3087 * minor incompatible changes:
3088 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3089 DIRECTORY no longer implicitly promotes NIL slots of its
3090 pathname argument to :WILD. In particular, when you ask for the
3091 contents of a directory (which you used to be able to do without
3092 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3093 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3094 ** changes in behavior that ANSI explicitly defines to be
3095 implementation dependent:
3096 *** The new compiler-only implementation still conforms with ANSI,
3097 but acts a little different than before. Besides the obvious
3098 changes in performance tradeoffs (that the cost per form passed
3099 to EVAL has gone up, and the cost per form executed by EVAL
3100 has gone down), the behavior of the system changes a little
3101 because there are no longer any interpreted function objects.
3102 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3103 e.g. doing COMPILE on the output of interactive DEFUN is
3105 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3107 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3108 has changed from (:CL) to NIL.
3109 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3110 e.g. #\Newline and #\Tab, don't have names specified in the
3111 ANSI Common Lisp standard, is now based on their ASCII symbolic
3112 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3113 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3114 are no longer used for output.
3115 ** changes in internal implementation constants:
3116 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3117 4 million. (If your application spends a lot of time GCing and
3118 you have a lot of RAM, you might want to experiment with
3119 increasing it even more.)
3120 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3121 However, almost all old code should still continue to work without
3122 immediate update, as SB-C-CALL is now a (deprecated) nickname
3124 ** Old operator names in the style DEF-FOO are now deprecated in
3125 favor of new corresponding names DEFINE-FOO, for consistency with
3126 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3127 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3128 internal symbols, but a few supported extensions like
3129 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3130 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3131 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3132 etc. as you get deeper into recursive calls to the debugger
3133 command loop, instead of the old "5]", "5]]", "5]]]"
3134 sequence. (I was motivated to do this when squabbles between
3135 ILISP and SBCL left me very deeply nested in the debugger. In the
3136 short term, this change will probably provoke more ILISP/SBCL
3137 squabbles, but hopefully it will be an improvement in the long run.)
3138 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3139 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3140 function, since declaiming return types involving aliens is
3141 (1) annoyingly messy to do by hand and (2) vital to efficient
3142 compilation of code which calls such functions.
3143 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3144 longer reexported by the SB-EXT package. They're solely useful
3145 for alien code, so it seems more logical that you should get
3146 them from the SB-ALIEN package, not in SB-EXT.
3147 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3148 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3149 features. Instead, the code that they used to control is always
3150 built into the system.
3151 * many other bug fixes
3152 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3153 advantage of the new EVAL-WHEN stuff and to clean them up in
3154 general, and they are now more ANSI-compliant in a number of
3155 ways. Martin Atzmueller is responsible for a lot of this.
3156 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3158 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3159 *** correct ERROR type for various file operations
3160 *** some fixes for Lisp streams
3161 *** DEFMETHOD syntax checking
3162 *** changing old weird representation of debug information as
3163 strings (which, among their other deficiencies, don't transform
3164 correctly when you rename packages, and don't change their
3165 print representation when you change things like *PACKAGE*
3166 and *PRINT-LENGTH*) to symbols and lists of symbols
3167 He also made several improvements and fixed several bugs in DESCRIBE.
3168 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3170 *** misbehavior of WRITE-STRING/WRITE-LINE
3171 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3172 and several other LOOP problems as well
3173 *** DIRECTORY when similar filenames are present
3174 *** DEFGENERIC with :METHOD options
3175 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3176 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3177 *** argument ordering in FIND with :TEST option
3178 *** mishandled package designator argument in APROPOS-LIST
3179 *** various problems in the backquote readmacro
3180 *** a bug in APROPOS
3181 *** probably some others that I'm not describing very well here,
3182 since the CVS log documents them by reference to sbcl-devel
3183 messages, and the SourceForge archives aren't working well.:-(
3184 ** Dan Barlow improved the Alpha port (and is making progress on the
3185 PPC port, for those of you who think different).
3186 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3187 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3188 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3189 continues to work on the SPARC port (for those of you in a position
3190 to look down upon our little PC-compatible boxes from a great height).
3191 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3192 and uses that copy, rather than the current dynamic value, when
3193 it's trying to decide whether to truncate output. Thus e.g.
3194 (let ((*print-lines* 50))
3195 (pprint-logical-block (stream nil)
3197 (let ((*print-lines* 8))
3198 (print (aref possiblybigthings i) stream)))))
3199 should now truncate the logical block only at 50 lines, instead of
3200 often truncating it at 8 lines, as it did before.
3201 * The doc/cmucl/ directory, containing old CMU CL documentation
3202 from the time of the fork, is no longer part of the base system.
3203 SourceForge has shut down its anonymous FTP service, and with it
3204 my original plan for distributing the old CMU CL documentation
3205 there. For now, if you need these files you can download an old
3206 SBCL source release and extract them from it.
3207 * The fasl file version number changed again, for dozens of reasons,
3208 some of which are apparent above.
3210 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3211 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3212 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3213 patch, so that DISASSEMBLE output is much nicer.
3214 * The code in the SB-PROFILE package now seems reasonably stable.
3215 I still haven't decided what the final interface should look like
3216 (I'd like PROFILE to interact cleanly with TRACE, since both
3217 facilities use function encapsulation) but if you have a need
3218 for profiling now, you can probably use it successfully with
3219 the current CMU-CL-style interface.
3220 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3221 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3222 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3223 link now returns the dangling link itself, and for similar
3224 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3225 link itself. (In these cases the old code signalled an error and
3226 looped endlessly, respectively.) Thus, DIRECTORY now works even
3227 in the presence of dangling and cyclic symbolic links.
3228 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3229 is now a supported extension again, since the consensus on
3230 sbcl-devel was that it can be useful for ordinary development
3231 work, not just for debugging SBCL itself.
3232 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3233 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3234 that currently-defined functions might be redefined later with
3235 different return types.
3236 * Hash tables can be printed readably, as inspired by CMU CL code
3237 of Eric Marsden and SBCL code of Martin Atzmueller.
3238 * better error handling in CLOS method combination, thanks to
3239 Martin Atzmueller porting Pierre Mai's CMU CL patches
3240 * more overflow fixes for >16Mbyte I/O buffers
3241 * A bug in READ has been fixed, so that now a single Ctrl-D
3242 character suffices to cause end-of-file on character streams.
3243 In particular, now you only need one Ctrl-D at the command
3244 line (not two) to exit SBCL.
3245 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3247 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3248 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3249 does the right thing.
3250 * The implementation of some type tests, especially for CONDITION
3251 types, is now tidier and maybe faster, due to CMU CL code
3252 originally by Douglas Crosher, ported by Martin Atzmueller.
3253 * Some math functions have been fixed, and there are new
3254 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3255 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3256 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3257 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3258 likely to be built into sbcl-0.7.0.)
3259 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3260 CL, ported to SBCL by Martin Atzmueller)
3261 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3262 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3263 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3264 * There's a new slam.sh hack to shorten the edit/compile/debug
3265 cycle for low-level changes to SBCL itself, and a new
3266 :SB-AFTER-XC-CORE target feature to control the generation of
3267 the after-xc.core file needed by slam.sh.
3268 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3269 COMPILE-FILE is no longer supported, so that now every function
3270 gets an entry point, so that block compilation looks a little
3271 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3272 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3273 returns the number of bytes consed since the system started,
3274 rather than the number consed since the first time the function
3275 was called. (The new definition parallels ANSI functions like
3276 CL:GET-INTERNAL-RUN-TIME.)
3277 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3278 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3279 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3280 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3281 * Fasl file version numbers are now independent of the target CPU,
3282 since historically most system changes which required version
3283 number changes have affected all CPUs equally. Similarly,
3284 the byte fasl file version is now equal to the ordinary
3287 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3288 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3289 is no longer recognized. I apologize for this, because it was
3290 listed in SB-EXT as a supported extension, but I found that
3291 its existing behavior was poorly specified, as well as incorrectly
3292 specified, and it looked like too much of a mess to straighten it
3293 out. I have enough on my hands trying to get ANSI stuff to work..
3294 * many patches ported from CMU CL by Martin Atzmueller, with
3295 half a dozen bug fixes in pretty-printing and the debugger, and
3296 half a dozen others elsewhere
3297 * fixed bug 13: Floating point infinities are now supported again.
3298 They might still be a little bit flaky, but thanks to bug reports
3299 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3300 as flaky as they were.
3301 * The --noprogrammer command line option is now supported. (Its
3302 behavior is slightly different in detail from what the old man
3303 page claimed it would do, but it's still appropriate under the
3304 same circumstances that the man page talks about.)
3305 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3306 are now supported, and enabled by default. Thus, the compiler can
3307 handle many floating point and complex operations much less
3308 inefficiently. (Thus e.g. you can implement a complex FFT
3310 * The compiler now detects type mismatches between DECLAIM FTYPE
3311 and DEFUN better, and implements CHECK-TYPE more correctly, and
3312 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3313 thanks to patches from Martin Atzmueller.
3314 * various fixes to make the cross-compiler more portable to
3315 ANSI-conforming-but-different cross-compilation hosts (notably
3316 Lispworks for Windows, following bug reports from Arthur Lemmens)
3317 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3318 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3319 fixed thanks to Pierre Mai's CMU CL patches.
3320 * fixes to keep the system from overflowing internal counters when
3321 it tries to use i/o buffers larger than 16M bytes
3322 * fixed bug 45a: Various internal functions required to support
3323 complex special functions have been merged from CMU CL sources.
3324 (When I was first setting up SBCL, I misunderstood a compile-time
3325 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3326 * improved support for type intersection and union, fixing bug 12
3327 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3328 more obscure bugs as well
3329 * some steps toward byte-compiling non-performance-critical
3330 parts of the system, courtesy of patches from Martin Atzmueller
3331 * Christophe Rhodes has made some debian packages of sbcl at
3332 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3333 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3334 stable, but are nonetheless usable. When he's ready, I'd be happy
3335 to add them to the SourceForge "File Releases" section. (And if
3336 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3337 * new fasl file format version number (because of changes in
3338 internal representation of (OR ..) types to accommodate the new
3339 support for (AND ..) types, among other things)
3341 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3342 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3344 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3346 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3347 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3348 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3349 * DESCRIBE now works on structure objects again.
3350 * Most function call argument type mismatches are now handled as
3351 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3352 know whether the function will be redefined before the call is
3353 executed. (The compiler could flag local calls with full WARNINGs,
3354 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3355 it doesn't keep track of enough information to know whether calls
3356 are local in this sense.)
3357 * Compiler output is now more verbose, with messages truncated
3358 later than before. (There should be some supported way for users
3359 to override the default verbosity, but I haven't decided how to
3360 provide it yet, so this behavior is still controlled by the internal
3361 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3362 src/compiler/ir1util.lisp.)
3363 * Fasl file format version numbers have increased again, because
3364 support for the Gray streams extension changes the layout of the
3365 system's STREAM objects.
3366 * The Gray subclassable streams extension now works, thanks to a
3367 patch from Martin Atzmueller.
3368 * The full LOAD-FOREIGN extension (not just the primitive
3369 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3370 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3371 but like most other programs, it defaults to copying the Unix
3372 environment from the original process instead of starting the
3373 new process in an empty environment.
3374 * Extensions which manipulate the Unix environment now support
3375 an :ENVIRONMENT keyword option which doesn't smash case or
3376 do other bad things. The CMU-CL-style :ENV option is retained
3377 for porting convenience.
3378 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3379 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3381 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3383 * A patch from Martin Atzmueller seems to have solved the SIGINT
3384 problem, and as far as we know, signal-handling now works cleanly.
3385 (If you find any new bugs, please report them!)
3386 * The system no longer defaults Lisp source file names to types
3387 ".l", ".cl", or ".lsp", but only to ".lisp".
3388 * The compiler no longer uses special default file extensions for
3389 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3390 seems to expect a single default extension for all compiled code,
3391 and there's no compelling reason to try to stretch the standard
3392 to allow two different extensions.) Instead, byte-compiled files
3393 default to the same extension as native-compiled files.
3394 * Fasl file format version numbers have increased again, because
3395 a rearrangement of internal implementation packages made some
3396 dumped symbols in old fasl files unreadable in new cores.
3397 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3398 many fewer weird special cases.
3399 * Bug #17 (differing COMPILE-FILE behavior between logical and
3400 physical pathnames) has been fixed, and some related misbehavior too,
3401 thanks to a patch from Martin Atzmueller.
3402 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3403 by Tim Moore, ported to SBCL by Martin Atzmueller.
3404 * Martin Atzmueller fixed several filesystem-related problems,
3405 including bug #36, in part by porting CMU CL patches, which were
3406 written in part by Paul Werkowski.
3407 * More compiler warnings in src/runtime/ are gone, thanks to
3408 more patches from Martin Atzmueller.
3409 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3412 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3414 * DESCRIBE now works on CONDITION objects.
3415 * The debugger now handles errors which arise when trying to print
3416 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3418 * The build system now uses an additional file, customize-target-features.lisp,
3419 to allow local modifications to the target *FEATURES* list. (The point of
3420 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3421 debugging features enabled, without having to worry about propagating it
3422 into everyone's system when I do a "cvs update".) When no
3423 customize-target-features.lisp file exists, the target *FEATURES* list
3424 should be constructed the same way as before.
3425 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3426 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3427 warn about behavior which is undefined under the ANSI spec. Note: This
3428 is specified by ANSI, but it's not very popular with programmers.
3429 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3430 macro in the SBCL sources for an example of a workaround which you
3431 might use to make portable ANSI-standard code which does what you want.
3432 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3433 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3434 when it's used as a non-toplevel form. (This is required in order
3435 to implement the DEFCONSTANT-EQX macro.)
3436 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3437 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3438 funny things, due to bugs in the implementation of EVAL-WHEN
3439 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3440 non-toplevel DEFCONSTANTs are uncommon.)
3441 * The core file version number and fasl file version number have been
3442 incremented, because the old noncompliant DEFCONSTANT behavior involved
3443 calling functions which no longer exist, and because I also took the
3444 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3445 * fixed bug 1 (error handling before read-eval-print loop starts), and
3446 redid debugger restarts and related debugger commands somewhat while
3448 ** The QUIT debugger command is gone, since it did something
3449 rather different than the SB-EXT:QUIT command, and since it never
3450 worked properly outside the main toplevel read/eval/print loop.
3451 Invoking the new TOPLEVEL restart provides the same functionality.
3452 ** The GO debugger command is also gone, since you can just invoke
3453 the CONTINUE restart directly instead.
3454 ** The TOP debugger command is also gone, since it's redundant with the
3455 FRAME 0 command, and since it interfered with abbreviations for the
3457 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3459 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3460 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3461 some porting work by Martin Atzmueller)
3462 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3463 with a boatload of patches which he ported to SBCL. Now that those
3465 ** The system tries to make sure that its low-priority messages
3466 are prefixed by semicolons, to help people who like to use
3467 syntax highlighting in their ILISP buffer. (This patch
3468 was originally due to Raymond Toy.)
3469 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3470 patches originally written by Raymond Toy.
3471 ** The compiler understands coercion between single-value and
3472 multiple-VALUES type expressions better, getting rid of some very
3473 weird behavior, thanks to patches originally by Robert MacLachlan
3474 and Douglas Crosher.
3475 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3476 lambda lists, thanks to a patch originally by Pierre Mai.
3477 ** The system no longer bogusly warns about "abbreviated type
3479 ** The compiler gets less confused by inlining and RETURN-FROM,
3480 thanks to some patches originally by Tim Moore.
3481 ** The system no longer hangs when dumping circular lists to fasl
3482 files, thanks to a patch originally from Douglas Crosher.
3483 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3484 undefined function error.
3485 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3486 in structure slot initforms) for now, documented workaround instead:-|
3487 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3488 * fixed bug 5: added stubs for various Gray stream functions called
3489 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3490 installed, at least appropriate type errors are generated
3491 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3492 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3493 consistently in DEFMETHOD forms.
3494 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3495 it doesn't seem to affect SBCL after all
3496 * The C runtime system now builds with better optimization and many
3497 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3499 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3501 * The system is now under CVS at SourceForge (instead of the
3502 CVS repository on my home machine).
3503 * The new signal handling code has been tweaked to treat register
3504 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3505 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3506 downstream, e.g. in debug-int.lisp, has implicit dependencies
3507 on the unsignedness of integer representation of machine words,
3508 and that caused the system to bomb out with infinite regress
3509 when trying to recover from type errors involving signed values,
3510 e.g. (BUTLAST '(1 2 3) -1).)
3511 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3512 (This was one of the bugs Peter Van Eynde reported back in July.)
3513 * The system now uses code inspired by Colin Walters' O(N)
3514 implementation of MAP (from the cmucl-imp@cons.org mailing
3515 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3516 inline the MAP operation, and there is more than one
3517 sequence argument to the MAP call (so that it can't just
3518 do ETYPECASE once and for all based on the type of the
3519 single sequence argument). (The old non-inline implementation
3520 of the general M-argument sequence-of-length-N case required
3521 O(M*N*N) time when any of the sequence arguments were LISTs.)
3522 * The QUIT :UNIX-CODE keyword argument has been renamed to
3523 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3524 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3526 * Raymond Wiker's patches to port dynamic loading from Linux to
3527 FreeBSD have been added.
3528 * The BUGS file is now more nearly up to date, thanks in large part
3529 to Martin Atzmueller's review of it.
3530 * The debugger now flushes standard output streams before it begins
3531 its output ("debugger invoked" and so forth).
3532 * The core version number and fasl file version number have both
3533 been incremented, because of incompatible changes in the layout
3535 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3536 used to not be called for a saved Lisp image.)
3537 * Martin Atzmueller's version of a patch to fix a compiler crash,
3538 as posted on sbcl-devel 13 September 2000, has been installed.
3539 * Instead of installing Martin Atzmueller's patch for the
3540 compiler transform for SUBSEQ, I deleted the compiler transform,
3541 and transforms for some similar consing operations.
3542 * A bug in signal handling which kept TRACE from working on OpenBSD
3544 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3545 away by constant folding
3546 * The system now defines its address space constants in one place
3547 (in the Lisp sources), and propagates them automatically elsewhere
3548 (through GENESIS and the sbcl.h file). Therefore, patching the
3549 address map is less unnecessarily tedious and error-prone. The
3550 Lisp names of address space constants have also been systematized.
3551 * CVS tags like dollar-Header-dollar have been removed from
3552 the sources, because they have never saved me trouble and
3553 they've been source of trouble working with patches and other
3554 diff-related operations.
3555 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3556 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3558 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3560 * The system has been ported to OpenBSD.
3561 * The system now compiles with a simple "sh make.sh" on the systems
3562 that it's supported on. I.e., now you no longer need to tweak
3563 text in the target-features.lisp-expr and symlinks in src/runtime/
3564 by hand, the make.sh takes care of it for you.
3565 * The system is no longer so grossly inefficient when compiling code
3566 involving vectors implemented as general (not simple) vectors (VECTOR T),
3567 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3568 which dares to use the various sequence functions on non-simple
3569 vectors, takes less of a performance hit.
3570 * There is now a primitive type predicate VECTOR-T-P
3571 to test for the (VECTOR T) type, so that e.g.
3572 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3573 can now be compiled with some semblance of efficiency. (The old code
3574 turned the type declaration into a full call to %TYPEP at runtime!)
3575 * AREF on (VECTOR T) is still not fast, since it's still compiled
3576 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3577 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3578 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3579 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3580 without making full calls to SUBTYPEP at runtime.
3581 (Some analogous efficiency issues for non-simple vectors specialized to
3582 element types other than T, or for non-simple multidimensional arrays,
3583 have not been addressed. They could almost certainly be handled the
3584 same way if anyone is motivated to do so.)
3585 * The changes in array handling break binary compatibility, so
3586 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3587 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3588 instead of (VALUES T T).
3589 * By following the instructions that Dan Barlow posted to sbcl-devel
3590 on 2 July 2000, I was able to enable primitive dynamic object
3591 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3592 functionality is not implemented (since it calls ld to resolve
3593 library references automatically, requiring RUN-PROGRAM for its
3594 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3595 try to resolve library references) is now supported.
3596 * The system now flushes the standard output streams when it terminates,
3597 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3598 them at several other probably-convenient times, e.g. in each pass of
3599 the toplevel read-eval-print loop, and after evaluating a form given
3600 as an "--eval" command-line option. (These changes were motivated by a
3601 discussion of stream flushing issues on cmucl-imp in August 2000.)
3602 * The source transform for TYPEP of array types no longer assumes
3603 that an array whose element type is a not-yet-defined type
3604 is implemented as an array of T, but instead punts, so that the
3605 type will be interpreted at runtime.
3606 * There is now some support for cross-compiling in make.sh: each of
3607 the phases of make.sh has its own script. (This should be transparent
3608 to people doing ordinary, non-cross-compile builds.)
3609 * Since my laptop doesn't have hundreds of megabytes of memory like
3610 my desktop machine, I became more motivated to do some items on
3611 my to-do list in order to reduce the size of the system a little:
3612 ** Arrange for various needed-only-at-cold-init things to be
3613 uninterned after cold init. To support this, those things have
3614 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3615 symbols with such names are now uninterned after cold init).
3616 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3617 fasl files for cold load.
3618 ** Remove the old compiler structure pooling code (which used to
3619 be conditional on the target feature :SB-ALLOC) completely.
3620 ** Redo the representation of some data in cold init to be more compact.
3621 (I also looked into supporting byte compiled code at bootstrap time,
3622 which would probably reduce the size of the system a lot, but that
3623 looked too complicated, so I punted for now.)
3624 * The maximum signal nesting depth in the src/runtime/ support code has
3625 been reduced from 4096 to 256. (I don't know any reason for the very
3626 large old value. If the new smaller value turns out to break something,
3627 I'll probably just bump it back up.)
3628 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3630 * Many, many bugs reported by Peter Van Eynde have been added to
3631 the BUGS list; some have even been fixed.
3632 * While enabling dynamic object file loading, I tried to make the
3633 code easier to understand, renaming various functions and variables
3634 with less ambiguous names, and changing some function calling
3635 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3636 * While trying to figure out how to do the OpenBSD port, I tried to
3637 clean up some of the code in src/runtime/. In particular, I dropped
3638 support for non-POSIX signal handling, added various comments,
3639 tweaked the code to reduce the number of compilation warnings, and
3640 renamed some files to increase consistency.
3641 * To support the new automatic configuration functionality in make.sh,
3642 the source file target-features.lisp-expr has been replaced with the
3643 source file base-target-features.lisp-expr and the machine-generated
3644 file local-target-features.lisp-expr.
3645 * fixed a stupid quoting error in make.sh so that using CMU CL
3646 "lisp -batch" as cross-compilation host works again
3648 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3650 * DESCRIBE no longer tries to call itself recursively to describe
3651 bound/fbound values, so that it no longer fails on symbols which are
3652 bound to themselves (like keywords, T, and NIL).
3653 * DESCRIBE now works on generic functions.
3654 * The printer now prints less-screwed-up representations of closures
3655 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3656 * A private symbol is used instead of the :EMPTY keyword previously
3657 used to mark empty slots in hash tables. Thus
3658 (DEFVAR *HT* (MAKE-HASH-TABLE))
3659 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3660 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3661 now does what ANSI says that it should. (You can still get
3662 similar noncompliant behavior if bang on the hash table
3663 implementation with all the symbols you get back from
3664 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3665 This breaks binary compatibility, since tests for equality to
3666 :EMPTY are wired into things like the macroexpansion of
3667 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3669 * There's now a minimal placeholder implementation for CL:STEP,
3670 as required by ANSI.
3671 * An obscure bug in the interaction of the normal compiler, the byte
3672 compiler, inlining, and structure predicates has been patched
3673 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3674 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3676 * Missing ordinary arguments in a macro call are now detected even
3677 when the macro lambda list contains &KEY or &REST.
3678 * The debugger no longer complains about encountering the top of the
3679 stack when you type "FRAME 0" to explicitly instruct it to go to
3680 the top of the stack. And it now prints the frame you request even
3681 if it's the current frame (instead of saying "You are here.").
3682 * As specified by ANSI, the system now always prints keywords
3683 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3685 * The default initial SIZE of HASH-TABLEs is now smaller.
3686 * Type information from CLOS class dispatch is now propagated
3687 into DEFMETHOD bodies, so that e.g.
3688 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3690 is now basically equivalent to
3691 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3692 (DECLARE (TYPE SINGLE-FLOAT X))
3694 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3695 operation, without having to do run-time type dispatch.
3696 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3697 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3698 * Since I don't seem to be making big file reorganizations very often
3699 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3700 is overflowing my ability to conveniently back them up), I've finally
3701 checked the system into CVS. (The CVS repository is on my home system,
3702 not at SourceForge -- putting it on SourceForge might come later.)
3703 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3704 high-level GC-NOTIFY-FOO functions send their output. (There's
3705 still very little control of where low-level verbose GC functions
3706 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3707 less than it used to -- the GC-NOTIFY-FOO functions are now under
3708 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3709 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3710 in only one place in the source code, and propagates it automatically
3711 everywhere that it's needed. Thus e.g. when I bump the version from
3712 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3713 * The C source files now include boilerplate legalese and documentation
3714 at the head of each file (just as the Lisp source files already did).
3715 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3716 to his page will be replaced with a link to his new CLiki service.
3718 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3720 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3721 * The build process now looks for GNU make under the default name "gmake",
3722 instead of "make" as it used to. If GNU make is not available as "gmake"
3723 on your system, you can change this default behavior by setting the
3724 GNUMAKE environment variable.
3725 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3726 can build without error under CMU CL.
3728 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3730 * There is now a partial SBCL user manual (with some new text and some
3731 text cribbed from the CMU CL manual).
3732 * The beginnings of a profiler have been added (starting with the
3733 CMU CL profiler and simplifying and cleaning up). Eventually the
3734 main interface should be through the TRACE macro, but for now,
3735 it's still accessed through vaguely CMU-CL-style functions and macros
3736 exported from the package SB-PROFILE.
3737 * Some problems left over from porting CMU CL to the new
3738 cross-compilation bootstrap process have been cleaned up:
3739 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3740 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3741 part when getting rid of the compiler *BACKEND* stuff.)
3742 ** The value of *NULL-TYPE* was screwed up, because it was
3743 being initialized before the type system knew the final
3744 definition of the 'NULL type. This screwed up several key
3745 optimizations in the compiler, causing inefficiency in all sorts
3746 of places. (I found it because I wanted to understand why
3747 GET-INTERNAL-RUN-TIME was consing.)
3748 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3749 PROCLAIM FTYPE information. Unfortunately this broke binary
3750 compatibility again, since now the forms output by DEFGENERIC
3751 to refer to functions which didn't exist in 0.6.3.
3752 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3753 can use the new (as of 0.6.3) transform for SOME into MAP into
3755 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3756 values to (INTEGER 0 1000000), so that the time code will no longer
3757 occasionally get blown up by Linux returning 1000000 microseconds
3758 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3759 its output conform to the ANSI spec. (Alas, this makes its output
3760 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3761 * A full call to MAP NIL with a single sequence argument no longer conses.
3762 * fixes to problems pointed out by Martin Atzmueller:
3763 * The manual page no longer talks about multiprocessing as though
3764 it were currently supported.
3765 * The ILISP support patches have been removed from the distribution,
3766 because as of version 5.10.1, ILISP now supports SBCL without us
3767 having to maintain patches.
3768 * added a modified version of Raymond Toy's recent CMU CL patch for
3769 EQUALP comparison of HASH-TABLE
3771 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3773 * The system still can't cross-compile itself with
3774 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3775 entails), but at least it can compile more of itself that way
3776 than it used to be able to, and various buglets which were uncovered
3777 by trying to cross-compile itself that way have now been fixed.
3778 * This release breaks binary compatibility again. This time
3779 at least I've incremented the FASL file format version to 2, so that the
3780 problem can be detected reliably instead of just causing weird errors.
3781 * various new style warnings:
3782 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3783 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3784 ** using the lexical binding of a variable named in the *FOO* style
3785 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3786 as specified by ANSI.
3787 * *RANDOM-STATE* is no longer automatically initialized from
3788 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3789 default behavior of the system is to repeat its behavior every time
3790 it's run. If you'd like to change this behavior, you can always
3791 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3792 old convention there was no comparably easy way to get the system to
3793 repeat its behavior every time it was run.
3794 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3795 FTYPE declarations has been removed, in favor of their ANSI
3796 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3797 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3798 the types of their sequence arguments can be determined at compile time.
3799 This is done through a new open code expansion for MAP which eliminates
3800 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3801 argument types can be determined at compile time.
3802 * The optimizer now transforms COERCE into an identity operation when it
3803 can prove that the coerced object is already of the correct type. (This
3804 can be a win for machine generated code, including the output of other
3805 optimization transforms, such as the MAP transform above.)
3806 * Credit information has been moved from source file headers into CREDITS.
3807 * Source file headers have been made more standard.
3808 * The CASE macro now compiles without complaining even when it has
3811 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3813 * (Note that the way that the PCL macroexpansions were rewritten
3814 to accommodate the change in DEFGENERIC below breaks binary
3815 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3816 not run under sbcl-0.6.2. Once we get out of alpha releases,
3817 i.e. hit release 1.0.0, we'll probably try to maintain binary
3818 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3819 and sbcl-1.4.4. Until then, however, it might be fairly common
3820 for maintenance releases to break binary compatibility.)
3821 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3823 * The reporting of unhandled signals has been changed to print some
3824 explanatory text as well as the report form. (Previously only
3825 the report form was printed.)
3826 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3827 it defines, so that the compiler no longer issues undefined function
3828 warnings for compiled-but-not-yet-loaded generic functions.
3829 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3830 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3831 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3832 section of the ANSI spec are supported.
3833 * The "" nickname for the "KEYWORD" package has been removed.
3834 The reader still handles symbol tokens which begin with a package marker
3835 as keywords, but it doesn't expose its mechanism for doing so in the
3836 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3837 * The system now issues STYLE-WARNINGs for contradictory TYPE
3838 proclamations. (Warnings for contradictory FTYPE proclamations would
3839 be nice too, but those can't be done usefully unless the type system
3840 is made smarter about FUNCTION types.)
3841 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3842 systematized, so that "*target-*.lisp is supposed to exist only on the
3843 target and imply that there's a related file which exists on the
3844 host, and *host-*.lisp is supposed to exist only on the host and imply
3845 that there's a related file which exists on the target. This involves a
3846 lot of renaming. Hopefully the acute confusion caused by the renaming
3847 will be justified by the reduction in chronic confusion..
3848 ** runtime-type.lisp -> early-target-type.lisp
3849 ** target-type.lisp -> late-target-type.lisp
3850 ** early-host-format.lisp -> early-format.lisp
3851 ** late-host-format.lisp -> late-format.lisp
3852 ** host-error.lisp -> misc-error.lisp
3853 ** early-error.lisp -> early-target-error.lisp
3854 ** late-error.lisp -> late-target-error.lisp
3855 ** host-defboot.lisp -> early-defboot.lisp
3856 ** code/misc.lisp -> code/target-misc.lisp
3857 ** code/host-misc.lisp -> code/misc.lisp
3858 ** code/numbers.lisp -> code/target-numbers.lisp
3859 ** code/early-numbers.lisp -> numbers.lisp
3860 ** early-host-type.lisp -> early-type.lisp
3861 ** late-host-type.lisp -> late-type.lisp
3862 ** host-typep.lisp -> typep.lisp
3863 ** load.lisp -> target-load.lisp
3864 ** host-load.lisp -> load.lisp
3865 ** host-disassem.lisp -> disassem.lisp
3866 ** host-insts.lisp -> insts.lisp
3867 ** byte-comp.lisp -> target-byte-comp.lisp
3868 ** host-byte-comp.lisp -> byte-comp.lisp
3869 ** host-signal.lisp -> signal.lisp
3870 ** host-defstruct.lisp -> defstruct.lisp
3871 ** late-target-type.lisp -> deftypes-for-target.lisp
3872 Furthermore, several other previously target-only files foo.lisp (e.g.
3873 hash-table.lisp and random.lisp) have been split into a target-and-host
3874 foo.lisp file and a target-only target-foo.lisp file, with their key type
3875 definitions in the target-and-host part, so that the cross-compiler will
3876 know more about target types.
3877 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3878 gone away. In their place are various *BACKEND-FOO* variables
3879 corresponding to the slots of the old structure.
3880 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3881 into the target SBCL has been fixed.
3882 * The chill.lisp system for loading cold code into a running SBCL
3884 * Support for the CMU CL "scavenger hook" extension has been removed.
3885 (It was undocumented and unused in the CMU CL sources that SBCL was
3886 derived from, and stale in sbcl-0.6.1.)
3887 * Various errors in the cross-compiler type system were detected
3888 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3889 (enabling various consistency checks). Many of them were fixed,
3890 but some hard problems remain, so the compiler is back to
3891 running without *TYPE-SYSTEM-INITIALIZED* for now.
3892 * As part of the cross-compiler type system cleanup, I implemented
3893 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3894 * I have started adding UNCROSS calls throughout the type system
3895 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3896 on cross-compiler input files will be able to go away, and various
3898 * CONSTANTP now returns true for quoted forms (as explicitly required
3901 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3903 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3904 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3905 (MAKE-LIST -1). In the longer term, it ought to become true
3906 that declarations are assertions even at SAFETY 1. For now, it's not
3907 quite true even at SAFETY 3, but it's at least more nearly true..
3908 (Note that this change seems to increases the size of the system by
3909 O(5%) and to decrease the speed of the compiler by 20% or more.)
3910 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3911 for the problem of printing dozens of lines of distracting information
3912 about low-level system machinery as part of the top stack frame
3913 on entry to the debugger when an undefined function was called.
3914 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3915 is not reset to COMMON-LISP-USER.
3916 * Compilation of stuff related to dyncount.lisp has been made conditional
3917 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3918 smaller. The various dyncount-related symbols have been moved into
3919 a new "SB-DYNCOUNT" package.
3920 * tty-inspect.lisp has been renamed to inspect.lisp.
3921 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3922 feature has gone away. (When we eventually port to other flavors of
3923 libc and/or Unix, we'll try to make the differences between flavors
3924 invisible at the user level.)
3925 * Various other *FEATURES* tags, and/or their associated conditionals,
3926 have been removed if obsolescent, or given better documentation, or
3927 sometimes given more-mnemonic names.
3929 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3931 * tidied up "make.sh" script
3932 * tidied up system directory structure
3933 * better "clean.sh" behavior
3934 * added doc/FOR-CMUCL-DEVELOPERS
3935 * many many small tweaks to output format, e.g. removing possibly-confusing
3936 trailing #\. character in DESCRIBE-INSTANCE
3937 * (EQUALP #\A 'A) no longer signals an error.
3938 * new hashing code, including EQUALP hashing
3939 * tidied up Lisp initialization and toplevel
3940 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3941 * command line argument processing
3942 * added POSIX-GETENV function to deal with Unix-ish environment variables
3943 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3944 terminating SBCL on EOF
3945 * non-verbose GC by default
3946 * There is no more "sbcl" shell script; the sbcl file is now the C
3947 runtime executable (just like CMU CL).
3948 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3950 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3951 symbol and package information as Lisp native symbol and package objects
3952 instead of strings naming symbols and strings naming packages. This way,
3953 whenever packages are renamed (as in warm init), debug information is
3954 transformed along with everything else.
3955 * tweaked the optimization policy declarations which control the building
3956 of SBCL itself. Now, among other things, the system no longer saves
3957 source location debugging information. (This helps two problems at once
3958 by reducing SBCL size and by keeping SBCL from trying to look for its
3959 sources -- which may not exist -- when reporting errors.)
3960 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3961 debugging and testing purposes
3962 * cleaned up printing, making the printer call PRINT-OBJECT for
3963 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3964 methods, giving nearly-ANSI behavior
3965 * converted almost all special variables to use *FOO* naming convention
3966 * deleted PARSE-TIME functionality, since it can be done portably
3967 * moved some files out of cold init into warm init
3968 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3970 * regularized formatting of source files
3971 * added an install.sh script
3972 * fixed ridiculous memory usage of cross-compiler by making
3973 compiler/alloc.lisp not try to do pooling unless it can hook
3974 itself into the GC of the cross-compilation host. Now the system
3975 builds nicely on my old laptop.
3976 * added :SB-ALLOC in target-features.lisp-expr
3977 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3978 was not implemented)
3979 * re-did condition handling and note reporting in the compiler. Notes
3980 are no longer handled by signalling conditions. Style warnings
3981 and warnings are handled more correctly and reported in such a way
3982 that it's easy to find one or the other in your output (so that you
3983 can e.g. figure out which of many problems caused COMPILE-FILE to
3985 * changed the severity of several compiler warnings from full WARNING
3986 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3987 compiler note reporting so that it doesn't use the condition system
3988 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3989 COMPILE-FILE command)
3990 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3991 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3992 no longer supported, since their implementation was deeply intertwingled
3993 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3994 * removed bogus "support" for compiler macros named (SETF FOO), and
3995 removed the compiler macro for SETF INFO (but only after making a fool
3996 of myself on the cmucl-imp mailing list by posting a bogus patch for
3997 DEFINE-COMPILER-MACRO..)
3998 * Compiled files containing forms which have side effects on the Lisp
3999 reader (such as DEFPACKAGE forms) are now handled more correctly.
4000 (Compiler queuing of top level lambdas has been suppressed by setting
4001 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4002 * deleted various currently-unused source files, e.g. gengc.lisp. They
4003 may be added back at some point e.g. when porting to other architectures,
4004 but until they are it's distracting to distribute them and to try to
4006 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4007 there were so many of them they're just distractions, and UNCROSS is
4008 known to be able to handle the current sources
4009 * moved PROFILE functionality into TRACE, so that it will be clear
4010 how the wrapping and unwrapping of functions when you profile them
4011 interacts with the wrapping and unwrapping of functions when you
4012 trace them. (Actually, the functionality isn't there yet, but at least
4013 the interface specification is there. Hopefully, the functionality will
4014 arrive with some future maintenance release.)
4015 * removed host-oops.lisp
4016 * changed signature of QUIT function to allow UNIX-CODE argument
4017 * fixed READ-SEQUENCE bug
4018 * tweaked verbose GC output so that it looks more like the progress
4019 output that ANSI specifies for functions like LOAD
4020 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4021 * added <http://sbcl.sourceforge.com> to the banner information printed by