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.
9 * improvements to the Windows port:
10 ** floating point exceptions are now reported correctly.
11 ** stack exhaustion detection works partially.
12 ** more accurate GET-INTERNAL-REAL-TIME.
13 ** better RUN-PROGRAM behaviour in case of errors in the child
15 ** PROBE-FILE now simplifies pathnames correctly.
16 ** DIRECTORY now works correctly with :WILD-INFERIORS.
18 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
19 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
20 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
22 * bug fix: two potential GC deadlocks affecting threaded builds.
23 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
24 returns T (reported by Anton Kazennikov)
25 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
26 with non-variable places
27 * bug fix: remove a race condition in the setting of
28 funcallable-instance functions, this should make threaded CLOS
29 code more stable against memory faults.
30 * bug fix: corruption of specials when unbinding is interrupted by an
31 asynchronous unwind (reported by Hannu Koivisto)
32 * improvement: the debugger will now also display local variables that
33 are only used once, for code compiled with a DEBUG optimization quality
36 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
37 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
38 * incompatible change: External-format support for FFI calls. The
39 SB-ALIEN:C-STRING no longer implies an ASCII
40 external-format. Instead the string is subject to external-format
41 conversion. Additionally return values of type SB-ALIEN:C-STRING
42 are by default of type (SIMPLE-ARRAY CHARACTER), not
43 SIMPLE-BASE-STRING. For an alien type that behaves like the old
44 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
45 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
46 * incompatible change: SB-EXT package no longer contains the
47 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
48 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
50 * incompatible change: the single-stepper is no longer available
51 on Alpha, Mips and Sparc platforms.
52 * minor incompatible change: the direct superclasses of
53 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
54 not (STANDARD-OBJECT FUNCTION). This makes the
55 class-precedence-lists of GENERIC-FUNCTION and
56 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
58 * minor incompatible change: the default stream external format on
59 non-unicode SBCL is no longer detected from the locale, but is always
61 * new feature: Add a version of evaluator that uses an interpreter instead
62 of the compiler. EVAL still uses the compiler by default, to switch it
63 to use the interpreter, set the value of the variable
64 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
65 * minor incompatible change: the single-stepper REPL has been merged
66 with the normal debugger (see the "Stepping" heading of the debugger help
67 for more details). The debugger command STEP will no longer switch
68 to the single-stepper REPL.
69 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
71 * bug fix: loading of default sysinit file works. (thanks to Leonid
73 * bug fix: better detection of circularities in the file-compiler.
74 (reported by Marco Monteiro)
75 * bug fix: the CL pathname functions now work with files that have
76 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
77 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
78 systems with Unix98 pty semantics.
79 * bug fix: ASDF-INSTALL will now work with bsd tar.
80 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
82 * bug fix: timers expiring in dead threads no longer cause a
83 type-error (reported by Paul "Nonny Mouse").
84 * bug fix: thanks to more lightweight single-stepper instrumentation,
85 code compiled with (DEBUG 3) will compile and execute significantly faster,
86 and will have more accurate type-inferencing than before
87 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
88 whose bindings are modified
89 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
90 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
91 ** replace dummy implementations of CL:MACHINE-INSTANCE and
92 CL:SOFTWARE-VERSION with proper version
94 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
95 * feature: implemented the READER-METHOD-CLASS and
96 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
98 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
100 * optimization: faster LOGCOUNT implementation on x86 and x86-64
101 (thanks to Lutz Euler)
102 * optimization: hashing of general arrays and vectors has been
103 improved. (reported by Any Fingerhut)
104 * enhancement: SB-INTROSPECT is now able to find definitions of
105 profiled functions. (thanks to Troels Henriksen)
106 * enhancement: compiler-macro expansion applies now to FUNCALL forms
107 as well, allowing compiler-macros for SETF-functions to expand.
108 * enhancement: step-instrumentation no longer wraps known
109 single-value functions with multiple-value context, allowing
110 better type inference.
111 * fixed bug #337: use of MAKE-METHOD in method combination now works
112 even in the presence of user-defined method classes. (reported by
113 Bruno Haible and Pascal Costanza)
114 * fixed bug #339(c): if there are applicable methods not part of any
115 long-form method-combination group, call INVALID-METHOD-ERROR.
116 (reported by Bruno Haible)
117 * fixed bug #361: the :FUNCTION initarg in the protocol for
118 initialization of methods can now be used to override
119 internally-produced optimized functions. (reported by Bruno
121 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
122 system-provided lambda expression no longer cause warnings about
123 unbound #:|pv-table| symbols.
124 * bug fix: improved the handling of type declarations and the
125 detection of violations for keyword arguments with non-constant
127 * bug fix: potentially erronous calls to PATHNAME and
128 MERGE-PATHNAMES were being flushed in some cases.
129 (reported by Richard Kreuter)
130 * bug fix: compiled calls to TYPEP were mishandling obsolete
131 instances. (reported by James Bielman and Attila Lendvai)
132 * bug fix: format strings with ~> without matching ~< no longer
133 trigger an AVER, but signal an understandable error instead.
134 (reported by Antonio Martinez)
135 * bug fix: specifying an output-file without a pathname-type for
136 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
137 of input-file instead of "fasl". (reported by Robert Dodier)
138 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
139 (reported by James Y Knight).
140 * bug fix: compiler-macros-function did not consider the environment
141 argument for shadowing by local functions.
142 * bug fix: compiler-macros expansion was inhibited by local INLINE
144 * bug fix: inline expansions of known functions were subject to
145 step-instrumentation in high debug policies, leading to problems
147 * bug fix: compiler failed to differentiate between different CONS
149 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
150 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
151 * bug fix: #368: incorrect use of expressed vs. upgraded array
153 * bug fix: #306a: more precise unions of array types.
154 * thread-safety improvements:
155 ** CONDITION-WAIT could return early on Linux, if the thread was
156 interrupted and subsequently continued with SIGCONT.
157 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
159 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
160 * added support for the ucs-2 external format. (contributed by Ivan
162 * minor incompatible change: pretty printing of objects of type
163 (cons symbol) is, in the default pprint-dispatch-table, now
164 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
166 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
167 called later in a class's lifetime, possibly as late as when the
168 first instance of the class is created. Previously,
169 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
170 class became finalizeable.
171 * fixed bug: FILE-POSITION sometimes returned inconsistent results
172 for multibyte external-format streams. (thanks to "vbzoli")
173 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
174 with :ALLOCATION :CLASS inherited from superclasses of the
176 * fixed bug: anonymous classes can now be created using the :NAME
177 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
178 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
179 * fixed bug: core-files saved with :EXECUTABLE T can again be
180 executed when SBCL_HOME isn't set. (reported by James Knight)
181 * fixed bug: toplevel LOCALLY forms with declarations could
182 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
183 * fixed bug: printing from several different threads using different
184 values of *print-case* could cause invalid output, due to
185 some internal special variables of the printer not being bound
186 thread-locally (reported by Max Mikhanosha)
187 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
188 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
189 (reported by Pascal Costanza)
190 * fixed bug: FINALIZE-INHERITANCE is called from
191 REINITIALIZE-INSTANCE on classes when the class has previously
192 been finalized, as required by AMOP.
193 * minor code generation optimizations:
194 ** better register allocation in CLOS dispatching functions
195 ** overflow detection when coercing signed bytes to fixnums on x86-64
196 ** is now implemented with one IMUL instruction instead of three shifts
197 ** more efficient bit-vector access on x86 and x86-64
198 ** more efficient access to raw structure slots on x86 and x86-64
199 * fixed some bugs revealed by Paul Dietz' test suite:
200 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
203 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
204 * feature: thread support on Solaris/x86, and experimental thread support
206 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
208 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
209 default method for SLOT-UNBOUND.
210 * minor incompatible change: prevent the user from specializing the
211 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
212 somewhat counter to the intent of the protocol, I (CSR) think, and
213 additionally it just doesn't work in SBCL as currently
214 implemented, thanks to optimizations (that are always valid for
215 the other three SLOT-VALUEish functions, but not for the setter).
216 * minor incompatible change: the :SB-LDB feature is now enabled by
217 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
218 the low-level debugger.
219 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
220 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
221 down to an SB-SYS:FD-STREAM, allowing direct communication in
222 more cases. Temporary files and pipes are still used as a fallback
224 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
225 on the non-reentrant C functions localtime(3) and gmtime(3).
226 * bug fix: global optimization policy was not visible in LOCALLY and
228 * bug fix: class objects can be used as specializers in methods.
229 (reported by Pascal Costanza)
230 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
231 with a type error. (reported by blitz_ on #lisp)
232 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
233 failed with a type error. (reported by Pascal Bourguignon)
234 * bug fix: merging pathnames against defaults with :DIRECTORY
235 starting with '(:RELATIVE :BACK) should preserve the :BACK.
236 (reported by James Y Knight)
237 * bug fix: saving large (>2GB) cores on x86-64 now works
238 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
239 constant index and a (simple-array (signed-byte 32)) array
240 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
241 error (patch by Robert J. Macomber)
242 * feature: TIME now displays the amount of run-time spent in GC
243 * bug fix: The debugger now does a better job of respecting
244 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
245 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
246 workaround for bug 403.)
247 * bug fix: floating point exception handling now works on FreeBSD
248 (thanks to NIIMI Satoshi)
249 * fixed some bugs revealed by Paul Dietz' test suite:
250 ** MISC.641: LET-conversion were not supposed to work in late
252 * improvements to the Win32/x86 port:
253 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
254 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
256 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
257 * new feature: source path information is generated for macro-expansion
258 errors for use in IDE's like Slime (thanks to Helmut Eller)
259 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
260 * bug fix: compiler does not loop forever on an invalid type in
262 * improvement: compilation of most CLOS applications is significantly
264 * optimization: added a limited bytecode compiler for simple toplevel
265 forms, speeding up compilation and FASL loading
266 * bug fix: the statistical profiler now properly distinguishes anonymous
269 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
270 * minor incompatible change: in sbcl-0.9.11 (but not earlier
271 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
272 into subclasses of the finalized class. Now user calls to
273 FINALIZE-INHERITANCE finalize just the one class, and calls by the
274 system return before any subclasses are finalized.
275 * minor incompatible change: The reader no longer ignores errors
276 regarding non-existent packages in #+ and #- feature tests.
277 * new feature: command line options --no-sysinit, --no-userinit to
278 inhibit loading the corresponding init files
279 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
280 for sending data through UDP sockets (thanks to François-René Rideau)
281 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
282 error is signalled instead (thanks to François-René Rideau)
283 * minor incompatible change: Error signalling behaviour of lexical
284 operations violating package locks has changed slightly. Refer to
285 documentation on package locks for details.
286 * enhancement: EVAL can process IF-expressions without resorting to the
288 * bug fix: Some operations involving SIN, COS, or TAN weren't being
289 constant-folded properly due to missing out-of-line functions.
290 (reported by Mika Pihlajamäki)
291 * bug fix: LISTEN sometimes returned T even in cases where no data was
292 immediately available from the stream
293 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
294 were reversed. (reported by Levente Mészáros)
295 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
296 when given a vector displaced to an adjustable array. (reported
298 * bug fix: derivation of float boundaries from numbers outside the
299 appropriate float range (reported by John Wiseman)
300 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
302 * bug fix: precomputing cache entries for generic functions with
303 some subclasses of specializers as yet invalid does not attempt to
304 fill a cache line with a negative offset. (reported by Levente
306 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
307 allow condition class objects as arguments to DOCUMENTATION and
308 (SETF DOCUMENTATION); only find and set documentation for
309 structure names for the STRUCTURE doc-type. (suggested by Gary
311 * improvements to the Win32/x86 port:
312 ** (user-homedir-pathname) and default initialization file
313 locations now know about the user's "Documents and Settings"
314 directory (thanks to Yaroslav Kavenchuk)
315 ** run-program is implemented (thanks to Mike Thomas)
316 ** sockets support (thanks to Timothy Ritchey)
317 ** better backtrace support (thanks to Alastair Bridgewater)
318 ** sb-grovel supported
319 ** asdf-install and sb-posix work somewhat
320 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
321 * improvements to the Solaris/x86 port:
322 ** works on Solaris 11/Solaris Express
323 ** floating-point exception handling support
324 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
325 * fixed some bugs revealed by Paul Dietz' test suite:
326 ** REMOVE-METHOD returns its generic function argument even when
327 the method is not one of the generic functions' methods.
328 ** objects declared as MEMBER types can be admissible arguments to
330 ** printing characters should simply be printed by the FORMAT ~:C
332 ** compiler failure when compiling functions with hairy constant
333 defaults for optional parameters.
334 ** compiler produces wrong code when MAYBE-INLINE-expanding a
335 function, which is already optimized.
337 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
338 * new platform: experimental support for SBCL x86/Darwin, including
339 MacOS X 10.4.5 on Intel.
340 * new feature: Unicode character names are now known to the system
341 (through CHAR-NAME and NAME-CHAR).
342 * new feature: the filesystem location of SBCL's core file is
343 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
344 * minor incompatible change: the contrib modules SB-POSIX and
345 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
346 this change is to make it easier to distribute
347 location-independent binaries.
348 * bug fix: as implied by AMOP, standardized classes no longer have
349 slots named by external symbols of public packages. (reported by
351 * optimization: calling functions via constant symbols -- as in
352 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
353 via the function object as in (FUNCALL #'FOO).
354 * optimization: on x86 and x86-64, the calling convention for the
355 case of dealing with an unknown number of values has been altered
356 to be friendlier to the prediction heuristics implemented,
357 particularly with reference to CALL and RET pairing. (thanks to
358 Alastair Bridgewater)
359 * optimization: on x86, the code for access to thread-local symbol
360 values and binding/unbinding of thread-local symbols is smaller.
361 (thanks to Alastair Bridgewater)
362 * enhancement: CONSTANTP is now able to determine constantness of
363 more complex forms, including calls to constant-foldable standardized
364 functions and some special forms beyond QUOTE.
365 * fixed bug: occasional GC crashes on Solaris/x86
366 * optimization: x86-64 supports stack allocation of results of simple
367 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
368 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
369 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
370 (thanks to James Knight)
371 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
372 by the standard (thanks to Stephen Compall)
374 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
375 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
376 be used for bundling the runtime and the core file into one
377 executable binary. This feature is not currently supported on all SBCL
378 platforms. (thanks to James Bielman and NIIMI Satoshi)
379 * new feature: a generational or ephemeral garbage collector is now
380 the default on the PowerPC platform (both Linux and Darwin). The
381 old Cheney (stop and copy) collector is a build-time option.
382 (thanks to Cyrus Harmon, after Raymond Toy)
383 * minor incompatible change: the method by which SBCL finds its
384 contributed modules has changed; it no longer relies on symbolic
385 links from an $SBCL_HOME/systems directory, but searches directly
386 in the subdirectories of $SBCL_HOME.
387 * enhancement: the dynamic heap size on the Linux/PPC platform is
388 markedly larger, even using the older Cheney garbage collector.
389 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
390 certain complicated string types. (reported by Gary King)
391 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
392 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
393 * fixed bug: DOCUMENTATION on structures no longer signals an
394 error if no structure type of the right name exists. (reported by
396 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
397 applicable non-standard primary or :AROUND method on
398 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
399 no longer cause unbound variable errors. (reported by Kevin Reid)
400 * fixed bug: metacircle resolution in cases where methods have slots
401 added before the slots from STANDARD-METHOD. (reported by Jean
403 * fixed bug: the Power PC assembly code for calling into lisp
404 sometimes computed its return address wrongly (depending on the
405 15th bit of the address, fixed at link time). This used to
406 manifest itself in a segmentation violation while building PCL.
407 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
408 many others over the years)
409 * fixed bug: ROOM no longer reports silly numbers for stack usage if
410 the stack pointer (treated as a signed integer) is negative.
411 (thanks to Peter van Eynde)
413 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
414 * new platform: experimental support for the Windows operating
415 system has been added. (thanks to Alastair Bridgewater)
416 * fixed several bugs in and robustified the PPC FFI (including
417 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
418 * bug fix: allow non-simple string symbol names (reported by Paul
420 * bug fix: interrupt handling on NetBSD (thanks to Richard M
422 * bug fix: saving a core corrupted callbacks on x86/x86-64
423 * bug fix: closed a loophole in metacircularity detection and
424 grounding in the PCL implementation of CLOS.
425 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
426 (reported by tomppa on #lisp)
427 * bug fix: clear the direction flag on Lisp -> C transitions, as
428 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
429 (reported by Andrej Grozin and Hendrik Maryns)
430 * optimization: major improvements to GC efficiency on GENCGC platforms
431 * optimization: faster implementation of EQUAL
432 * optimization: emit more efficient opcodes for some common
433 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
434 * optimization: several other minor code-generation improvements on x86-64
435 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
437 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
438 * minor incompatible change: (SETF CLASS-NAME) and (SETF
439 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
440 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
441 by ANSI as a generic function, and for consistency (SETF
442 GENERIC-FUNCTION-NAME) is treated likewise.
443 * fixed bug #233.b: make constraint propagation notice when a variable
444 value is changed after it is referenced but before it is used
445 * fixed bug #296: no more arbitrary behaviour from filenames with
446 odd characters as --load arguments. (reported by Adam Warner)
447 * fixed bug #390: :CHARACTER-SET pathname components now work as
448 expected. (reported by Tim Daly Jr)
449 * fixed bug #391: complicated :TYPE intersections in slot
450 definitions no longer cause an error in PCL internals.
451 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
452 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
453 returning the number of octets which would be written to the
454 file-stream. (thanks to Robert J. Macomber)
455 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
456 arguments to the internal stream functions. (thanks to David
458 * bug fix: stack exhaustion is now handled on the x86/SunOS
459 platform. (thanks to Daisuke Homma)
460 * bug fix: more accurate ROOM results on GENCGC platforms
461 * optimization: improved type inference for arithmetic-for
462 index variables in LOOP
463 * optimization: faster floating-point SQRT on x86-64
464 * fixed some bugs revealed by Paul Dietz' test suite:
465 ** DOCUMENTATION returns NIL instead of "" for method combinations
466 that don't have a docstring
468 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
469 * minor incompatible change: (SETF CLASS-NAME) and (SETF
470 GENERIC-FUNCTION-NAME) are no longer generic functions, and
471 therefore are not customizeable by user code (as seems to be at
472 least permitted and maybe required by AMOP). As a consolation,
473 however, the SBCL implementation of these functions now calls
474 REINITIALIZE-INSTANCE as specified by AMOP.
475 * bug fix: slot-definition documentation is propagated as per ANSI
476 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
477 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
478 argument, as do eslotd initialization methods. (from Pascal
479 Costanza's "Closer" project)
480 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
481 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
483 * bug fix: it is now possible to have more than one subclass of
484 STANDARD-GENERIC-FUNCTION without causing stack overflow.
485 (reported by Bruno Haible, Pascal Costanza and others)
486 * bug fix: the dependent update protocol now works for generic
487 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
489 * bug fix: condition-class instances corresponding to
490 DEFINE-CONDITION forms are now created eagerly. (reported by
491 Kalle Olavi Niemitalo on comp.lang.lisp)
492 * bug fix: floating point printing is more accurate in some
493 circumstances. (thanks to Simon Alexander)
494 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
495 merged with *DEFAULT-PATHNAME-DEFAULTS*.
496 * bug fix: callbacks on OS X now preserve stack-alignment.
497 (thanks to Cyrus Harmon)
498 * enhancement: the x86-64 disassembler is much better at
499 disassembling SSE instructions. (thanks to Lutz Euler)
500 * enhancement: CHECK-TYPE on a variable now allows type-inference
501 (thanks to Vincent Arkesteijn)
502 * optimization: improved performance of EUC-JP external format.
503 (thanks to NIIMI Satoshi)
504 * optimization: performance improvements to IO on file streams of
505 :ELEMENT-TYPE CHARACTER
506 * optimization: much faster memory allocation on x86-64
507 * optimization: faster garbage collections (latency and throughput)
509 * optimization: faster bignum subtraction and division on x86 and x86-64
510 (thanks to Lutz Euler)
511 * optimization: more accurate type-inference for loops that use a
512 floating point index variable or a negative step.
514 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
515 * bug fix: add a workaround to SBCL looping infinitely at startup on
516 Linux kernels with apparently buggy implementations of personality().
517 (thanks to Svein Ove Aas)
518 * bug fix: Unicode symbols are correctly printed in LDB backtraces
519 (thanks to David Lichteblau)
520 * bug fix: local bindings shadow global symbol macros for
521 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
522 * optimization: non-open coded uses of numeric comparison operators
523 (e.g. >) no longer cons when called with more than one parameter
524 on platforms supporting dynamic-extent allocation.
525 * enhancement: saving cores with foreign code loaded is now supported
526 on MIPS/Linux in addition to the previously supported platforms.
527 * bug fix: FIND-RESTART now tests for activity, not applicability when given
528 a restart object as identifier. (reported by Helmut Eller for CMUCL)
529 * bug fix: division by zero in sb-sprof when no samples were collected
530 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
531 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
532 (thanks to Alexander Kjeldaas)
533 * new feature: ignoring the return values of destructive functions
534 like NREVERSE signals a compile-time style-warning.
535 (thanks to Kevin Reid)
536 * new feature: the alignment of alien structure fields can be
537 explicitly specified. (thanks to Cyrus Harmon)
538 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
539 and dump core on SIGQUIT
541 ** incompatible change: threads do not inherit values of specials
542 from their parents (see manual)
543 ** bug fix: threads stacks belonging to dead threads are freed by the
544 next exiting thread, no need to gc to collect thread stacks anymore
545 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
546 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
547 ** bug fix: dynamic variable and thread start related gc lossage
548 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
550 * fixed some bugs revealed by Paul Dietz' test suite:
551 ** SUBTYPEP is slightly more accurate on heinously complicated
552 CONS types where some of the members have uncertain (in the
553 NIL, NIL sense) type relationships to each other.
555 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
556 no longer purified unless :PURIFY T is explicitly specified.
557 ** Non-purified cores are significantly smaller than before
559 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
560 * new feature: timers based on Zach Beane's excellent timer package
561 * added support for the following external formats: koi8-u,
562 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
563 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
564 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
565 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
566 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
567 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
568 * incompatible change: a threaded SBCL will no longer revert to
569 non-threaded mode on non-NPTL systems, but refuse to start entirely.
570 * bug fix: interrupts are disabled until startup is complete; no
571 more sigsegvs when receiving a signal to soon
572 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
574 * bug fix: add a workaround for the memory randomization features in
575 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
576 workaround will only be in effect on systems with the proc filesystem
578 * bug fix: printing objects of type HASH-TABLE signals a
579 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
581 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
582 have been running for over 50 days. (reported by Gilbert Baumann)
583 * bug fix: the logic for getting names of functions gets less
584 confused when confronded with alternate-metaclass
585 funcallable-instances. (reported by Cyrus Harmon)
586 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
587 consistent, even on internal alternate-metaclass objects.
588 * bug fix: flush closure information collected by physical
589 environment analysis prepass before the main pass. (bug reported
591 * bug fix: compiler pack phase does not modify a hashtable, which is
592 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
593 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
594 may be expanded. (reported by Alexander <malishev> on c.l.l)
595 * bug fix: no more style-warnings when compiling code using LAST
596 under high-SPEED low-SPACE settings. (reported by David Wallin
598 * fixed some incompatibilities between SBCL's MOP and the MOP
600 ** the METAOBJECT class is now implemented;
601 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
602 STANDARD-OBJECT, as required;
603 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
604 now compatible (as required by VALIDATE-SUPERCLASS); there
605 remains a consistency requirement over the presence or absence
606 of the FUNCTION class in the superclass list of finalized
607 classes; see the manual for more details;
608 ** the :AROUND method for COMPUTE-SLOTS on
609 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
610 requested slot ordering.
612 ** bug fix: parent thread now can be gc'ed even with a live
614 ** bug fix: binding a special with PROGV to no value is not
615 the same as that symbol not having been bound (thanks to
617 ** bug fix: binding specials is thread safe (thanks to
619 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
620 and SB-SPROF do not die with 'no handler for signal XX in
621 interrupt_handle_now(..)' anymore
622 ** bug fix: WITH-TIMEOUT works with multiple running threads
623 * fixed some bugs revealed by Paul Dietz' test suite:
624 ** ENSURE-GENERIC-FUNCTION should take a method class object for
625 the :method-class keyword argument.
627 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
628 * new port: the Solaris operating system on x86 processors is now
629 mostly supported, though some rough edges in the environment
630 remain. (thanks to Daisuke Homma)
631 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
632 space for its heap. (thanks to Thiemo Seufer)
633 * optimization: SBCL on MIPS platforms now supports dynamic-extent
634 closures. (thanks to Thiemo Seufer)
635 * minor incompatible change: eof selects abort in the debugger.
636 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
637 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
639 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
640 deprecated, and will go away in a future revision of SBCL.
641 * minor incompatible change: GC-ON and GC-OFF are no longer
642 implemented with a counter, it does not matter how many times gc
643 is switched on or off
644 * bug fix: discriminating functions for generic function classes
645 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
646 make invalid assumptions about method precedence order. (reported
648 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
649 platforms. (thanks to Thiemo Seufer)
650 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
651 * bug fix: correct pathname printing with printer escaping is on.
652 (thanks to Kevin Reid)
653 * bug fix: complex VOP definitions in "user-space" no longer trigger
654 package locks. (reported by Zach Beane)
655 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
656 for funcallable-instance functions, and is no different from
658 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
659 characters in the current readtable. (reported by Nicholas Neuss)
660 * bug fix: SUBTYPEP on various CONS types returns more a more
661 accurate acknowledgment of its certainty.
662 * optimizations: REMOVE-DUPLICATES now runs in linear time on
663 lists in some cases. This partially fixes bug 384.
664 * flush all standard streams before prompting in the REPL and the
666 * bug fix: signal handling and triggering gc do not conflict
667 directly anymore, in particular a high frequency sb-sprof does
668 not prevent gc from running
669 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
670 approximation for timezone and DST information between the
671 universal time epoch and the smallest negative 32-bit time_t.
672 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
673 year 1899 so long as the encoded time is non-negative.
674 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
675 the correct answer for non-constant rotate amounts when performing
676 an inline 32-bit rotation.
678 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
679 there is only one thread in the session
680 ** bug fix: memory leak for streams created in one thread and
681 written to in another
682 ** bug fix: lockup when compiled with gcc4
683 ** bug fix: race that allows the gc to be triggered when gc is
685 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
686 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
688 ** bug fix: finished threads are gc'ed properly
689 * fixed some bugs revealed by Paul Dietz' test suite:
690 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
691 has arguments with a different set of applicable methods from
692 the orignal arguments.
693 ** The type error thrown by MAP now has a correct expected-type
695 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
696 name a compiled function.
697 ** Three MISC tests where a large bignum was improperly coerced to
698 a float that couldn't represent that bignum during type
699 derivation were fixed.
700 ** SUBTYPEP can now handle types involving the negation of a
701 list-form FUNCTION type.
702 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
703 as (COMPLEX (AND RATIO (NOT FIXNUM))).
704 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
706 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
707 * New feature: Experimental support for bivalent streams: streams
708 opened with :element-type :default now allow character and binary
709 (unsigned-byte 8) I/O
710 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
711 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
713 * bug fix: no more highly sporadic "couldn't check whether ~S is
714 readable" when reading a stream and an interrupt hits in the middle
715 of a select system call
716 * compiler better recognizes complex arrays (reported by Eduardo
718 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
719 for variables on SBCL built with linkage-tables. (reported by Luis
721 * various error reporting improvements.
722 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
723 (Thanks to Thiemo Seufer)
724 * bug fix: nested reader invokations work correctly
725 * bug fix: it is possible to have simultaneous references to foreign
726 code and foreign data with the same name.
728 ** added x86-64 support
729 ** incompatible change: the threading api now works with thread
730 objects instead of thread ids
731 ** bug fix: threads are protected from signals and interruption when
732 starting up or going down
733 ** bug fix: a race where an exiting thread could lose its stack to gc
734 ** bug fix: don't halt on infinite error in threads if possible
735 ** fixed numerous gc deadlocks introduced in the pthread merge
736 ** bug fix: fixed thread safety issues in read and print
737 ** bug fix: debugger doesn't hang on session lock if interrupted at
738 an inappropriate moment
739 ** bug fix: run-program is now thread safe(r)
740 ** bug fix: inner with-recursive-lock no longer releases the mutex
741 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
742 * fixed some bugs revealed by Paul Dietz' test suite:
743 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
744 EXPECTED-TYPE slots filled.
745 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
747 ** MAKE-CONDITION accepts classes as type-designators.
748 ** COMPILE may never return NIL.
749 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
750 range before calling Unix time functions
752 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
753 * numerous signal handling fixes to increase stability
754 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
755 * minor incompatible change: we now correctly canonize default
756 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
757 as per the MOP, rather than the historical (INITARG INITFUNCTION
758 INITFORM). (reported by Bruno Haible)
759 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
760 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
761 for more information.
762 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
763 pathname is a directory pathname.
764 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
765 * dynamic space size on PPC has been increased to 768Mb. (thanks to
767 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
768 :METACLASS argument in addition to a class name. (reported by
769 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
770 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
771 (reported by Bruno Haible; patch from Gerd Moellmann)
772 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
774 * bug fix: more cleanups to the floating point exception handling on
775 x86-64 (thanks to James Knight)
776 * bug fix: the compiler does not try to constant fold calls of
777 COERCE and BIT-* functions when they return freshly constructed
778 objects. (reported by Nikodemus Siivola and Paul Dietz)
779 * optimization: improved the MIPS versions of generic (in the
780 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
781 * optimization: direct conversion of (unsigned-byte 32) to floats on
782 the PowerPC platform.
783 * optimization: structure instances with raw slots now use less
784 memory, and probably show better memory locality. (thanks to
786 * optimization: DYNAMIC-EXTENT declarations for lists and closures
787 are treated as requests for stack allocation on the x86-64,
788 Alpha-32, PPC and SPARC platforms.
789 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
790 the wrong file descriptor; implementation of SOCKET-OPEN-P.
791 (thanks to Tony Martinez)
793 ** gcing a dead thread can no longer lead to lockups
794 ** threads block signals until they are set up properly
795 ** errno is no longer shared by threads
796 ** interrupt-thread restores the eflags register on x86
797 ** fixed some lockups due to gc/thread interaction
798 * fixed some bugs revealed by Paul Dietz' test suite:
799 ** invalid dotted lists no longer raise a read error when
801 ** signal an error if a symbol that names a declaration is used
802 as the name of a type, or vice versa
803 ** allow using the (declare (typespec var*)) abbreviation for
804 (declare (type typespec var*)) with all type specifiers
805 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
806 ** always return NIL from PPRINT-POP when OBJECT is NIL
807 ** don't signal errors when pretty-printing malformed LABELS,
808 FLET or MACROLET forms
809 ** declarations in a DOLIST body are also in scope for the
811 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
813 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
816 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
817 * fixed cross-compiler leakages that prevented building a 32-bit
818 target with a 64-bit host compiler.
819 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
820 opened with :IF-EXISTS :OVERWRITE.
821 * fixed bug 281: error for an invalid qualifier in a short-form method
822 combination method is not signalled until the faulty method is called.
823 * bug fix: iteration variable type inferrer failed to deal with open
824 intervals. (reported by Alan Shields)
825 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
826 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
827 or more runtime options were provided to the sbcl binary.
828 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
829 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
830 * optimization: there's now a fast-path for fixnum arguments in the
831 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
832 * optimization: the code generated on x86-64 is more compact thanks
833 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
834 * bug fix: floating-point exception handling now partly works on
835 x86-64. (Thanks to James Knight)
836 * improvement to the MIPS backend from Thiemo Seufer: C-style
837 64-bit long long arguments and return values to alien functions
839 * fixed some bugs revealed by Paul Dietz' test suite:
840 ** the type-error signalled from WARN has a filled-in DATUM slot.
841 ** the type-error required when a stream is not associated with
842 a file has the stream as its datum.
843 ** type-errors on single-floats on x86-64 no longer have
844 :INVALID-OBJECT as the datum
845 ** the type-errors signalled for invalid function names now have
846 a correct expected type
847 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
848 for structures defined with a :TYPE. Documentation strings for
849 typed structures are no longer immediately discarded
850 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
851 broadcast streams and synonym streams. FILE-LENGTH now also works
852 on broadcast streams.
854 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
855 * incompatible change: the --noprogrammer option, deprecated since
856 version 0.7.5, has been removed. Please use the equivalent
857 --disable-debugger option instead.
858 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
860 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
861 inoperational for a while now) has been completely removed.
862 * null lexical environments are now printed as #<NULL-LEXENV>,
863 significantly reducing the amount of clutter in typical backtraces.
864 * documentation on weak pointers, finalization, and after GC hooks
865 has been added to the manual.
866 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
867 as other specialized array types, is much faster. SUBSEQ and
868 COPY-SEQ on such arrays have also been sped up.
869 * optimization: EQL is now more efficient when at least other argument
870 is known to be of type (OR FIXNUM (NOT NUMBER)).
871 * fixed bug: the runtime is now less vulnerable to changes in the
872 size of the SBCL object on OS X, and virtual memory is reserved for
873 all spaces that need to be at a fixed address.
874 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
875 * fixed bug: finalizers and after GC hooks that cause consing are now
877 * fixed bug: compiler error messages and summaries are now printed to
878 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
879 * fixed inference of the upper bound of an iteration variable.
880 (reported by Rajat Datta).
881 * fixed bug 211e: calling local functions with duplicated constant
882 keyword argument no longer causes a bogus style warning about an
884 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
885 ftype declaration to be disregarded. (reported by Dave Roberts)
886 * fixed bug 373: caused by erronous compilation of references to alien
887 variables in the runtime on ppc/darwin.
888 * fixed bug 376: CONJUGATE type deriver.
889 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
890 (reported by Baughn on #lisp)
891 * a cross-compiler bug on non-x86 platforms has been identified and
892 fixed. (thanks to Bruno Haible)
893 * improvements to the MIPS runtime code for increased stability.
894 (thanks to Thiemo Seufer)
895 * increased the maximimum compact environment size to allow
896 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
897 * partial workaround for bug 135: don't name the function we're
898 calling for hairy cases of EVAL, so as not to accumulate one environment
899 entry per call to EVAL. (reported by Kevin Reid)
900 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
901 * optimization: type testing for non-vector arrays should be faster.
902 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
904 * bug fix: setting 31st element of a bit vector to zero did not work
906 * bug fix: redefining a class definition which failed due to a
907 previous accessor / function clash now works (but see BUGS entry
908 #380 for more problems in this area). (thanks to Zach Beane)
909 * the long form of DEFINE-METHOD-COMBINATION disables method group
910 checking when given a single method group with pattern *. (thanks
912 * on x86 compiler supports stack allocation of results of simple
913 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
914 * support for building and running on Mac OS X 10.4 "Tiger" has been added
915 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
916 * fixed some bugs related to Unicode integration:
917 ** the restarts for recovering from input and output encoding
918 errors only appear when there is in fact such an error to
920 * fixed some bugs revealed by Paul Dietz' test suite:
921 ** MISC.549 and similar: late transformation of unsafe type
922 assertions into derived types caused unexpected code
924 ** SCALE-FLOAT type deriver is less wrong.
925 ** type derivers for EXP, LOG and similar functions compute result
926 types for complex arguments better.
927 ** (MISC.563) CONJUGATE type deriver works for very restricted
929 ** out-of-line type testers for character strings are available.
930 ** EQUAL compiler transform understands specialness of objects
932 ** accessing double-floats stored on the stack now works on x86-64.
933 ** debugger internals could sometimes create invalid lispobjs,
934 resulting in GC crashes.
935 ** MISC.548: type check weakening can convert required type into
937 ** initialization forms for bindings are not in scope of free special
940 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
941 * incompatible change: thread support for non-NPTL systems has
942 been removed - locking is buggy and unreliable. A threaded
943 SBCL build will now warn at startup and refuse to create
944 new threads, unless futex support is detected in the kernel
945 * incompatible change: the top level REPL now has only an ABORT
946 restart associated with it, not TOPLEVEL and ABORT as it used to.
947 TOP and TOPLEVEL are now available as debugger commands for
948 returning to the top level.
949 * incompatible change: forms evaluated in the REPL now use the
950 global optimization policy.
951 * incompatible change: user- and system-initialization files are
952 no longer processed with LOAD, but by READ and EVAL; hence the
953 global optimization policy, startup package, readtable, etc,
955 * The .fasl file format number has been incremented because of
956 various incompatible changes.
957 * internal entry point details and argument counts no longer appear
958 in backtraces unless explicitly requested by setting
959 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
960 * built-in and standard functions no longer have names like "top
961 level local call to FOO".
962 * fixed bug 32: functions defined in non-null lexical environments
963 now have more legible printed representation
964 * fixed bug 33: functions defined in non-null lexical environemnts
965 are now more amenable to inspection by INSPECT.
966 * workaround for bug 354: XEPs no longer appear in backtraces unless
967 explicitly requested.
968 * fixed bug: receiving the signal which results from attempting to
969 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
970 SIGBUS on other *BSDs) on architectures where the C stack is also
971 the Lisp stack (x86 and x86-64) from foreign code no longer leads
972 to debugger confusion or wild instability. (reported by Cheuksan
974 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
975 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
976 (reported by Lutz Euler)
977 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
978 compiled code. (reported by Paul Dietz)
979 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
980 specialized parameter to a method to something that is not TYPEP
981 the specializer is now possible.
982 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
983 face of package deletion.
984 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
985 pathname. (reported by David Tolpin on c.l.l)
986 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
987 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
988 STRING-STREAM and FILE-STREAM, respectively.
989 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
990 than previously; in addition, its test suite is now run on build.
991 (thanks to Luke Gorrie)
992 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
993 * optimization: added a immediate representation for single-floats
995 * fixed some bugs revealed by Paul Dietz' test suite:
996 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
997 ** attempting to create a package with a colliding nickname causes
998 correctable errors to be signalled.
999 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1000 ** several x86-64 backend bugs related to sign-extension of immediate
1003 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1004 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1005 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1007 * fixed loading of multiply forward-referenced layouts.
1008 (thanks to Cheuksan Wang)
1009 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1010 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1011 (see COMPILE-FILE documentation for details of :PRINT option)
1012 * fixed bugs 19 and 317: fixed-format floating point printing is
1013 more accurate. This also fixes a bug reported by Adam Warner
1014 related to the ~@F format directive.
1015 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1017 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1018 dispatch table if the from-char is a dispatch macro character.
1019 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1020 by the wordsize no longer ignore the last word. (reported by Lutz
1022 * fixed bug in type checking of dynamic-extent variables. (reported
1024 * optimization: sequence traversal functions use their freedom to
1025 coerce function designators to functions.
1026 * optimization: code with many calls to CLOS methods specialized on
1027 CLOS classes has had redundant type checks removed.
1028 * optimization: added declarations to speed up operations that access
1029 the internal character database (for example STRING-UPCASE)
1030 * optimization: comparison operations between floats and sufficiently small
1031 fixnums no longer create extra rationals
1032 * fixed some bugs related to Unicode integration:
1033 ** portions of multibyte characters at the end of buffers for
1034 character-based file input are correctly transferred to the
1035 start of the buffer at the next read.
1036 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1037 passing it through to OPEN.
1038 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1039 argument given to internal calls to OPEN.
1040 ** the built-in comment readers (introduced by character sequences
1041 ";" and "#|") are more forgiving to encoding errors; they will
1042 STYLE-WARN and then attempt to resync the stream at a character
1043 boundary. (thanks to Teemu Kalvas)
1044 * fixed some bugs in the x86-64 port:
1045 ** Negative short int return values from c-calls are sign-extended
1047 ** The stack is aligned to 16-bytes for c-calls, as required by
1048 the ABI. (thanks to Cheuksan Wang)
1049 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1050 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1051 for enums. (thanks to Vincent Arkesteijn)
1052 ** Multiple small optimizations and bugfixes for floating point
1054 * fixed some bugs revealed by Paul Dietz' test suite:
1055 ** Space, Tab, Linefeed, Return and Page have the invalid
1056 secondary constituent character trait.
1057 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1059 ** WITH-INPUT-FROM-STRING should only update the index place on
1061 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1062 works more reliably.
1063 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1064 with one-dimensional arrays with fill pointers.
1065 ** TYPE-OF failed on a complex with an integer realpart and a
1067 ** compiler failure during type inference for the code of form
1068 (IF (EQL X (THE ...)) ...) (MISC.535).
1070 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1071 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1072 platform. The port supports 61-bit fixnums, large memory spaces
1073 and reloading shared object files.
1074 * enhancement: saving cores with foreign code loaded is now
1075 supported on ppc/Darwin in addition to the previously supported
1077 * enhancement: the statistical profiler now walks deeper into the
1078 call stack for more meaningful call-graphs and accrued time
1079 reports (x86/x86-64 only). It also now reports time spent in
1081 * enhancement: it is now possible to trace most individual methods
1082 of a generic function in addition to tracing the generic function
1084 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1086 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1087 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1088 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1089 produces an error. (thanks to Vincent Arkesteijn)
1090 * bug fix: NAMESTRING on pathnames with :WILD components in their
1091 directories works correctly. (thanks to Artem V. Andreev)
1092 * fixed bug 125: compiler preserves identity of closures. (reported
1094 * bug fixed: functions with &REST arguments sometimes failed with
1095 "Undefined function" when compiled with (DEBUG 3). (reported by
1097 * bug fix: overflow during compiling of setting element of a bit
1098 vector with constant index and value. (reported by Timmy Douglas)
1099 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1100 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1101 * on x86 compiler partially supports stack allocation of dynamic-extent
1103 * GO and RETURN-FROM do not check the extent of their exit points
1104 when compiled with SAFETY 0.
1105 * fixed some bugs related to Unicode integration:
1106 ** encoding and decoding errors are now much more robustly
1107 handled; it should now be possible to recover even from invalid
1108 input or output to the terminal. (thanks to Teemu Kalvas)
1109 ** provided a first cut at implementing STRING-TO-OCTETS and
1110 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1111 ** altered the SB-MD5 contributed module slightly, changing the
1112 interface just enough for it to be supportable for builds where
1113 lisp characters are not eight bits.
1114 * fixed some bugs revealed by Paul Dietz' test suite:
1115 ** the FORMATTER-generated functions for ~V[ conditionals require
1116 the correct number of arguments.
1117 ** READ-FROM-STRING returns the mandated second value when applied
1118 to displaced strings.
1119 ** the #\Rubout and #\Backspace characters are treated as invalid
1120 constituent characters by the tokenizer.
1122 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1123 * new feature: reloading changed shared object files with
1124 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1125 * new feature: references to foreign variables and functions
1126 can now be compiled and loaded before the corresponding shared
1127 object file is loaded, as long as the foreign definitions are
1128 available at runtime.
1129 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1130 supported. (thanks to Dan Debertin)
1131 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1132 just keywords. (thanks to Vincent Arkesteijn)
1133 * fixed bug #331: structure-class instances corresponding to
1134 DEFSTRUCT forms are now created eagerly.
1135 * fixed bug #345: backtraces from calls to undefined functions work
1136 on x86 as well. Related bug #61 is now also partially fixed on x86
1137 (backtraces from throws to unknown catch tags.)
1138 * bug fix: lambda-list parsing is now stricter vrt. order and number
1139 of lambda-list keywords.
1140 * bug fix: as specified by AMOP, an error is signalled if a
1141 class-option appears multiple times in a DEFCLASS form. (reported
1143 * bug fix: RANDOM can be compiled when the compiler derives the type
1144 of its numeric argument as a disjoint set of small integers.
1145 (reported by Paul Dietz)
1146 * bug fix: starting a core saved with shared objects loaded when
1147 those objects are not available no longer causes threaded SBCL to
1148 hang. (reported by Sean Ross)
1149 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1151 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1152 argument, which must always be NIL. (reported by Kalle Niemitalo)
1153 * bug fix: printing 1.0d+23 no longer results in an error.
1154 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1155 * bug fix: structure slot setters preserve evaluation order. (thanks
1157 * fixed some bugs related to Unicode integration:
1158 ** RUN-PROGRAM can allow its child to take input from a Lisp
1159 stream. (reported by Stefan Scholl)
1160 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1161 locales. (reported by Ken Causey)
1162 * fixed some bugs revealed by Paul Dietz' test suite:
1163 ** INCF, DECF and REMF evaluate their place form as specified in
1165 ** FORMATTER expands ~{ ~} iteration directives with V or #
1166 parameters correctly.
1167 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1168 consequent uses no arguments correctly.
1169 ** the system has a partial understanding of the (COMPLEX RATIO)
1171 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1174 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1175 * new feature: a build-time option (controlled by the :SB-UNICODE
1176 keyword feature, enabled by default) for building the system with
1177 support for the entire 21-bit character space defined by the
1179 * new feature: the system now has rudimentary external-format
1180 support; the primary user-visible change at this time is that
1181 characters with the high bit set (such as lower-case-e-acute) will
1182 print correctly to a terminal in a UTF-8 environment.
1183 * minor incompatible change: BASE-CHAR no longer names a class;
1184 however, CHARACTER continues to do so, as required by ANSI.
1185 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1186 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1187 has been moved to the SB-EXT package (temporarily re-exported from
1189 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1190 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1192 * fixed bug #351: better error-handlind and reporting for malformed
1194 * fixed bug #350: bignum-printing is now more memory-efficient,
1195 allowing printing of very large bignums, eg. (expt 2 10000000).
1196 (reported by Bruno Haible)
1197 * fixed bug #302: better primitive-type selection for intersection
1199 * fixed bug #308: non-graphic characters now all have names, as
1200 required. (reported by Bruno Haible)
1201 * bug fix: redefining a class with different superclasses now correctly
1202 removes it from the direct-subclasses of its previous superclasses.
1203 (reported by David Morse)
1204 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1205 new value now works. (reported by Bruno Haible)
1206 * bug fix: correct canonicalization of multiple non-standard slot
1207 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1208 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1209 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1210 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1211 now exists, an signals an error.
1212 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1213 during macroexpansion for non-iteration variables as well. (reported
1214 by Bruno Haible for CMUCL)
1215 * bug fix: Cyclic structures and unprintable objects in compiler
1216 messages no longer cause errors. (reported by Bruno Haible)
1217 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1218 and READ-FROM-STRING all now return a primary value of NIL if
1219 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1220 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1221 (reported by Bruno Haible for CMUCL)
1222 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1223 do not designate a condition. (reported by Bruno Haible for
1225 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1226 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1227 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1228 permitting resolution in favour of any of the conflicting symbols.
1229 (reported by Bruno Haible for CMUCL)
1230 * bug fix: EQUAL compiler optimizations is less aggressive on
1231 strings which can potentially compare true despite having distinct
1232 specialized array element types.
1233 * bug fix: unit enumerations can be defined without dividing by
1234 zero. (thanks to Vincent Arkesteijn)
1235 * FORMAT compile-time argument count checking has been enhanced.
1236 (report from Bruno Haible for CMUCL)
1237 * a partial workaround for the bug 262: the compiler does not try to
1238 inline-expand a local function doing RETURN-FROM from a deleted
1239 BLOCK. (thanks to Peter Denno for the bug report and to David
1240 Wragg for the simple test case)
1241 * fixed some bugs revealed by Paul Dietz' test suite:
1242 ** RENAME-PACKAGE allows all package designators as new package
1244 ** constraint propagation and lambda variable substitution are
1245 more cautious in dealing with partially deleted code.
1246 ** compiler sometimes lost reoptimization passes.
1247 ** CERROR, given a condition as condition designator, passes its
1248 remaining arguments to the continue format control without
1250 ** Case-altering FORMAT directives work correctly on non-ASCII
1252 ** The REAL type specifier handles bounds outside the floating
1253 point ranges without signalling FLOATING-POINT-OVERFLOW.
1254 ** Functions with IR1-transformations can create intercomponent
1255 references to global functions.
1256 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1258 ** FORMAT ~R treats a nil value for its first parameter correctly.
1260 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1261 * enhancement: saving cores with foreign code loaded is now
1262 supported on x86/NetBSD and sparc/Linux in addition to the previously
1263 supported platforms.
1264 * bug fix: on some platforms repeated installations caused multiple
1265 copies of HTML documentation to be installed -- should not happen
1266 any more. (reported by Stefan Scholl)
1267 * bug fix: parsing self-recursive alien record types multiple times
1268 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1269 original patch by Helmut Eller for CMUCL)
1270 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1271 (thanks to Richard Kreuter)
1272 * bug fix: defining classes whose accessors are methods on existing
1273 generic functions in other (locked) packages no longer signals
1274 bogus package lock violations. (reported by François-René Rideau)
1275 * bug fix: special variables as DEFMETHOD parameters no longer have
1276 associated bogus type declarations. (reported by David Wragg and
1278 * bug fix: read-write consistency on streams of element-type
1279 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1280 * bug fix: redefiniton of the only method of a generic function with
1281 no DEFGENERIC no longer emits a full WARNING. In addition,
1282 redefinition of generic functions with no DEFGENERIC to an
1283 incompatible lambda list now signals an error. (thanks to Zach
1285 * bug fix: DEFGENERIC now works even when there's a function of the
1286 same name in an enclosing lexical environment. (thanks to Zach
1288 * fixed compiler failure, caused by instrumenting code during
1289 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1290 * optimization: added loop analysis and improved register allocation
1291 * fixed some bugs revealed by Paul Dietz' test suite:
1292 ** POSITION on displaced vectors with non-zero displacement
1293 returns the right answer.
1294 ** (SIMPLE-STRING) is a valid type specifier for sequence
1296 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1298 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1299 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1301 ** The justification version of the FORMAT ~< directive treats
1302 non-zero minpad parameter correctly.
1304 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1305 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1306 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1307 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1308 the supported interface.
1309 * new feature: Single-stepping of code compiled with DEBUG 2 or
1310 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1311 * new feature: saving cores with foreign code loaded is now
1312 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1313 Timothy Moore's work for CMUCL)
1314 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1315 parameters to * instead of NIL if no initform is supplied.
1316 (reported by Johan Bockgaard)
1317 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1318 correctly when FUNCALL appears as the car of the form. Note:
1319 despite this FUNCALL forms are not currently subject to
1320 compiler-macro expansion. (port of Raymond Toy's fix for the
1321 same from CMUCL, reported by Johan Bockgaard)
1322 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1323 (thanks for Teemu Kalvas)
1324 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1325 case instead of terminating the iteration (thanks for Julian
1326 Squires, Sean Champ and Raymond Toy)
1327 * bug fix: incorrect expansion of defgeneric that caused a style
1328 warning. (thanks for Zach Beane)
1329 * bug fix: slot accessor effective method computation works properly
1330 for classes with multiple non-standard applicable methods on
1331 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1332 * on x86 compiler supports stack allocation of results of LIST and
1333 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1334 CMUCL implementation by Gerd Moellmann)
1335 * fixed some bugs revealed by Paul Dietz' test suite:
1336 ** FORMAT strings with both the ~<~:;~> form of the justification
1337 directive and pretty-printing directives cause an error.
1339 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1340 * incompatible change: the internal functions
1341 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1342 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1343 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1344 instead of the old functions.
1345 * new feature: on platforms where "dladdr" is available foreign
1346 function names now appear in backtraces. (based on Helmut Eller's
1348 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1349 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1351 * documentation: ASDF manual is now distributed with SBCL.
1352 * bug fix: SBCL can now load its contributed modules with REQUIRE
1353 even if the system-provided entries have been removed from
1354 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1356 * bug fix: backtraces involving undefined functions or assembly
1357 routines are more informative. (thanks to Brian Downing)
1358 * bug fix: mutually referent alien structures now work correctly.
1359 (reported by Rick Taube)
1360 * bug fix: structures defined by WITH-ALIEN can be referred to
1361 within other definitions in the same WITH-ALIEN.
1362 * bug fix: division operators (MOD, TRUNCATE and the like) with
1363 constant zero divisors and integer dividends no longer generate
1365 * bug fix: provide default methods for INPUT-STREAM-P and
1366 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1367 * bug fix: improve the detection and resolution of MOP metacycles
1368 (where effective method computation depends on the generic function
1369 having its effective method computed). (reported by Bruno Haible)
1370 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1371 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1372 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1373 * bug fix: pathnames with relative directory components can be
1374 represented relative to default pathnames.
1375 * optimization: in taking the GCD of bignums, reduce the two bignums
1376 to approximately the same size (using Euclid's algorithm) before
1377 applying the more sophisticated binary GCD. (thanks to Juho
1379 * optimization: COUNT on bitvectors now operates word-at-a-time.
1380 * optimization: ASH with a positive, but not necessarily constant,
1381 (leftwards) shift, when in a modular context, is compiled to a
1383 * fixed some bugs revealed by Paul Dietz' test suite:
1384 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1385 if the corresponding argument is NIL.
1386 ** FORMAT directives accepting a minpad parameter treat negative
1388 ** literal commas as character parameters to FORMAT directives are
1390 ** literal spaces directly after ~<Newline> directives within a
1391 format-logical-block (~:< ~@:>) do not induce :FILL-style
1392 conditional newlines.
1393 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1394 as specified: it no longer includes conditional newlines).
1395 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1397 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1398 :PER-LINE-PREFIX argument does not evaluate to a string.
1400 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1401 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1402 the manual for details; for now, package locks can be disabled by
1403 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1404 if no major problems are found then it is likely that they will be
1405 compiled in unconditionally.
1406 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1407 unsupported operators on all platforms. To load a shared library
1408 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1409 object file, link it into a shared library outside of SBCL and
1410 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1412 * bug workaround: ROOM T and the GC were not working together
1413 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1414 reporting facility aren't preserved. That reporting has been
1415 disabled, so now until and unless someone figures out how to make it
1416 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1417 * minor incompatible change: as threatened around sbcl-0.8.0, the
1418 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1419 The symbols are also exported from SB-PCL for backwards
1420 compatibility, but more so than before SB-PCL should be treated as
1421 an implementation-internal package.
1422 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1424 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1425 It is known as of this release not to work on the Alpha, however.
1426 * fixed bug #167: errors signalled due to illegal syntax in method
1427 bodies are now more legible.
1428 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1429 designators and can hence be used with TYPEP.
1430 * fixed bug #333: CHECK-TYPE now ensures that the type error
1431 signalled, if any, has the right object to be accessed by
1432 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1433 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1434 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1436 * fixed bug #334: programmatic addition of slots using specialized
1437 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1438 and :ALLOCATION :CLASS slots.
1439 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1440 float exponents. (rereported by Peter Seibel)
1441 * fixed a bug: #\Space (and other whitespace characters) are no
1442 longer considered to be macro characters in standard syntax by
1443 GET-MACRO-CHARACTER.
1444 * fixed bug: initialization of condition class metaobjects no longer
1445 causes an instance of the condition to be created. (reported by
1447 * fixed bug: it is now possible to have slots such that
1448 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1449 description is neither :INSTANCE nor :CLASS.
1450 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1451 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1452 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1453 non-standard SLOT-DEFINITION-ALLOCATION values.
1454 * fixed bugs: various race conditions handling exiting threads.
1455 CL-PPCRE's thread torture test now passes.
1456 * fixed arguably-a-bug: GC time proportional to number of threads in
1457 system even when most of them are idle
1458 * optimization: improved performance of BIT and SBIT on bit-vectors.
1459 * .fasl file incompatibility: The fasl file version number has
1460 been incremented because of changes associated with package locks.
1462 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1463 * minor incompatible change: the system no longer provides
1464 optimization hints (conditions which are TYPEP
1465 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1466 mode; these hints are still emitted when higher SPEED optimization
1468 * new contrib module: a sampling profiler (profiling by statistical
1469 sampling, rather than by instrumenting functions) is available as
1470 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1471 * the behaviour of the standard function ED is now customizeable by
1472 third parties through a hook variable: see ED's documentation
1473 string for information on the protocol.
1474 * the compiler no longer emits efficiency notes for (FUNCALL X)
1475 when the type of X is uncertain under default optimization
1477 * fixed bug 276: mutating a binding of a specialized parameter to a
1478 method to something that is not TYPEP the specializer is now
1480 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1481 exhibit the required behaviour on the x86 platform. (thanks to
1482 Peter van Eynde, Eric Marsden and Bruno Haible)
1483 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1484 even for difficult arguments. (reported by Peter Graves)
1485 * fixed bug 141a: the backquote printer now descends quoted
1487 * fixed another bug in backquote printing: no more destructive
1488 modification of the form's list structure. (reported by Brian
1490 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1491 move between its address being taken and the call to
1492 interrupt_thread, fixing a crashing race condition.
1493 * the SB-POSIX contrib implementation has been adjusted so that it
1494 no longer exhibits ridiculously poor performance when constructing
1495 instances corresponding to C structs.
1497 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1498 * minor incompatible change: the sb-grovel contrib now treats C
1499 structures as alien (in the sense of SB-ALIEN) objects rather than
1500 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1501 has implications for memory management of client code
1502 (sb-grovel-returned objects must now be manually managed) and for
1503 type safety (alien objects now have full types).
1504 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1505 used to control emission of compiler diagnostics, rather than the
1506 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1507 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1508 quality should be considered deprecated.
1509 * install.sh now installs the user manual as well
1510 * (not quite a new documentable feature, but worth considering in
1511 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1512 beginnings of a semantically meaningful condition hierarchy is
1513 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1515 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1516 designator as the defaults argument.
1517 * fixed bug: Displaced arrays whose displaced-to array has become
1518 too small now cause ARRAY-DIMENSION to signal an error, providing
1519 for safer bounds-checking. (reported by Bruno Haible)
1520 * fixed bug: DEFCLASS slot definitions with identical :READER and
1521 :WRITER names now signal a reasonable error. (reported by Thomas
1523 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1525 * fixed bug: Invalid slot specification errors now print correctly.
1526 (thanks to Zach Beane)
1527 * fixed bug 320: Shared to local slot value transfers in class
1528 redefinitions now happen corrently from superclasses as
1529 well. (reported by Bruno Haible)
1530 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1532 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1533 improper lists correctly. (reported by Bruno Haible)
1534 * fixed bug 313: source-transform for <fun-name> was erroneously
1535 applied to a call of a value of a variable with name <fun-name>.
1536 (reported by Antonio Menezes Leitao)
1537 * fixed bug 307: The obsolete instance protocol ensures that
1538 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1539 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1540 remove dead unknown-values globs from the middle of the stack.
1541 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1542 easier to resume long computations after using *BREAK-ON-SIGNALS*
1543 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1544 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1545 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1546 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1547 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1548 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1550 * fixed a (fairly theoretical) bug in string printing: if
1551 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1552 does not have array-element-type equal to the most general string
1554 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1555 function name. (thanks to Nikodemus Siivola)
1556 * fixed bug: compile-time format string checker failed on
1557 non-closed ~{. (reported by Thomas F Burdick)
1558 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1559 don't warn on legal loop code involving more than one aggregate
1560 boolean. (thanks to Nikodemus Siivola)
1561 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1562 in keyword position to MAKE-PACKAGE was not being reported as
1564 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1565 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1567 * fixed bugs 280 and 312: the checking for multiple definitions in a
1568 file is less likely to become confused by uses of inline
1570 * fixed bug: the #S reader macro performs the keyword coercion
1571 specified for slot names. (reported by Kalle Niemitalo)
1572 * fixed bug: lambda lists may contain symbols whose names start with
1573 & but are not lambda-list-keywords; their occurrence triggers a
1575 * fixed bug 321: define-method-combination argument lambda lists do
1576 not cause invalid code to be generated when &optional or &aux
1577 variables have default values. (reported by Bruno Haible)
1578 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1579 have CLOS classes; this appears also to have fixed reported
1580 instability in stack exhaustion detection.
1581 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1582 MUFFLE-WARNING when no associated restart is present is now
1584 * optimization: rearranged the expansion of various defining macros
1585 so that each expands into only one top-level form in a
1586 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1588 * optimization: used a previously little-used slot in symbols to
1589 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1591 * fixed some bugs revealed by Paul Dietz' test suite:
1592 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1594 ** Fixed an optimization bug related to inheritance of initforms
1595 from local to shared slots.
1596 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1597 ** CAST optimizer forgot to flush argument derived type.
1598 ** print/read consistency on floats is now orders of magnitude
1599 more likely. (thanks also to Bruno Haible for a similar report
1601 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1602 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1603 for objects of type REAL. Make it so.
1604 ** SXHASH is specified (infelicitously) to respect similarity,
1605 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1606 it so. (thanks to Markus Ziegler)
1607 ** on the Alpha, the compiler succeeds in compiling functions
1608 returning a known number of arguments greater than 63.
1609 ** fixed handling of invalid NIL arguments in keyword position
1611 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1613 ** provided a readably-printable representation for RANDOM-STATE
1615 ** ensured that pathnames, where they have a namestring, always
1616 print using #P"..." syntax.
1618 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1619 * Support for the forthcoming 2.0 version of the NetBSD kernel
1620 running on x86 hardware has been added. (thanks to Perry
1621 E. Metzger most immediately, and others for their past work)
1622 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1623 current release version; SBCL's OpenBSD support had been broken
1624 since about the time of OpenBSD's switch to ELF binary format.)
1625 * [placeholder for DX summary]
1626 ** user code with &REST lists declared dynamic-extent, under high
1627 speed or space and low safety and debug optimization policy.
1628 * The manual has been converted to Texinfo format and the debugger
1629 chapter from the cmucl manual has been added.
1630 * A facility has been added to extract documentation strings from
1631 sbcl and store them as Texinfo-formatted snippets for inclusion in
1632 the manual (via Texinfo's @include directive)
1633 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1634 the test case to Dave Roberts)
1635 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1636 pointers. (reported by Sean Ross)
1637 * bug fix: PROFILE output is printed nicely even for large numerical
1638 values. (thanks to Zach Beane)
1639 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1640 greater than 32 handle EOF correctly.
1641 * bug fix: on X86 an immediate argument of the IMUL instruction is
1642 correctly printed in disassembly. (thanks to Lutz Euler)
1643 * bug fix: class slots in redefined classes preserve their old
1644 values. (thanks to Bruno Haible and Nikodemus Siivola)
1645 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1646 succeeds. (reported by Marco Baringer)
1647 * fixed some bugs revealed by Paul Dietz' test suite:
1648 ** READ-SEQUENCE now works on ECHO-STREAMs.
1649 ** RATIONALIZE works more according to its specification. (thanks
1652 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1653 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1654 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1655 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1656 should matter to you only if you rebind the printer control
1657 variables and then find you want different bindings in the
1658 debugger than in the ordinary execution of your program.)
1659 * The runtime build system has been tweaked to support building
1660 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1661 assembler and linker. (thanks to Nikodemus Siivola)
1662 * Unbound, undefined, undeclared variables now trigger full
1663 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1664 more often programmer error than deliberate exploitation of undefined
1666 * optimization: the hash algorithm for strings has changed to one
1667 that is less vulnerable to spurious collisions. (thanks to Juho
1669 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1670 needless bounds checking. (thanks to Juho Snellman)
1671 * optimization: implemented multiplication as a modular
1672 (UNSIGNED-BYTE 32) operation on the PPC backend.
1673 * fixed some bugs revealed by Paul Dietz' test suite:
1674 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1676 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1677 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1678 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1680 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1682 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1684 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1685 their output stream on EOF from read.
1686 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1687 have been read to end-of-file.
1688 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1690 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1691 description of determination of which consecutive characters
1693 ** Printing the "Space" character with escaping on now yields "#\\ ",
1694 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1695 ** Reading floating-point numbers with *READ-BASE* set to a number
1696 less than 10 works correctly.
1697 ** Reading floating-point numbers with *READ-BASE* set to a number
1698 more than 10 works correctly.
1699 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1700 the readtable currently in effect.
1702 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1703 * minor incompatible change: parsing of namestrings on a physical
1704 (Unix) host has changed; numbers after the final #\. in a
1705 namestring are no longer interpreted as a version field. This is
1706 intented to be largely invisible to the user, except that the
1707 meaning of the namestring "*.*.*" has changed: it now refers to a
1708 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1709 should usually be replaced by
1710 (make-pathname :name :wild :type :wild :version :wild)
1711 with the added benefit that this is more likely to be portable.
1712 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1713 option to OPEN now signals an error if the file being opened
1714 exists; this may have an impact on existing code.
1715 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1716 signals received, on the PowerPC platforms (both Linux and
1717 Darwin). (thanks to Pierre Mai for pointing out the location of
1719 * several fixes on OS X: The system now builds and runs cleanly on
1720 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1721 (thanks to Brian Mastenbrook)
1722 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1723 second-resolution: integer multiples of 1/3600 between -24 and 24.
1724 (thanks to Vincent Arkesteijn)
1725 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1726 type when called with 1 argument; PEEK-CHAR checked type of
1727 PEEK-TYPE only after having read first character from a
1728 stream. (reported by Peter Graves)
1729 * bug fix: the garbage collector now has much better locality
1730 behaviour, and in particular no longer treats all memory as being
1731 exhausted when many small objects point to each other in a deeply
1733 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1734 recognized as being TYPEP their class.
1735 * bug fix: the PUSHNEW documentation string has been corrected.
1736 (thanks to Vincent Arkesteijn)
1737 * bug fix: defaulting of the value for the last of an atypically
1738 large number of multiple values being bound was not being
1739 performed correctly on the Alpha or PPC platforms
1740 * optimization: implemented multiplication as a modular
1741 (UNSIGNED-BYTE 32) operation on the x86 backend.
1742 * optimization: SEARCH on simple-base-strings can now be open-coded.
1743 (see also contrib/compiler-extras.lisp for inspiration for
1744 teaching the compiler about the Boyer-Moore algorithm).
1745 * value, returned by MAX (and MIN) called with several EQUALP, but
1746 not EQL, arguments now does not depend on compiler settings.
1747 * fixed some bugs revealed by Paul Dietz' test suite:
1748 ** in stack analysis liveness information is propagated from
1749 non-local entry points.
1750 ** pathwise CAST removing failed when the CAST node did not start
1752 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1753 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1755 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1756 host is already defined.
1757 ** RENAME-FILE works on streams instead of signalling an internal
1759 ** PEEK-CHAR uses the current readtable when determining whether
1760 or not a character is whitespace.
1761 ** MERGE-PATHNAMES handles the case when the pathname does not
1762 specify a name while the default-pathname specifies a version.
1763 ** Pathnames now stand a better chance of respecting print/read
1765 ** Attempting to use standardized file system operators with a
1766 pathname with invalid :DIRECTORY components signals a
1768 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1770 ** DIRECTORY on logical pathnames is more correct.
1771 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1772 signal a TYPE-ERROR if their argument is not a stream.
1773 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1774 designator argument does not designate a stream.
1775 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1776 examining the synonym.
1777 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1779 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1780 element-type larger than ([UN]SIGNED-BYTE 32).
1782 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1783 * When built with the :SB-FUTEX feature, threaded builds now take
1784 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1785 for faster/more reliable mutex and condition variable support.
1786 * Incompatible change (but one you probably shouldn't have been using
1787 anyway): the interface and code for arbitrating between multiple
1788 threads in the same user session has been redesigned.
1789 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1790 error when called without an explicit environment argument.
1791 (thanks to Nikodemus Siivola)
1792 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1793 confused about their position. (thanks to Adam Warner and Gerd
1795 * bug fix: Pretty printing backquoted forms with unquotations in the
1796 argument list position of various code constructs such as LAMBDA
1797 now works correctly. (reported by Paul Dietz)
1798 * bug fix: Pretty printing unquotations no longer loses all
1799 stream position information.
1800 * optimization: performance of string output streams is now less
1801 poor for multiple small sequence writes.
1802 * optimization: performance of CSUBTYPEP in the presence of complex
1803 expressions involving CONS and NOT many times has been improved.
1804 (reported by Paul Dietz)
1805 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1807 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1808 streams for the REPL, for the debugger, and so on. (thanks to
1810 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1811 * fixed some bugs revealed by Paul Dietz' test suite:
1812 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1813 is ignored if it is too small, rather than propagating through
1814 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1815 ** extremely complex negations of CONS types were not being
1816 sufficiently canonicalized, leading to inconsistencies in
1818 ** VALUES tranformer lost derived type.
1820 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1821 * fixed a bootstrapping bug: the build process no longer assumes
1822 that the various BOOLE-related constants have the same value in
1823 host and target lisps. (noted by Paul Dietz' test suite on an
1824 SBCL binary built from CLISP)
1825 * The system can now be dynamically linked on the MIPS platform,
1826 which enables dynamic loading of foreign code from Lisp. (thanks
1827 to Ralf Baechle for discussions on the MIPS ABI)
1828 * The system now records debugging information for its own source
1829 files in a filesystem-position-independent manner, relative to
1830 the "SYS" logical host.
1831 * fixed a compiler bug: MV-LET convertion did not check references
1832 to the "max args" entry point. (reported by Brian Downing)
1833 * tweaked disassembly notes to be less confident about proclaiming
1834 some instruction as an LRA. (thanks to Brian Downing)
1835 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1836 now each have their own history, command character, and other
1837 characteristics. (thanks to David Lichteblau)
1838 * fixed some bugs revealed by Paul Dietz' test suite:
1839 ** compiler failure in compiling LOGAND expressions including a
1841 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1842 shift greater than 32.
1843 ** FLUSH-DEST did not mark blocks for type check regeneration.
1844 ** HANDLER-CASE failed to accept declarations in handler clauses
1845 in some circumstances.
1847 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1848 * New code in contrib/sb-introspect (still probably not entirely
1849 stable yet) provides some support for smart Lisp development
1850 environments like SLIME.
1851 * The conditions signalled for errors occurring when loading .fasl
1852 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1853 in a way which should help ASDF recover gracefully.
1854 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1855 stuff has been cleaned up. If you code contrib/ stuff, this might
1856 affect you, and you can look at contrib/README, contrib/STANDARDS,
1857 and/or the 0.8.4.27 diff to check.
1858 * In full calls the compiler now does not generate checks for declared
1859 argument types for all arguments.
1860 * various threading fixes
1861 ** and some experimental patches which didn't make it into
1862 the main tree for this release, but which are shipped in
1863 contrib/experimental-thread.patch as a possible fix for some
1864 failures (deadlock, spinning...) in GC-intensive multithreaded
1866 * fixed PPC build problem (source code incompatibility of different
1867 library versions): added offsetof() hackery which attempts to divine
1868 where glibc maintainers put uc_mcontext today
1869 * fixed bug 282: compiler does not trust type assertions while passing
1870 arguments to a full call.
1871 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1872 &OPTIONAL VALUES type parameter.
1873 * fix bug 214: algorithm for noting rejected templates is now more
1874 similar to that of template seletion. (also reported by rydis on
1876 * fixed bug 141b: printing backquoted information readably and prettily
1877 inserts a space where necessary.
1878 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1879 (CHARACTER 10) are now reported as errors, rather than propagated
1880 as unknown types. (reported by piso on #lisp)
1881 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1882 with values NIL and :ERROR. (thanks to Milan Zamazal)
1883 * fixed bug 191c: CLOS now does proper keyword argument checking as
1884 described in CLHS 7.6.5 and 7.6.5.1.
1885 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1886 longer raise an error; further, using a list as a for-as-arithmetic
1887 counter now raises a meaningful error.
1888 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1889 now understood by sequence creation functions such as MAKE-SEQUENCE
1891 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1892 a STRING-INPUT-STREAM.
1893 * compiler enhancement: SIGNUM is now better able to derive the type
1895 * type declarations inside WITH-SLOTS are checked. (reported by
1897 * fixed some bugs revealed by Paul Dietz' test suite:
1898 ** incorrect optimization of TRUNCATE for a positive first
1899 argument and negative second.
1900 ** compiler failure in let-convertion during flushing dead code.
1901 ** compiler failure while deriving type of TRUNCATE on an
1902 interval, containing 0.
1903 ** ASH of a negative bignum by a negative bignum count now returns
1905 ** intersection of CONS types now canonicalizes properly, fixing
1906 inconsistencies in SUBTYPEP.
1908 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1909 * incompatible change: The --disable-debugger command line
1910 option now clobbers the debugger at a more fundamental
1911 level, by redefining #'INVOKE-DEBUGGER instead of by
1912 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1913 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1914 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1915 enter the debugger REPL and then suffer recursive errors
1916 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1917 while under the new system, BREAK in --disable-debugger mode
1918 terminates the system just as an unhandled error would.
1919 * fixed compiler performance when processing loops with a step >1;
1920 * bug fix: DOCUMENTATION now retrieves generic function
1921 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1922 support has been systematized, and now supports the methods
1923 specified by ANSI, along with a default method and a method for
1924 slot documentation. (reported by Nathan Froyd)
1925 * bug fix: effective methods associated with a generic function are
1926 no longer cached over a change of that generic function's method
1927 combination. (reported by Andreas Fuchs)
1928 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1929 names a non-existent directory. (thanks to Andreas Fuchs)
1930 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1931 the PPC platform, lead to stack corruption; this has been fixed.
1932 (reported by Rainer Joswig)
1933 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1934 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1935 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1936 arithmetic in general, is now much more reliable on the PPC
1938 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1939 platform now returns the right answer.
1940 * optimization: restored some effective method precomputation in
1941 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1942 precomputation is now tunable.
1943 * optimization: compiler-internal data structure use has been
1944 reviewed, and changes have been made that should improve the
1945 performance of the compiler by about 20%.
1946 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1947 simple-base-strings and simple-bit-vectors is improved.
1948 * optimization: the optimization of 32-bit logical and arithmetic
1949 functions introduced in version 0.8.3 on the x86 has been
1950 implemented on the mips, ppc and sparc platforms; an
1951 implementation of the same facility, but for 64-bit arithmetic,
1952 has been added for the alpha.
1953 * microoptimization: the compiler is better able to make use of the
1954 x86 LEA instruction for multiplication by constants.
1955 * bug fix: in some situations compiler did not report usage of
1956 generic arithmetic in (SPEED 3) policy.
1957 * bug 145b fix: compiler used wrong type specifier while converting
1958 MEMBER-types to numeric.
1959 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1961 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1962 index argument now works properly on the Alpha platform.
1963 * bug fix: floating point exception treatment on the Alpha platform
1965 * bug fix: FILE-POSITION works much better on string input and
1966 output streams. (thanks to Nikodemus Siivola)
1967 * bug fix: many threading/garbage collection symptoms sorted.
1968 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1969 might be pseudo-atomic.
1970 * internal change: Stopping for GC is now done with signals not
1971 ptrace. GC is now done in whichever thread wanted it, instead of
1973 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1975 * bug fix: result form in DO is not contained in the implicit
1977 * incompatible change: ICR structure is changed; the value part of
1978 CONTINUATION is now called LVAR; corresponding functions are
1979 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1980 * added type deriver for ISQRT (thanks to Robert E. Brown).
1981 * bug fix: better support for loading from the command line when an
1982 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1984 * fixed some bugs revealed by Paul Dietz' test suite:
1985 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1986 ** ROUND and FROUND now give the right answer when given very
1987 small float arguments.
1988 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1990 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1991 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1992 ** LCM should return a non-negative integer.
1993 ** PARSE-INTEGER returned the index of a terminator instead of the
1994 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1995 ** PARSE-INTEGER returned an incorrect index being applied to a
1997 ** LCM with two arguments of 0 returns 0 rather than signalling
1999 ** unsigned addition of a 32-bit constant with the high bit set no
2000 longer causes an internal compiler error on the x86.
2001 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2002 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2003 with negative last argument.
2004 ** byte specifiers with zero size and position no longer cause
2005 an error during type derivation.
2006 ** bignum multiplication on the Alpha platform now returns the
2008 * porting: The system now builds on SuSE AMD64, although it still
2009 generates a 32-bit binary.
2010 * .fasl file incompatibility: The fasl file version number has
2011 been incremented (because of the changes to internal compiler
2012 data structures referred to above).
2014 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2015 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2016 more accurately, on the Darwin kernel running on PowerPC hardware.
2017 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2018 * Compiler code deletion notes now signal a condition of type
2019 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2020 an associated MUFFLE-WARNING restart.
2021 * The compiler now performs limited argument count validation of
2022 constant format strings in FORMAT, and where appropriate in ERROR,
2023 CERROR and WARN. (thanks to Gerd Moellmann)
2024 * New ASDF-INSTALL contrib can be used for automatic download and
2025 installation of third-party Lisp code from CCLAN or other sites
2027 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2028 which forces another thread to execute a function supplied by the
2030 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2031 now accept and act upon their :ELEMENT-TYPE keyword argument.
2032 (reported by Martin Atzmueller, Edi Weitz)
2033 * bug fix: FILE-POSITION now accepts position designators up to
2034 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2035 is the greater. (thanks to Patrik Nordebo)
2036 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2037 Kalle Olavi Niemitalo)
2038 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2039 printable. (reported by Eric Marsden)
2040 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2041 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2042 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2043 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2044 (thanks to Stig Sandoe)
2045 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2046 UNREAD-CHAR now correctly includes the unread character in the
2047 target sequence. (thanks to Gerd Moellmann)
2048 * bug fix in threaded builds: the system can now be suspended and
2049 resumed by shell job control with minimal disruption.
2050 * bug fixes in times and timezones >2038AD
2051 * better handling of "where is GNU make?" problem in build scripts
2052 (thanks to Nikodemus Siivola)
2053 * new optimization: inside a named function any reference to a
2054 function with the same name is considered to be a self-reference;
2055 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2056 optimization quality.
2057 * new optimization on x86: logical functions and + now have
2058 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2059 used when the result is truncated to 32 bits.
2060 * VALUES declaration is partially enabled.
2061 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2062 * fixed some bugs revealed by Paul Dietz' test suite:
2063 ** The system now obeys the constraint imposed by
2064 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2065 types form a lattice under type intersection.
2066 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2067 ** ASSOC now ignores NIL elements in an alist.
2068 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2069 and (1+ MOST-POSITIVE-FIXNUM) answers.
2070 ** The addition of a method with invalid qualifiers to a generic
2071 function does not cause an error to be signalled immediately;
2072 a warning is signalled, and the error is generated only on
2073 calling the generic function.
2074 * changed .fasl file version number, in part to add type codes for
2075 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2076 obscure ANSI requirements
2078 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2079 * fixed bug 148: failure to inline-expand a local function left
2080 garbage, confusing the compiler.
2081 * fixed bugs 3cd: structure slot readers perform type check if the
2082 slot can have an invalid value (i.e. it is either not initialized
2083 or can be written with a less specific slot writer).
2084 * bug fix: the compiler now traps array references to elements off
2085 the end of an array; previously, the bounds checking in some
2086 circumstances could go off-by-one.
2087 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2089 * type declarations for array element types now obey the description
2090 on the CLHS page "Declaration TYPE", as per discussions on
2091 sbcl-help around 2003-05-08. This means that a declaration
2092 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2093 declaration, all references to BAR will be asserted or assumed
2094 (with THE, so dependent on compiler policy) to involve objects of
2095 type FOO. Note that no such declaration is implied in
2096 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2097 * declared types of functions from the "Conditions"
2098 chapter. (reported by Paul Dietz)
2099 * bug fix: CERROR accepts a function as its first argument.
2100 * bug fix: NTH an NTHCDR accept a bignum as index
2101 arguments. (reported by Adam Warner)
2102 * optimization: character compare routines now optimize comparing
2103 against a constant character. (reported by Gilbert Baumann)
2104 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2105 argument now works properly on the MIPS platform.
2106 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2107 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2108 declared to have a functional type, does not cause an error
2110 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2111 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2112 * fixed bug: if last continuation of a deleted block has a
2113 destination, this destination should be deleted too. (reported by
2115 * fixed a bug in the bootstrap process: the host compiler's values
2116 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2117 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2118 test case from Patrik Nordebo)
2119 * improved the ability of the disassembler on the PPC platform to
2120 provide helpful disassembly notes.
2121 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2122 the class in more cases than previously.
2123 * bug fix: FILE-POSITION now understands :START and :END for
2124 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2125 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2126 8) data. (thanks to David Lichteblau)
2127 * bug fix: it is possible to add a method to a generic function
2128 without lambda list.
2129 * bug fix: reader failed to signal END-OF-FILE inside an
2130 object representation. (reported by Nikodemus Siivola)
2131 * fixed some bugs revealed by Paul Dietz' test suite:
2132 ** LAST and [N]BUTLAST should accept a bignum.
2133 ** condition slot accessors are methods.
2134 ** (VECTOR NIL) is a subtype of STRING.
2136 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2137 * minor incompatible change: some nonsensical specialized lambda
2138 lists (used in DEFMETHOD) which were previously ignored now signal
2140 * minor incompatible change: the system is now aware of the types of
2141 variables in the COMMON-LISP package, and will signal errors for
2142 most violations of these type constraints (where previously they
2143 were silently accepted).
2144 * minor incompatible change: COMPILE-FILE now uses the freedom
2145 afforded (ANSI 3.2.2.3) to use derived function types for
2146 functions defined in the same file. This also permits the system
2147 to warn on static type mismatches and function
2148 redefinition. (Currently it does not work with high DEBUG level.)
2149 * minor incompatible change: VALUES declaration is disabled.
2150 * When issuing notes, the compiler now signals a condition of type
2151 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2152 restart for use in user handlers. It is expected that the
2153 COMPILER-NOTE condition will eventually become a condition
2154 supertype to a hierarchy of note types, which will then be
2155 handleable in a similar fashion. However, at the moment, no such
2156 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2157 but it's an implementation detail, not a classification for the
2159 * Changes in type checking closed the following bugs:
2160 ** type checking of unused values (192b, 194d, 203);
2161 ** template selection based on unsafe type assertions (192c, 236);
2162 ** type checking in branches (194bc).
2163 * A short form of VALUES type specifier has ANSI meaning (it has
2164 increased the number of situations when SBCL cannot perform type
2166 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2167 keywords or constants is permissible.
2168 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2169 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2170 classes. (thanks to Antonio Martinez)
2171 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2172 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2173 argument to a pretty-print stream. Instead, it leaves any such
2174 operations to DESCRIBE-OBJECT methods.
2175 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2177 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2178 lambda lists. (thanks to David Lichteblau)
2179 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2180 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2181 (thanks to Gerd Moellmann)
2182 * fixed bug 246: increased compilation speed of long
2183 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2185 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2187 * DEFCONSTANT now throws a condition of type
2188 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2189 constant to a non-EQL value; CONTINUE and ABORT restarts
2190 respectively change and preserve the value.
2191 * fixed bug 63: The code walker, part of the implementation of CLOS,
2192 is now better at handling symbol macros.
2193 * bug fix: There is no longer an internal implementation type named
2194 CL:LENGTH. (reported by Raymond Toy)
2195 * bug fix: In macro-like defining macros/special operators the
2196 implicit block does not enclose the lambda list.
2197 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2198 enclosed in parentheses) are not suitable as type specifiers, and
2199 their use properly signals an error now.
2200 * bug fix: An argument count mismatch for a type specifier in code
2201 being compiled no longer causes an unhandled error at compile
2202 time, but signals a compile-time warning.
2203 * fixed simple vector readable printing
2204 * bug fix: DESCRIBE takes more care over whether the class
2205 precedence list slot of a class is bound before accessing it.
2206 (reported by Markus Krummenacker)
2207 * bug fix: FORMATTER can successfully compile pretty-printer format
2208 strings which use variants of the ~* directive inside.
2209 * bug fix: SEARCH now applies its TEST predicate to the elements of
2210 the arguments in the correct order. (thanks to Wolfhard Buss)
2211 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2212 in type inference. (thanks to Robert E. Brown)
2213 * bug fix: Reading in symbols with an explicit package name of ""
2214 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2215 * fixed some bugs revealed by Paul Dietz' test suite:
2216 ** NIL is now allowed as a structure slot name.
2217 ** Arbitrary numbers, not just REALs, are allowed in certain
2218 circumstances in LOOP for-as-arithmetic clauses.
2219 ** Multiple class redefinitions before slot access no longer
2220 causes a type error.
2221 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2222 association between the name and a class.
2223 ** Generic functions with non-standard method-combination and over
2224 five methods all of which return constants no longer return NIL
2225 after the first few invocations. (thanks to Gerd Moellmann)
2226 ** CALL-NEXT-METHOD with no arguments now passes the original
2227 values of the arguments, even in the presence of assignment.
2228 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2230 ** DISASSEMBLE works with closures and funcallable instances.
2231 ** ADD-METHOD now returns the generic function, not the new method.
2232 ** FIND-METHOD signals an error if the lengths of the specializers
2233 is incompatible with the generic function, even if the ERRORP
2235 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2236 which its argument is a member.
2237 ** DEFCLASS only redefines the class named by its class-name
2238 argument if that name is the proper name of the class;
2239 otherwise, it creates a new class.
2240 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2241 of the UNBOUND-SLOT condition to the name of the slot.
2242 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2243 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2244 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2245 SLOT-MAKUNBOUND in the specified fashion.
2247 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2248 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2249 cross-compilation host. As a consequence, we can now bootstrap our
2250 way up to SBCL starting with a bare gcc toolchain and human-readable
2251 source code (first the source to CLISP, then the source to SBCL).
2252 * A contributed module containing a partial implementation of the
2253 simple-streams interface has been included. (thanks to Rudi
2255 * A contributed module implementing the RFC1321 Message Digest
2256 Algorithm, known as MD5, has been included.
2257 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2258 no longer has any effect, as the code controlled by this feature
2259 has been deleted. (As far as we know, no-one has ever built using
2260 this feature, and its semantics were confused in any case).
2261 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2262 work on conditions (as required by the ANSI specification),
2263 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2264 expected behaviour on conditions. Users should note, however,
2265 that such behaviour is not required by the ANSI specification,
2266 and so use of this behaviour may render their code unportable.
2267 * fixed some bugs revealed by Paul Dietz' test suite:
2268 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2270 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2271 ** SXHASH on condition objects no longer returns NIL.
2272 ** :ALLOCATION :CLASS slots are better treated; their values are
2273 updated on class redefinition, and initforms inherited from
2274 superclasses are applied.
2275 ** REMOVE-METHOD returns its generic function argument even when
2276 no method was removed.
2277 ** SHARED-INITIALIZE now initializes the values of the requested
2278 slots, including those with :ALLOCATION :CLASS.
2279 ** ALLOCATE-INSTANCE now works on structure classes defined via
2280 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2282 ** SLOT-EXISTS-P now works on conditions, as well as structures
2284 ** MAKE-LOAD-FORM now has the required methods on
2285 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2286 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2287 keyword, and now implements the SLOT-NAMES argument.
2288 ** methods with &OPTIONAL arguments no longer allow too many
2289 arguments to be passed in the call without error.
2290 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2291 option is consistent with the required arguments of the generic
2292 function lambda list.
2293 * bug fix: REQUIRE accepts a string designator. (Thanks to
2295 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2296 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2297 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2298 * bug fix: The compiler now checks for duplicated variables in macro
2300 * bug fix: SETQ on globals returns the correct value.
2301 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2302 not a COMPILER-ERROR (followed by some other strange error on
2303 choosing the CONTINUE restart).
2304 * bug fix: make.sh and friends are now more consistent in the way that
2305 they look for GNU "make".
2307 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2308 * experimental native threads support (on x86 Linux >=2.4 only).
2309 This is not compiled in by default: you need to add :SB-THREAD to
2310 the target features. See the "Beyond ANSI" chapter of the manual
2312 * fix for longstanding nonANSIism: The old distinction between
2313 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2314 The return value from CL:FIND-CLASS is now a CLOS class, and
2315 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2316 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2317 * An interface to the MetaObject Protocol, as described in Kiczales,
2318 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2319 MIT Press, 1991, is available from the SB-MOP package.
2320 * incompatible change: the SB-PCL package should now be considered
2321 a private implementation detail, and no longer a semi-private MOP
2323 * minor incompatible change: due to rearrangement for threads, the
2324 control stack and binding stack are now allocated at arbitrary
2325 addresses instead of being hardcoded per-port. Users affected by
2326 this probably have to be doing advanced things with shared
2327 libraries, and will know who they are.
2328 * minor incompatible change: Previously, all --eval forms used were
2329 processed with READ before any of them were processed with EVAL.
2330 Now each --eval form is processed with both READ and EVAL before
2331 the next --eval form is processed. (Thus package operations like
2332 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2333 work as the user might reasonably expect.)
2334 * minor incompatible change: *STANDARD-INPUT* is now only an
2335 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2337 * minor incompatible change: Y-OR-N-P is now character-oriented,
2338 not line oriented. Also, YES-OR-NO-P now works without errors.
2339 (thanks to Antonio Martinez)
2340 * sb-aclrepl module improvements: an integrated inspector, added
2341 repl features, and a bug fix to :trace command.
2342 * Known functions, which cannot be open coded by the backend, are
2343 considered to be able to check types of their arguments. (fixing
2344 a bug report by Nathan J. Froyd)
2345 * fixed a bug in computing method discriminating functions: It is
2346 now possible to define methods specialized on classes which have
2347 forward-referenced superclasses. (thanks to Gerd Moellmann)
2348 * fixed evaluation order in optional entries (reported by Gilbert
2350 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2351 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2352 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2353 required-by-AMOP NAME argument, as well as CLASS and
2354 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2355 * fixed bug 20: DEFMETHOD can define methods using names that are
2356 not the proper names of classes to designate class specializers.
2357 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2358 Unix file descriptors, instead of blowing up. (thanks to Antonio
2360 * Garbage collection refactoring: user-visible change is that a
2361 call to the GC function during WITHOUT-GCING will not do garbage
2362 collection until the end of the WITHOUT-GCING. If you were doing
2363 this you were probably losing anyway.
2364 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2365 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2366 TYPEP the latter but not the former.
2367 * The compiler issues a full WARNING for calls to undefined functions
2368 with names from the CL package.
2369 * MAP-INTO for a vector destination is open coded. (reported by
2370 Brian Downing on c.l.l)
2371 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2372 documentation string.
2373 * fixed some bugs revealed by Paul Dietz' test suite:
2374 ** COPY-ALIST now signals an error if its argument is a dotted
2376 ** Condition slots are now accessed more correctly in the presence
2377 of multiple initargs for a given slot.
2378 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2379 exclude restarts of the same name associated with a different
2381 ** DEFCLASS of forward-referenced classes with another
2382 forward-referenced class in the superclasses list no longer
2384 ** Condition slots are now initialized once each, not multiple
2385 times. (thanks to Gerd Moellmann)
2386 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2388 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2389 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2390 arguments contain duplicated elements.
2391 ** RESTART-CASE understands local macros.
2392 ** RESTART-CASE associates exactly its own restarts with a condition.
2393 ** ENDP in safe mode checks its argument to be of type LIST.
2394 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2395 in question is unbound.
2396 ** Optimized MAKE-INSTANCE functions no longer cause internal
2397 assertion failures in the presence of duplicate initargs.
2398 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2399 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2400 propagated even in the ctor optimized implementation of
2402 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2404 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2405 * a better implementation of SXHASH on (simple) bit vectors,
2406 measured both in execution speed and in distribution of results
2407 over the positive fixnums, has been installed. Likewise, a better
2408 implementation of EQUAL for simple bit vectors is now available.
2409 * fixed CEILING optimization for a divisor of form 2^k.
2410 * fixed bug 240 (emitting extra style warnings "using the lexical
2411 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2412 by Antonio Martinez)
2413 * fixed SXHASH, giving different results for NIL depending on type
2414 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2415 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2416 variable. (found by Rolf Wester)
2417 * SBCL does not ignore type declarations for special
2418 variables. (reported by rif on c.l.l 2003-03-05)
2419 * some bug fixes in contrib/sb-aclrepl/
2420 * fixed some bugs revealed by Paul Dietz' test suite:
2421 ** a bug in the CONS type specifier, whereby the CAR and CDR
2422 types got intertwined, has been fixed;
2423 ** the type system is now able to reason about the interaction
2424 between INTEGER and RATIO types more completely;
2425 ** APPEND, [N]REVERSE and NRECONC check that those their
2426 arguments, which must be proper lists, are really so;
2427 ** An array specialized to be unable to hold elements has been
2428 implemented, as required -- yes, really -- by ANSI;
2429 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2430 on malformed property lists;
2432 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2433 * incompatible packaging change: in line with Unix convention,
2434 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2435 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2436 * REQUIRE and PROVIDE are now optionally capable of doing something
2437 useful. See the documentation string for REQUIRE.
2438 * infrastructure for a managed SBCL contrib system: contributed
2439 modules in this release include:
2440 ** the ASDF system definition facility;
2441 ** an interface to the BSD Sockets API;
2442 ** an ACL-like convenience interface to the repl;
2443 (thanks to Kevin Rosenberg)
2444 ** an implementation of ROTATE-BYTE, with efficient implementation
2446 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2447 gives a closer approximation to the right answer than previously.
2448 (thanks to Raymond Toy)
2449 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2450 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2451 optional environment arguments, as required by ANSI.
2452 * fixed bugs in other functions taking environment objects, allowing
2453 calls with an explicit NIL environment argument to be compiled
2455 * fixed bug 228: primary return values from
2456 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2457 COMPILE or FUNCTION.
2458 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2459 :TYPE will no longer signal errors on innocuous objects.
2460 * fixed bug 231b: SETQ is better at respecting type declarations in
2461 the lexical environment.
2462 * fixed a bug in DEFCLASS: classes named by symbols with no or
2463 unprintable packages can now be defined.
2464 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2465 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2466 * fixed compiler failure related to checking types of functions.
2467 (reported by Robert E. Brown)
2468 * the compiler is now much more consistent in its error-checking
2469 treatment of bounding index arguments to sequence functions: in
2470 (SAFETY 3) code, errors will be signalled in almost all cases if
2471 invalid sequence bounding indices are passed to functions defined
2472 by ANSI to operate on sequences.
2473 * fixed a bug in the build procedure: documentation of SBCL-specific
2474 packages is now preserved and available in the final Lisp image.
2475 * lifted FDEFINITION lookup out of loops in the implementation of
2476 many list operations. (thanks to Robert E. Brown)
2477 * fixed a bug in the reader: the #n# reader macro now works for
2478 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2479 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2480 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2481 E. Brown and Tony Martinez respectively)
2482 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2483 SLOT-BOUNDP: in particular, it is now able to optimize them much
2484 better, and is now not vulnerable to having packages renamed.
2485 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2486 always called when a slot is not present in an instance. (thanks
2488 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2489 longer signal an error on unreading a character following EOF on
2490 the previous constituent stream. (thanks to Tony Martinez)
2491 * fixed some bugs revealed by Paul Dietz' test suite:
2492 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2493 not just nonnegative fixnums;
2494 ** the logical bit-array operators such as BIT-AND now accept an
2495 explicit NIL for their "opt-arg" argument (to indicate a
2496 freshly-consed result bit-array);
2497 ** ELT now signals an error on an invalid sequence index in safe
2499 ** the type system is now cleverer about negations of numeric
2500 types, and consequently understands the BIGNUM and RATIO types
2502 ** the type system is now cleverer about the interaction between
2503 INTEGER and RATIO types: while bugs still remain, many more
2504 cases are accurately computed;
2505 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2506 if it is in the last clause;
2507 ** CONSTANTLY now correctly returns a side-effect-free function in
2509 ** DECLARE is no longer treated as a special-operator; in
2510 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2511 * incremented fasl file version number due to the change in the
2512 DEFSTRUCT-SLOT-DESCRIPTION structure.
2514 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2515 * minor incompatible change: code processed by the "interpreter" or
2516 EVAL now has a compilation optimization policy of (DEBUG 2)
2517 (changed from (DEBUG 1)) to improve debuggability of interactive
2518 development, and to allow the use of the debug RETURN command in
2520 * an experimental implementation of the RETURN command for the
2521 debugger has been included. (thanks to Frederik Kuivinen)
2522 * fixed bug 62: constraints were not propagated into a loop.
2523 * fixed bug in embedded calls of SORT (reported and investigated by
2525 * fixed some bugs revealed by Paul Dietz' test suite:
2526 ** printing and reading of arrays with some dimensions having
2527 length 0 (thanks to Gerd Moellmann);
2528 ** BOA constructor with &AUX argument without a default value does
2529 not cause a type error;
2530 ** CONSTANTP now returns true for all self-evaluating objects.
2532 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2533 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2534 accessors that are related by inheritance, as specified in the
2535 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2536 Valtteri Vuorikoski)
2537 * The compiler is now able to inline functions that were defined in
2538 a complex lexical environment (e.g. inside a MACROLET).
2539 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2541 * Support for the upcoming FreeBSD-5.0 release has been included.
2542 (thanks to Dag-Erling Smorgrav)
2543 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2544 effect when it is not in a toplevel context.
2545 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2546 stand a better chance of being correct. (thanks to Gerd
2548 * fixed bug in COERCE, which now signals an error on coercing a
2549 rational to a bounded real type which excludes the expected
2551 * The compiler is now able to derive types more accurately from the
2552 COERCE and COMPILE functions.
2553 * fixed bug 223: functional binding is considered to be constant
2554 only for symbols in the CL package.
2555 * fixed bug 231: SETQ did not check the type of a variable being set
2556 (reported by Robert E. Brown)
2557 * A new optimization for MAKE-INSTANCE has been included, fixing
2558 various bugs (including relating to :ALLOCATION :CLASS slots and
2559 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2561 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2562 ** As required by ANSI, LOOP now disallows anonymous collection
2563 clauses such as COLLECT I in conjunction with aggregate boolean
2564 clauses such as THEREIS (= I 1);
2565 ** LOOP now signals an error when any variable is reused in the
2566 same loop (including the potentially useful construct analogous
2567 to WITH A = 1 WITH A = (1+ A);
2568 ** IT is only a special loop symbol within the first clause of a
2569 conditional loop clause;
2570 ** LOOP with a typed iteration variable over a hashtable now
2571 signals a type error iff it should.
2572 * fixed some other bugs revealed by Paul Dietz' test suite:
2573 ** FILE-STREAM now names the class previously known as FD-STREAM;
2574 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2575 argument) no longer signals an error;
2576 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2577 of the slot symbol, rather than using the current package
2578 ((:CONC-NAME "") continues to intern the slot's name in the
2580 * incremented fasl file version number, because of the incompatible
2581 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2582 of the new implementation of DEFINE-COMPILER-MACRO.
2584 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2585 * Support for building SBCL for MIPS platforms running in
2586 little-endian mode has now been checked in, and basic
2587 functionality on said platforms verified.
2588 * minor incompatible change: PCL now records the pathname of a file
2589 in which methods and the like are defined, rather than its
2591 * minor incompatible change: TRUENAME now considers the truename of
2592 a file naming a directory to be the pathname with :DIRECTORY
2593 component indicating that directory.
2594 * minor incompatible change: a NAMED clause in the extended form of
2595 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2596 reason for the previous behaviour is unclear.
2597 * more systematization and improvement of CLOS and MOP conformance
2598 in PCL (thanks to Gerd Moellman and Pierre Mai):
2599 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2601 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2602 primary methods with no specializers;
2603 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2605 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2606 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2607 FORWARD-REFERENCED-CLASSes; error reporting on
2608 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2610 ** SXHASH on CLOS instances now uses a slot internal to the
2611 instance to return different numbers on distinct instances,
2612 while preserving the same return value through invocations of
2614 ** DEFMETHOD signals errors when methods with longer incongruent
2615 lambda lists are added to generic functions;
2616 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2617 CLASS, as specified in AMOP;
2618 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2619 on the order returned by the primary method for classes of
2620 class STANDARD-CLASS;
2621 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2622 * fixed some bugs shown by Paul Dietz' test suite:
2623 ** DOLIST puts its body in TAGBODY;
2624 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2626 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2627 value producing form;
2628 ** if more variables are given to PROGV than values, extra
2629 variables are bound and made to have no value;
2630 ** NSUBSTITUTE on list arguments gets the right answer with
2632 ** ELT signals an error of type TYPE-ERROR when the index argument
2633 is not a valid sequence index;
2634 ** LOOP signals (at macroexpansion time) an error of type
2635 PROGRAM-ERROR when duplicate variable names are found;
2636 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2637 ** FUNCALL of special-operators now cause an error of type
2639 ** PSETQ now works as required in the presence of side-effecting
2640 symbol-macro places;
2641 ** NCONC accepts any object as its last argument;
2642 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2644 ** loop-for-as-package does not require a package to be explicitely
2646 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2648 * fixed bug 166: compiler preserves "there is a way to go"
2649 invariant when deleting code.
2650 * fixed bug 172: macro lambda lists with required arguments after
2651 &REST arguments now cause an error to be signalled. (thanks to
2653 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2654 bug. (thanks to Matthew Danish)
2655 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2657 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2658 arguments are lexically rebound. (thanks to Gerd Moellmann and
2660 * fixed bug 194: error messages are now more informative when there
2661 is no primary method applicable in a call to a generic
2662 function. (thanks to Gerd Moellmann)
2663 * fixed bug in command line argument checking (thanks to Julian
2665 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2666 (thanks to Matthew Danish)
2667 * incremented fasl file version number, because of the
2668 SXHASH-related changes in the layout of CLOS data structures
2670 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2671 * minor incompatible change: The runtime (the Unix executable named
2672 "sbcl") is now much pickier about the .core files it will load.
2673 Essentially it now requires .core files to descend from the same
2674 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2675 as the runtime does. (The intent is to prevent the crashes which
2676 can occur, and which can even be reported as mysterious failures,
2677 when people patch the sources or change the build parameters
2678 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2679 sbcl and .core files.)
2680 * fixed bug: VALUES-LIST is no longer optimized away.
2681 * fixed bug 142: The FFI conversion of C string values to Lisp
2682 string values no longer conses excessively. (thanks to Nathan
2683 Froyd porting Raymond Toy's fix to CMU CL)
2684 * began to systematize and improve MOP conformance in PCL (thanks to
2685 Nathan Froyd, Gerd Moellman and Pierre Mai):
2686 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2688 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2689 ** FINALIZE-INHERITANCE is now called on class finalization;
2690 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2691 argument precedence order.
2692 * fixed bug 202: The compiler no longer fails on functions whose
2693 derived types contradict their declared type.
2694 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2695 so it can be non-toplevel.
2696 * The fasl file version number has changed (because of the new
2697 implementation of DEFMACRO).
2698 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2699 safe code, that any length requirement by their type-specifier
2700 argument is valid. The exceptions to this are described in bug
2702 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2703 functions now check the types of their inputs as required by ANSI.
2704 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2705 introduced symbol is DECLAREd to be SPECIAL.
2706 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2707 * fixed bug: :COUNT argument to sequence functions may be negative
2708 * fixed bug: body of DO-SYMBOLS may contain declarations
2709 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2710 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2711 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2712 symbol macro only once
2713 * fixed printing of call frame when argument list is unavailable
2714 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2715 * compiler no longer signals WARNING on unknown keyword
2718 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2719 * A beta-quality port to the mips architecture running Linux,
2720 based on the old CMUCL backend, has been made. It has been tested
2721 on a big-endian kernel, and works sufficiently well to be able to
2722 rebuild itself; it has not been tested in little-endian mode.
2723 * fixed an inconsistency between gencgc.c and purify.c which made
2724 dumping/loading .core files unreliable
2725 * fixed bug 120a: The compiler now deals correctly with IFs where
2726 the consequent is the same as the alternative, instead of
2727 misderiving the return type. (thanks to Alexey Dejneka)
2728 * fixed bug 113: Logical pathnames are now dumpable (the logical
2729 host is resolved at load-time, throwing an error if it is not
2731 * fixed bug 174: FORMAT's error message is slightly clearer when a
2732 non-printing character is used in a format directive.
2733 * fixed several bugs in compiler checking of type declarations, i.e.
2734 violations of the Python "declarations are assertions" principle
2735 (thanks to Alexey Dejneka)
2736 * fixed several bugs in PCL's error checking (thanks to Gerd
2738 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2740 * fixed bug in compilation of functions as first class values
2741 (thanks to Antonio Martinez-Shotton)
2742 * The compiler's handling TYPE-ERRORs which it can prove will
2743 inevitably happen at runtime has been cleaned up and corrected
2744 in several ways. (thanks to Alexey Dejneka)
2745 * improved argument type checking for various basic arithmetic
2746 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2747 lavished on them in the past that they can be compiled in many
2748 ways in different special cases
2749 * fixed bug 181: compiler checks validity of user supplied type
2751 * cleaned up code flushing in optimization: Function calls which
2752 should signal errors for safety purposes (e.g. which ANSI says
2753 should signal errors when their arguments are of incorrect type)
2754 are no longer optimized away.
2755 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2756 * incremented fasl file version number, because changes in the
2757 implementation of sequence functions like COERCE caused
2758 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2759 in old inline expansions) to become undefined. (Actually these
2760 changes were later undone, so we might very well be binary
2761 compatible with 0.7.7 after all, but leaving the version number
2762 incremented seemed like the simplest and most conservative
2765 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2766 * An alpha-quality port to the parisc architecture running Linux,
2767 based on the old CMUCL backend, has been made. This, even more so
2768 than the other backends, should be considered still a work in
2769 progress; known problems include that the Linux kernel in 64-bit
2770 mode does not propagate the correct sigcontext structure to
2771 userspace, and consequently SBCL on a parisc64 kernel will not
2773 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2774 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2775 suppressed.) Also now that the compiler is looking at declarations
2776 in the environment, it checks optimization declarations as well,
2777 and suppresses inlining when (> DEBUG SPEED).
2778 * More fixes have been made to treatment of floating point exception
2779 treatment and other Unix signals. In particular, floating point
2780 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2781 * The detection and handling of control stack exhaustion (infinite
2782 or very deeply nested recursion) has changed. Stack exhaustion
2783 detection is now done by write-protecting pages at the OS level
2784 and applies at all optimization settings; when found, a
2785 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2786 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2787 be caught using IGNORE-ERRORS.
2788 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2789 correctly when transforming recursive LABELS functions to LETs.
2790 (thanks to Alexey Dejneka)
2791 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2792 that are names of constants or global variables.
2793 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2794 alien routines with docstrings.
2795 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2796 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2798 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2799 (thanks to Eric Marsden)
2800 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2801 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2802 * Bug fix: The SPARC backend can now compile functions involving
2803 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2804 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2805 to a functions expecting a PATHNAME-DESIGNATOR.
2806 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2807 lambda lists. (thanks to Alexey Dejneka)
2808 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2809 functions so that EXPORTing the name of the class doesn't cause
2810 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2811 to Antonio Martinez for reporting this)
2812 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2813 constant-folded without failing an assertion. (thanks to Einar
2814 Floystad Dorum for reporting this)
2815 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2816 types are now dealt with more correctly by the compiler.
2817 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2818 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2819 behaviour analogous to RENAME-FILE. This puts its behaviour more
2820 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2822 * The fasl file version number has changed again. (because of the
2823 bug fix involving the names of PCL MAKE-INSTANCE functions)
2825 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2826 * bug fix: Floating point exceptions are treated much more
2827 consistently on the x86/Linux and PPC/Linux platforms.
2828 * Array initialization with :INITIAL-ELEMENT is now much faster for
2829 cases when the compiler cannot open code the array creation, but
2830 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2831 array accesses have also seen a speed increase.
2832 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2833 specified. (This was at the root of some bad interactions between
2834 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2836 * bug fix: Internal error arguments for undefined functions are now
2837 computed correctly on the PPC/Linux platform.
2838 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2839 Raymond Toy's patch for CMU CL)
2840 * Support for the Solaris 9 operating environment has been included
2841 (thanks to Daniel Merritt)
2842 * A very ugly but hopefully complete draft of the missing FFI chapter
2843 of the manual has been created by reformatting the corresponding
2844 CMU CL manual chapter into (currently very ugly and incoherent)
2845 DocBook and bringing it up to date for SBCL behavior. Thus, the
2846 manual is now essentially complete, at least by my extreme
2847 once-and-only-once standards, whereby it's acceptable to refer to
2848 the doc strings of SB-EXT functions as the primary documentation.
2849 * The fasl file version number has changed again, due to cleanup of
2850 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2851 is no longer a static symbol.)
2853 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2854 * SBCL now builds with OpenMCL (version 0.12) as the
2855 cross-compilation host; also, more progress has been made toward
2856 bootstrapping under CLISP.
2857 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2859 * bug 158 fixed: The compiler can now deal with integer loop
2860 increments different from 1; fixing this turned out also to fix
2862 * bug 169 fixed: no more bogus warnings about using lexical bindings
2863 despite the presence of perfectly good SPECIAL declarations (thanks
2864 to David Lichteblau)
2865 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2866 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2867 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2869 * bug 180 fixed: Method combination specifications no longer ignore
2870 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2871 * bug fix: Structure type predicate functions now check their argument
2872 count as they should.
2873 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2874 correctly. (thanks to Pierre Mai)
2875 * minor incompatible change: The --noprogrammer option is deprecated
2876 in favor of the new --disable-debugger option, which is very similar.
2877 (The major difference is that it takes effect at a slightly different
2878 time at startup, causing handling of errors in --sysinit and
2879 --userinit files will be affected differently.) The
2880 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2881 been added to allow this functionality to be controlled from ordinary
2882 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2883 maintainers, who might want to run non-interactive scripts to
2884 build SBCL cores which will later be used interactively.)
2885 * minor incompatible change: The LOAD function no longer, when given
2886 a wild pathname to load, loads all files matching that pathname.
2887 Instead, an error of type FILE-ERROR is signalled.
2889 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2890 * bug 147 fixed: The compiler preserves its block link/count
2891 invariants more correctly now so that it doesn't crash. (thanks
2893 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2895 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2896 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2897 no-value-for-this-character as NIL (as specified by ANSI).
2898 * HOST-NAMESTRING on physical pathnames now returns a string that is
2899 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2900 (thanks to Christophe Rhodes)
2901 * The Alpha port handles icache flushing more correctly. (thanks to
2903 * More progress has been made toward bootstrapping under CLISP. (thanks
2904 to Christophe Rhodes)
2905 * The fasl file format has changed again, because dynamic loading
2906 on OpenBSD (which has non-ELF object files) motivated some cleanups
2907 in the way that foreign symbols are transformed and passed around.
2908 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2909 is no longer treated as whitespace by the reader, but instead as
2910 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2911 |A
\7fB|, instead of A as it used to.
2913 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2914 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2915 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2916 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2917 this as of 0.7.1.45, but was left out of the previous news section
2918 (thanks to Dan Barlow)
2919 * SBCL now runs on the Solaris operating system on SPARC architectures
2920 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2921 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2922 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2923 * SPARC backend cleanups, allowing builds of cores optimized for V8
2924 and V9 SPARCS, and also emission of code targeted to a particular
2925 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2927 * SBCL is closer to bootstrapping under CLISP, thanks to various
2928 fixes by Christophe Rhodes.
2929 * The fasl file format has changed again, to allow the compiler's
2930 INFO database to support symbol macros.
2931 * The user manual (in doc/) is formatted into HTML more nicely.
2932 (thanks to coreythomas)
2933 * The system is smarter about SUBTYPEP relationships, especially
2934 those involving NOT types (including types such as ATOM which are
2935 represented internally using NOT types). Thus SUBTYPEP is less
2936 likely to return (VALUES NIL NIL) in general, and in particular
2937 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2939 * The fasl file format has changed again, because the internal
2940 representation of types now includes a new slot to support the new
2941 SUBTYPEP-of-NOT-types logic.
2942 * (not a change in the main branch of SBCL, but a related prototype
2943 which can hopefully be merged into the main branch of SBCL in the
2944 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2945 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2946 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2947 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2948 dispatch macro character combinations. (thanks to Alexey Dejneka)
2949 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2950 that superficially look like logical namestrings correctly.
2951 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2953 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2954 default supplied-p) for &optional and &key arguments. (thanks to
2957 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2958 * incompatible change: The compiler is now less aggressive about
2959 tail call optimization, doing it only when (> SPACE DEBUG) or
2960 (> SPEED DEBUG). (This is an incompatible change because there are
2961 programs which relied on the old CMU-CL-style behavior to optimize
2962 away their unbounded recursion which will now die of stack overflow.)
2963 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2964 for non-GENCGC systems has been increased to 20M (since that
2965 seems much closer to the likely performance optimum for modern
2966 systems than the old 4M value was)
2967 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2968 and *DEBUG-PRINT-LEVEL*
2969 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2970 of CMU CL's support for SPARC, and various endianness and other
2971 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2972 * new syntactic sugar for the Unix command line: --load foo.bar is now
2973 an alternate notation for --eval '(load "foo.bar")'.
2975 ** The system now detects stack overflow and handles it gracefully,
2976 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2977 optimization settings. (This is a good thing in general, and
2978 its introduction in this version should be particularly timely
2979 for anyone whose code fails because of suppression of tail
2981 ** The system now hunts for the C variable "environ" in a more
2982 devious way, to avoid segfaults when the C library version differs
2983 between compile time and run time. (thanks to Christophe Rhodes)
2984 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2985 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2986 ** The compiler no longer issues bogus style warnings for undefined
2987 classes in the same source file as the DEFCLASSes which defined
2988 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2990 ** fixes in CONDITION class precedence list for undefined function
2991 errors (thanks to Alexei Dejneka)
2992 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2993 correctly. (thanks to Dan Barlow)
2994 ** portability fixes aiming at bootstrapping under CLISP (thanks
2995 to Dave McDonald and Christophe Rhodes)
2996 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2997 ** fixes in type translation and and type inference (thanks to
2999 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3000 ** various fixes in the new ports (thanks to Dan Barlow)
3001 * several changes related to debugging:
3002 ** suppression of tail recursion, as noted above
3003 ** stack overflow detection, as noted above
3004 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3005 is now the default. (For some time encapsulation has been more
3006 reliable than the breakpoint-based :ENCAPSULATE NIL
3007 implementation, at least on X86 systems; and I just noticed that
3008 encapsulation also seems closer to the spirit of the ANSI
3011 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3013 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3014 up properly again. (There was a packaging bug in 0.7.0 which
3015 left their definitions in SB-SYS::LOAD-FOREIGN and
3016 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3017 vital for most things which interface to C-level interfaces,
3018 like extensions working with sockets or databases or
3019 Perl-compatible regexes or whatever, and the need to fix
3020 this bug is the main reason that 0.7.1 was released so
3022 ** DEFGENERIC is now choosier about the methods it redefines, so that
3023 reLOADing a previously-LOADed file containing DEFGENERICs does
3024 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3025 cycle now works as it should. (thanks to Alexey Dejneka)
3026 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3027 fixed by Christophe Rhodes. (He actually submitted this patch
3028 months ago, and I delayed until after 0.7.0.)
3029 ** Bug 111 (internal compiler confusion about runtime checks on
3030 FUNCTION types) was fixed by Alexey Dejneka.
3031 * Some internal cleanups (getting rid of variables which aren't
3032 needed now that the byte interpreter is gone) caused the fasl
3033 file format number to change again.
3035 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3036 * major incompatible change: The default fasl file extension, i.e. the
3037 default extension for files produced by COMPILE-FILE, has changed
3038 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3040 ** There are many changes in the implementation of the compiler.
3041 SBCL is now essentially a compiler-only implementation of ANSI
3042 Common Lisp. EVAL still "interprets" a few special cases, but
3043 almost all the interesting cases are handled by creating
3044 a LAMBDA expression, calling COMPILE on it, then calling
3045 FUNCALL on the result.
3046 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3047 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3048 Since the code is newer, there might still be some new bugs
3049 (though not as many as before Martin Atzmueller's fixes:-). But
3050 the new code is substantially simpler and clearer, and hopefully
3051 any remaining bugs will be simpler, less fundamental, and more
3052 fixable then the bugs in the old code.
3053 ** The revised compiler is still a little unsteady on its feet.
3055 *** The debugging information it produces (particularly the names
3056 of FUNCTION objects) is sometimes much less useful than what
3057 the old compiler produced.
3058 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3059 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3060 within a MACROLET) has been temporarily weakened.
3061 ** There are new compiler optimizations for various functions:
3062 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3063 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3064 *** the math functions TRUNCATE, FLOOR, and CEILING
3065 *** the function-of-all-trades COERCE
3066 Mostly these should be transparent, but there's one
3067 potentially-annoying problem (bug 117): when the compiler
3068 inline-expands a function and does type analysis on the result,
3069 it can create control paths which have type mismatches, and
3070 when it can't prove that those control paths aren't taken,
3071 it will issue WARNINGs about the type mismatches. This is
3072 a particular problem in practice for the new sequence functions.
3073 It's not clear how this should be fixed, and for now, a
3074 workaround is given in the entry for 117 in the BUGS file.
3075 ** (Because of the interaction between the two previous items --
3076 occasional inlining problems and new inline expansions -- some
3077 of the new sequence function optimizations won't really kick in
3078 completely until debugging information, and then inlining, are
3079 straightened out in some future version.)
3080 * minor incompatible changes:
3081 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3082 DIRECTORY no longer implicitly promotes NIL slots of its
3083 pathname argument to :WILD. In particular, when you ask for the
3084 contents of a directory (which you used to be able to do without
3085 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3086 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3087 ** changes in behavior that ANSI explicitly defines to be
3088 implementation dependent:
3089 *** The new compiler-only implementation still conforms with ANSI,
3090 but acts a little different than before. Besides the obvious
3091 changes in performance tradeoffs (that the cost per form passed
3092 to EVAL has gone up, and the cost per form executed by EVAL
3093 has gone down), the behavior of the system changes a little
3094 because there are no longer any interpreted function objects.
3095 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3096 e.g. doing COMPILE on the output of interactive DEFUN is
3098 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3100 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3101 has changed from (:CL) to NIL.
3102 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3103 e.g. #\Newline and #\Tab, don't have names specified in the
3104 ANSI Common Lisp standard, is now based on their ASCII symbolic
3105 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3106 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3107 are no longer used for output.
3108 ** changes in internal implementation constants:
3109 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3110 4 million. (If your application spends a lot of time GCing and
3111 you have a lot of RAM, you might want to experiment with
3112 increasing it even more.)
3113 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3114 However, almost all old code should still continue to work without
3115 immediate update, as SB-C-CALL is now a (deprecated) nickname
3117 ** Old operator names in the style DEF-FOO are now deprecated in
3118 favor of new corresponding names DEFINE-FOO, for consistency with
3119 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3120 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3121 internal symbols, but a few supported extensions like
3122 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3123 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3124 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3125 etc. as you get deeper into recursive calls to the debugger
3126 command loop, instead of the old "5]", "5]]", "5]]]"
3127 sequence. (I was motivated to do this when squabbles between
3128 ILISP and SBCL left me very deeply nested in the debugger. In the
3129 short term, this change will probably provoke more ILISP/SBCL
3130 squabbles, but hopefully it will be an improvement in the long run.)
3131 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3132 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3133 function, since declaiming return types involving aliens is
3134 (1) annoyingly messy to do by hand and (2) vital to efficient
3135 compilation of code which calls such functions.
3136 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3137 longer reexported by the SB-EXT package. They're solely useful
3138 for alien code, so it seems more logical that you should get
3139 them from the SB-ALIEN package, not in SB-EXT.
3140 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3141 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3142 features. Instead, the code that they used to control is always
3143 built into the system.
3144 * many other bug fixes
3145 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3146 advantage of the new EVAL-WHEN stuff and to clean them up in
3147 general, and they are now more ANSI-compliant in a number of
3148 ways. Martin Atzmueller is responsible for a lot of this.
3149 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3151 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3152 *** correct ERROR type for various file operations
3153 *** some fixes for Lisp streams
3154 *** DEFMETHOD syntax checking
3155 *** changing old weird representation of debug information as
3156 strings (which, among their other deficiencies, don't transform
3157 correctly when you rename packages, and don't change their
3158 print representation when you change things like *PACKAGE*
3159 and *PRINT-LENGTH*) to symbols and lists of symbols
3160 He also made several improvements and fixed several bugs in DESCRIBE.
3161 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3163 *** misbehavior of WRITE-STRING/WRITE-LINE
3164 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3165 and several other LOOP problems as well
3166 *** DIRECTORY when similar filenames are present
3167 *** DEFGENERIC with :METHOD options
3168 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3169 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3170 *** argument ordering in FIND with :TEST option
3171 *** mishandled package designator argument in APROPOS-LIST
3172 *** various problems in the backquote readmacro
3173 *** a bug in APROPOS
3174 *** probably some others that I'm not describing very well here,
3175 since the CVS log documents them by reference to sbcl-devel
3176 messages, and the SourceForge archives aren't working well.:-(
3177 ** Dan Barlow improved the Alpha port (and is making progress on the
3178 PPC port, for those of you who think different).
3179 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3180 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3181 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3182 continues to work on the SPARC port (for those of you in a position
3183 to look down upon our little PC-compatible boxes from a great height).
3184 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3185 and uses that copy, rather than the current dynamic value, when
3186 it's trying to decide whether to truncate output. Thus e.g.
3187 (let ((*print-lines* 50))
3188 (pprint-logical-block (stream nil)
3190 (let ((*print-lines* 8))
3191 (print (aref possiblybigthings i) stream)))))
3192 should now truncate the logical block only at 50 lines, instead of
3193 often truncating it at 8 lines, as it did before.
3194 * The doc/cmucl/ directory, containing old CMU CL documentation
3195 from the time of the fork, is no longer part of the base system.
3196 SourceForge has shut down its anonymous FTP service, and with it
3197 my original plan for distributing the old CMU CL documentation
3198 there. For now, if you need these files you can download an old
3199 SBCL source release and extract them from it.
3200 * The fasl file version number changed again, for dozens of reasons,
3201 some of which are apparent above.
3203 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3204 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3205 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3206 patch, so that DISASSEMBLE output is much nicer.
3207 * The code in the SB-PROFILE package now seems reasonably stable.
3208 I still haven't decided what the final interface should look like
3209 (I'd like PROFILE to interact cleanly with TRACE, since both
3210 facilities use function encapsulation) but if you have a need
3211 for profiling now, you can probably use it successfully with
3212 the current CMU-CL-style interface.
3213 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3214 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3215 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3216 link now returns the dangling link itself, and for similar
3217 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3218 link itself. (In these cases the old code signalled an error and
3219 looped endlessly, respectively.) Thus, DIRECTORY now works even
3220 in the presence of dangling and cyclic symbolic links.
3221 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3222 is now a supported extension again, since the consensus on
3223 sbcl-devel was that it can be useful for ordinary development
3224 work, not just for debugging SBCL itself.
3225 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3226 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3227 that currently-defined functions might be redefined later with
3228 different return types.
3229 * Hash tables can be printed readably, as inspired by CMU CL code
3230 of Eric Marsden and SBCL code of Martin Atzmueller.
3231 * better error handling in CLOS method combination, thanks to
3232 Martin Atzmueller porting Pierre Mai's CMU CL patches
3233 * more overflow fixes for >16Mbyte I/O buffers
3234 * A bug in READ has been fixed, so that now a single Ctrl-D
3235 character suffices to cause end-of-file on character streams.
3236 In particular, now you only need one Ctrl-D at the command
3237 line (not two) to exit SBCL.
3238 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3240 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3241 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3242 does the right thing.
3243 * The implementation of some type tests, especially for CONDITION
3244 types, is now tidier and maybe faster, due to CMU CL code
3245 originally by Douglas Crosher, ported by Martin Atzmueller.
3246 * Some math functions have been fixed, and there are new
3247 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3248 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3249 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3250 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3251 likely to be built into sbcl-0.7.0.)
3252 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3253 CL, ported to SBCL by Martin Atzmueller)
3254 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3255 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3256 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3257 * There's a new slam.sh hack to shorten the edit/compile/debug
3258 cycle for low-level changes to SBCL itself, and a new
3259 :SB-AFTER-XC-CORE target feature to control the generation of
3260 the after-xc.core file needed by slam.sh.
3261 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3262 COMPILE-FILE is no longer supported, so that now every function
3263 gets an entry point, so that block compilation looks a little
3264 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3265 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3266 returns the number of bytes consed since the system started,
3267 rather than the number consed since the first time the function
3268 was called. (The new definition parallels ANSI functions like
3269 CL:GET-INTERNAL-RUN-TIME.)
3270 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3271 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3272 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3273 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3274 * Fasl file version numbers are now independent of the target CPU,
3275 since historically most system changes which required version
3276 number changes have affected all CPUs equally. Similarly,
3277 the byte fasl file version is now equal to the ordinary
3280 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3281 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3282 is no longer recognized. I apologize for this, because it was
3283 listed in SB-EXT as a supported extension, but I found that
3284 its existing behavior was poorly specified, as well as incorrectly
3285 specified, and it looked like too much of a mess to straighten it
3286 out. I have enough on my hands trying to get ANSI stuff to work..
3287 * many patches ported from CMU CL by Martin Atzmueller, with
3288 half a dozen bug fixes in pretty-printing and the debugger, and
3289 half a dozen others elsewhere
3290 * fixed bug 13: Floating point infinities are now supported again.
3291 They might still be a little bit flaky, but thanks to bug reports
3292 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3293 as flaky as they were.
3294 * The --noprogrammer command line option is now supported. (Its
3295 behavior is slightly different in detail from what the old man
3296 page claimed it would do, but it's still appropriate under the
3297 same circumstances that the man page talks about.)
3298 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3299 are now supported, and enabled by default. Thus, the compiler can
3300 handle many floating point and complex operations much less
3301 inefficiently. (Thus e.g. you can implement a complex FFT
3303 * The compiler now detects type mismatches between DECLAIM FTYPE
3304 and DEFUN better, and implements CHECK-TYPE more correctly, and
3305 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3306 thanks to patches from Martin Atzmueller.
3307 * various fixes to make the cross-compiler more portable to
3308 ANSI-conforming-but-different cross-compilation hosts (notably
3309 Lispworks for Windows, following bug reports from Arthur Lemmens)
3310 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3311 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3312 fixed thanks to Pierre Mai's CMU CL patches.
3313 * fixes to keep the system from overflowing internal counters when
3314 it tries to use i/o buffers larger than 16M bytes
3315 * fixed bug 45a: Various internal functions required to support
3316 complex special functions have been merged from CMU CL sources.
3317 (When I was first setting up SBCL, I misunderstood a compile-time
3318 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3319 * improved support for type intersection and union, fixing bug 12
3320 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3321 more obscure bugs as well
3322 * some steps toward byte-compiling non-performance-critical
3323 parts of the system, courtesy of patches from Martin Atzmueller
3324 * Christophe Rhodes has made some debian packages of sbcl at
3325 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3326 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3327 stable, but are nonetheless usable. When he's ready, I'd be happy
3328 to add them to the SourceForge "File Releases" section. (And if
3329 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3330 * new fasl file format version number (because of changes in
3331 internal representation of (OR ..) types to accommodate the new
3332 support for (AND ..) types, among other things)
3334 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3335 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3337 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3339 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3340 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3341 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3342 * DESCRIBE now works on structure objects again.
3343 * Most function call argument type mismatches are now handled as
3344 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3345 know whether the function will be redefined before the call is
3346 executed. (The compiler could flag local calls with full WARNINGs,
3347 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3348 it doesn't keep track of enough information to know whether calls
3349 are local in this sense.)
3350 * Compiler output is now more verbose, with messages truncated
3351 later than before. (There should be some supported way for users
3352 to override the default verbosity, but I haven't decided how to
3353 provide it yet, so this behavior is still controlled by the internal
3354 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3355 src/compiler/ir1util.lisp.)
3356 * Fasl file format version numbers have increased again, because
3357 support for the Gray streams extension changes the layout of the
3358 system's STREAM objects.
3359 * The Gray subclassable streams extension now works, thanks to a
3360 patch from Martin Atzmueller.
3361 * The full LOAD-FOREIGN extension (not just the primitive
3362 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3363 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3364 but like most other programs, it defaults to copying the Unix
3365 environment from the original process instead of starting the
3366 new process in an empty environment.
3367 * Extensions which manipulate the Unix environment now support
3368 an :ENVIRONMENT keyword option which doesn't smash case or
3369 do other bad things. The CMU-CL-style :ENV option is retained
3370 for porting convenience.
3371 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3372 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3374 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3376 * A patch from Martin Atzmueller seems to have solved the SIGINT
3377 problem, and as far as we know, signal-handling now works cleanly.
3378 (If you find any new bugs, please report them!)
3379 * The system no longer defaults Lisp source file names to types
3380 ".l", ".cl", or ".lsp", but only to ".lisp".
3381 * The compiler no longer uses special default file extensions for
3382 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3383 seems to expect a single default extension for all compiled code,
3384 and there's no compelling reason to try to stretch the standard
3385 to allow two different extensions.) Instead, byte-compiled files
3386 default to the same extension as native-compiled files.
3387 * Fasl file format version numbers have increased again, because
3388 a rearrangement of internal implementation packages made some
3389 dumped symbols in old fasl files unreadable in new cores.
3390 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3391 many fewer weird special cases.
3392 * Bug #17 (differing COMPILE-FILE behavior between logical and
3393 physical pathnames) has been fixed, and some related misbehavior too,
3394 thanks to a patch from Martin Atzmueller.
3395 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3396 by Tim Moore, ported to SBCL by Martin Atzmueller.
3397 * Martin Atzmueller fixed several filesystem-related problems,
3398 including bug #36, in part by porting CMU CL patches, which were
3399 written in part by Paul Werkowski.
3400 * More compiler warnings in src/runtime/ are gone, thanks to
3401 more patches from Martin Atzmueller.
3402 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3405 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3407 * DESCRIBE now works on CONDITION objects.
3408 * The debugger now handles errors which arise when trying to print
3409 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3411 * The build system now uses an additional file, customize-target-features.lisp,
3412 to allow local modifications to the target *FEATURES* list. (The point of
3413 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3414 debugging features enabled, without having to worry about propagating it
3415 into everyone's system when I do a "cvs update".) When no
3416 customize-target-features.lisp file exists, the target *FEATURES* list
3417 should be constructed the same way as before.
3418 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3419 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3420 warn about behavior which is undefined under the ANSI spec. Note: This
3421 is specified by ANSI, but it's not very popular with programmers.
3422 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3423 macro in the SBCL sources for an example of a workaround which you
3424 might use to make portable ANSI-standard code which does what you want.
3425 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3426 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3427 when it's used as a non-toplevel form. (This is required in order
3428 to implement the DEFCONSTANT-EQX macro.)
3429 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3430 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3431 funny things, due to bugs in the implementation of EVAL-WHEN
3432 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3433 non-toplevel DEFCONSTANTs are uncommon.)
3434 * The core file version number and fasl file version number have been
3435 incremented, because the old noncompliant DEFCONSTANT behavior involved
3436 calling functions which no longer exist, and because I also took the
3437 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3438 * fixed bug 1 (error handling before read-eval-print loop starts), and
3439 redid debugger restarts and related debugger commands somewhat while
3441 ** The QUIT debugger command is gone, since it did something
3442 rather different than the SB-EXT:QUIT command, and since it never
3443 worked properly outside the main toplevel read/eval/print loop.
3444 Invoking the new TOPLEVEL restart provides the same functionality.
3445 ** The GO debugger command is also gone, since you can just invoke
3446 the CONTINUE restart directly instead.
3447 ** The TOP debugger command is also gone, since it's redundant with the
3448 FRAME 0 command, and since it interfered with abbreviations for the
3450 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3452 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3453 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3454 some porting work by Martin Atzmueller)
3455 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3456 with a boatload of patches which he ported to SBCL. Now that those
3458 ** The system tries to make sure that its low-priority messages
3459 are prefixed by semicolons, to help people who like to use
3460 syntax highlighting in their ILISP buffer. (This patch
3461 was originally due to Raymond Toy.)
3462 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3463 patches originally written by Raymond Toy.
3464 ** The compiler understands coercion between single-value and
3465 multiple-VALUES type expressions better, getting rid of some very
3466 weird behavior, thanks to patches originally by Robert MacLachlan
3467 and Douglas Crosher.
3468 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3469 lambda lists, thanks to a patch originally by Pierre Mai.
3470 ** The system no longer bogusly warns about "abbreviated type
3472 ** The compiler gets less confused by inlining and RETURN-FROM,
3473 thanks to some patches originally by Tim Moore.
3474 ** The system no longer hangs when dumping circular lists to fasl
3475 files, thanks to a patch originally from Douglas Crosher.
3476 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3477 undefined function error.
3478 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3479 in structure slot initforms) for now, documented workaround instead:-|
3480 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3481 * fixed bug 5: added stubs for various Gray stream functions called
3482 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3483 installed, at least appropriate type errors are generated
3484 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3485 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3486 consistently in DEFMETHOD forms.
3487 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3488 it doesn't seem to affect SBCL after all
3489 * The C runtime system now builds with better optimization and many
3490 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3492 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3494 * The system is now under CVS at SourceForge (instead of the
3495 CVS repository on my home machine).
3496 * The new signal handling code has been tweaked to treat register
3497 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3498 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3499 downstream, e.g. in debug-int.lisp, has implicit dependencies
3500 on the unsignedness of integer representation of machine words,
3501 and that caused the system to bomb out with infinite regress
3502 when trying to recover from type errors involving signed values,
3503 e.g. (BUTLAST '(1 2 3) -1).)
3504 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3505 (This was one of the bugs Peter Van Eynde reported back in July.)
3506 * The system now uses code inspired by Colin Walters' O(N)
3507 implementation of MAP (from the cmucl-imp@cons.org mailing
3508 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3509 inline the MAP operation, and there is more than one
3510 sequence argument to the MAP call (so that it can't just
3511 do ETYPECASE once and for all based on the type of the
3512 single sequence argument). (The old non-inline implementation
3513 of the general M-argument sequence-of-length-N case required
3514 O(M*N*N) time when any of the sequence arguments were LISTs.)
3515 * The QUIT :UNIX-CODE keyword argument has been renamed to
3516 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3517 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3519 * Raymond Wiker's patches to port dynamic loading from Linux to
3520 FreeBSD have been added.
3521 * The BUGS file is now more nearly up to date, thanks in large part
3522 to Martin Atzmueller's review of it.
3523 * The debugger now flushes standard output streams before it begins
3524 its output ("debugger invoked" and so forth).
3525 * The core version number and fasl file version number have both
3526 been incremented, because of incompatible changes in the layout
3528 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3529 used to not be called for a saved Lisp image.)
3530 * Martin Atzmueller's version of a patch to fix a compiler crash,
3531 as posted on sbcl-devel 13 September 2000, has been installed.
3532 * Instead of installing Martin Atzmueller's patch for the
3533 compiler transform for SUBSEQ, I deleted the compiler transform,
3534 and transforms for some similar consing operations.
3535 * A bug in signal handling which kept TRACE from working on OpenBSD
3537 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3538 away by constant folding
3539 * The system now defines its address space constants in one place
3540 (in the Lisp sources), and propagates them automatically elsewhere
3541 (through GENESIS and the sbcl.h file). Therefore, patching the
3542 address map is less unnecessarily tedious and error-prone. The
3543 Lisp names of address space constants have also been systematized.
3544 * CVS tags like dollar-Header-dollar have been removed from
3545 the sources, because they have never saved me trouble and
3546 they've been source of trouble working with patches and other
3547 diff-related operations.
3548 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3549 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3551 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3553 * The system has been ported to OpenBSD.
3554 * The system now compiles with a simple "sh make.sh" on the systems
3555 that it's supported on. I.e., now you no longer need to tweak
3556 text in the target-features.lisp-expr and symlinks in src/runtime/
3557 by hand, the make.sh takes care of it for you.
3558 * The system is no longer so grossly inefficient when compiling code
3559 involving vectors implemented as general (not simple) vectors (VECTOR T),
3560 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3561 which dares to use the various sequence functions on non-simple
3562 vectors, takes less of a performance hit.
3563 * There is now a primitive type predicate VECTOR-T-P
3564 to test for the (VECTOR T) type, so that e.g.
3565 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3566 can now be compiled with some semblance of efficiency. (The old code
3567 turned the type declaration into a full call to %TYPEP at runtime!)
3568 * AREF on (VECTOR T) is still not fast, since it's still compiled
3569 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3570 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3571 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3572 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3573 without making full calls to SUBTYPEP at runtime.
3574 (Some analogous efficiency issues for non-simple vectors specialized to
3575 element types other than T, or for non-simple multidimensional arrays,
3576 have not been addressed. They could almost certainly be handled the
3577 same way if anyone is motivated to do so.)
3578 * The changes in array handling break binary compatibility, so
3579 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3580 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3581 instead of (VALUES T T).
3582 * By following the instructions that Dan Barlow posted to sbcl-devel
3583 on 2 July 2000, I was able to enable primitive dynamic object
3584 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3585 functionality is not implemented (since it calls ld to resolve
3586 library references automatically, requiring RUN-PROGRAM for its
3587 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3588 try to resolve library references) is now supported.
3589 * The system now flushes the standard output streams when it terminates,
3590 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3591 them at several other probably-convenient times, e.g. in each pass of
3592 the toplevel read-eval-print loop, and after evaluating a form given
3593 as an "--eval" command-line option. (These changes were motivated by a
3594 discussion of stream flushing issues on cmucl-imp in August 2000.)
3595 * The source transform for TYPEP of array types no longer assumes
3596 that an array whose element type is a not-yet-defined type
3597 is implemented as an array of T, but instead punts, so that the
3598 type will be interpreted at runtime.
3599 * There is now some support for cross-compiling in make.sh: each of
3600 the phases of make.sh has its own script. (This should be transparent
3601 to people doing ordinary, non-cross-compile builds.)
3602 * Since my laptop doesn't have hundreds of megabytes of memory like
3603 my desktop machine, I became more motivated to do some items on
3604 my to-do list in order to reduce the size of the system a little:
3605 ** Arrange for various needed-only-at-cold-init things to be
3606 uninterned after cold init. To support this, those things have
3607 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3608 symbols with such names are now uninterned after cold init).
3609 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3610 fasl files for cold load.
3611 ** Remove the old compiler structure pooling code (which used to
3612 be conditional on the target feature :SB-ALLOC) completely.
3613 ** Redo the representation of some data in cold init to be more compact.
3614 (I also looked into supporting byte compiled code at bootstrap time,
3615 which would probably reduce the size of the system a lot, but that
3616 looked too complicated, so I punted for now.)
3617 * The maximum signal nesting depth in the src/runtime/ support code has
3618 been reduced from 4096 to 256. (I don't know any reason for the very
3619 large old value. If the new smaller value turns out to break something,
3620 I'll probably just bump it back up.)
3621 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3623 * Many, many bugs reported by Peter Van Eynde have been added to
3624 the BUGS list; some have even been fixed.
3625 * While enabling dynamic object file loading, I tried to make the
3626 code easier to understand, renaming various functions and variables
3627 with less ambiguous names, and changing some function calling
3628 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3629 * While trying to figure out how to do the OpenBSD port, I tried to
3630 clean up some of the code in src/runtime/. In particular, I dropped
3631 support for non-POSIX signal handling, added various comments,
3632 tweaked the code to reduce the number of compilation warnings, and
3633 renamed some files to increase consistency.
3634 * To support the new automatic configuration functionality in make.sh,
3635 the source file target-features.lisp-expr has been replaced with the
3636 source file base-target-features.lisp-expr and the machine-generated
3637 file local-target-features.lisp-expr.
3638 * fixed a stupid quoting error in make.sh so that using CMU CL
3639 "lisp -batch" as cross-compilation host works again
3641 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3643 * DESCRIBE no longer tries to call itself recursively to describe
3644 bound/fbound values, so that it no longer fails on symbols which are
3645 bound to themselves (like keywords, T, and NIL).
3646 * DESCRIBE now works on generic functions.
3647 * The printer now prints less-screwed-up representations of closures
3648 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3649 * A private symbol is used instead of the :EMPTY keyword previously
3650 used to mark empty slots in hash tables. Thus
3651 (DEFVAR *HT* (MAKE-HASH-TABLE))
3652 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3653 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3654 now does what ANSI says that it should. (You can still get
3655 similar noncompliant behavior if bang on the hash table
3656 implementation with all the symbols you get back from
3657 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3658 This breaks binary compatibility, since tests for equality to
3659 :EMPTY are wired into things like the macroexpansion of
3660 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3662 * There's now a minimal placeholder implementation for CL:STEP,
3663 as required by ANSI.
3664 * An obscure bug in the interaction of the normal compiler, the byte
3665 compiler, inlining, and structure predicates has been patched
3666 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3667 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3669 * Missing ordinary arguments in a macro call are now detected even
3670 when the macro lambda list contains &KEY or &REST.
3671 * The debugger no longer complains about encountering the top of the
3672 stack when you type "FRAME 0" to explicitly instruct it to go to
3673 the top of the stack. And it now prints the frame you request even
3674 if it's the current frame (instead of saying "You are here.").
3675 * As specified by ANSI, the system now always prints keywords
3676 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3678 * The default initial SIZE of HASH-TABLEs is now smaller.
3679 * Type information from CLOS class dispatch is now propagated
3680 into DEFMETHOD bodies, so that e.g.
3681 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3683 is now basically equivalent to
3684 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3685 (DECLARE (TYPE SINGLE-FLOAT X))
3687 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3688 operation, without having to do run-time type dispatch.
3689 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3690 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3691 * Since I don't seem to be making big file reorganizations very often
3692 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3693 is overflowing my ability to conveniently back them up), I've finally
3694 checked the system into CVS. (The CVS repository is on my home system,
3695 not at SourceForge -- putting it on SourceForge might come later.)
3696 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3697 high-level GC-NOTIFY-FOO functions send their output. (There's
3698 still very little control of where low-level verbose GC functions
3699 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3700 less than it used to -- the GC-NOTIFY-FOO functions are now under
3701 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3702 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3703 in only one place in the source code, and propagates it automatically
3704 everywhere that it's needed. Thus e.g. when I bump the version from
3705 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3706 * The C source files now include boilerplate legalese and documentation
3707 at the head of each file (just as the Lisp source files already did).
3708 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3709 to his page will be replaced with a link to his new CLiki service.
3711 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3713 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3714 * The build process now looks for GNU make under the default name "gmake",
3715 instead of "make" as it used to. If GNU make is not available as "gmake"
3716 on your system, you can change this default behavior by setting the
3717 GNUMAKE environment variable.
3718 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3719 can build without error under CMU CL.
3721 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3723 * There is now a partial SBCL user manual (with some new text and some
3724 text cribbed from the CMU CL manual).
3725 * The beginnings of a profiler have been added (starting with the
3726 CMU CL profiler and simplifying and cleaning up). Eventually the
3727 main interface should be through the TRACE macro, but for now,
3728 it's still accessed through vaguely CMU-CL-style functions and macros
3729 exported from the package SB-PROFILE.
3730 * Some problems left over from porting CMU CL to the new
3731 cross-compilation bootstrap process have been cleaned up:
3732 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3733 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3734 part when getting rid of the compiler *BACKEND* stuff.)
3735 ** The value of *NULL-TYPE* was screwed up, because it was
3736 being initialized before the type system knew the final
3737 definition of the 'NULL type. This screwed up several key
3738 optimizations in the compiler, causing inefficiency in all sorts
3739 of places. (I found it because I wanted to understand why
3740 GET-INTERNAL-RUN-TIME was consing.)
3741 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3742 PROCLAIM FTYPE information. Unfortunately this broke binary
3743 compatibility again, since now the forms output by DEFGENERIC
3744 to refer to functions which didn't exist in 0.6.3.
3745 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3746 can use the new (as of 0.6.3) transform for SOME into MAP into
3748 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3749 values to (INTEGER 0 1000000), so that the time code will no longer
3750 occasionally get blown up by Linux returning 1000000 microseconds
3751 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3752 its output conform to the ANSI spec. (Alas, this makes its output
3753 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3754 * A full call to MAP NIL with a single sequence argument no longer conses.
3755 * fixes to problems pointed out by Martin Atzmueller:
3756 * The manual page no longer talks about multiprocessing as though
3757 it were currently supported.
3758 * The ILISP support patches have been removed from the distribution,
3759 because as of version 5.10.1, ILISP now supports SBCL without us
3760 having to maintain patches.
3761 * added a modified version of Raymond Toy's recent CMU CL patch for
3762 EQUALP comparison of HASH-TABLE
3764 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3766 * The system still can't cross-compile itself with
3767 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3768 entails), but at least it can compile more of itself that way
3769 than it used to be able to, and various buglets which were uncovered
3770 by trying to cross-compile itself that way have now been fixed.
3771 * This release breaks binary compatibility again. This time
3772 at least I've incremented the FASL file format version to 2, so that the
3773 problem can be detected reliably instead of just causing weird errors.
3774 * various new style warnings:
3775 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3776 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3777 ** using the lexical binding of a variable named in the *FOO* style
3778 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3779 as specified by ANSI.
3780 * *RANDOM-STATE* is no longer automatically initialized from
3781 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3782 default behavior of the system is to repeat its behavior every time
3783 it's run. If you'd like to change this behavior, you can always
3784 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3785 old convention there was no comparably easy way to get the system to
3786 repeat its behavior every time it was run.
3787 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3788 FTYPE declarations has been removed, in favor of their ANSI
3789 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3790 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3791 the types of their sequence arguments can be determined at compile time.
3792 This is done through a new open code expansion for MAP which eliminates
3793 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3794 argument types can be determined at compile time.
3795 * The optimizer now transforms COERCE into an identity operation when it
3796 can prove that the coerced object is already of the correct type. (This
3797 can be a win for machine generated code, including the output of other
3798 optimization transforms, such as the MAP transform above.)
3799 * Credit information has been moved from source file headers into CREDITS.
3800 * Source file headers have been made more standard.
3801 * The CASE macro now compiles without complaining even when it has
3804 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3806 * (Note that the way that the PCL macroexpansions were rewritten
3807 to accommodate the change in DEFGENERIC below breaks binary
3808 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3809 not run under sbcl-0.6.2. Once we get out of alpha releases,
3810 i.e. hit release 1.0.0, we'll probably try to maintain binary
3811 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3812 and sbcl-1.4.4. Until then, however, it might be fairly common
3813 for maintenance releases to break binary compatibility.)
3814 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3816 * The reporting of unhandled signals has been changed to print some
3817 explanatory text as well as the report form. (Previously only
3818 the report form was printed.)
3819 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3820 it defines, so that the compiler no longer issues undefined function
3821 warnings for compiled-but-not-yet-loaded generic functions.
3822 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3823 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3824 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3825 section of the ANSI spec are supported.
3826 * The "" nickname for the "KEYWORD" package has been removed.
3827 The reader still handles symbol tokens which begin with a package marker
3828 as keywords, but it doesn't expose its mechanism for doing so in the
3829 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3830 * The system now issues STYLE-WARNINGs for contradictory TYPE
3831 proclamations. (Warnings for contradictory FTYPE proclamations would
3832 be nice too, but those can't be done usefully unless the type system
3833 is made smarter about FUNCTION types.)
3834 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3835 systematized, so that "*target-*.lisp is supposed to exist only on the
3836 target and imply that there's a related file which exists on the
3837 host, and *host-*.lisp is supposed to exist only on the host and imply
3838 that there's a related file which exists on the target. This involves a
3839 lot of renaming. Hopefully the acute confusion caused by the renaming
3840 will be justified by the reduction in chronic confusion..
3841 ** runtime-type.lisp -> early-target-type.lisp
3842 ** target-type.lisp -> late-target-type.lisp
3843 ** early-host-format.lisp -> early-format.lisp
3844 ** late-host-format.lisp -> late-format.lisp
3845 ** host-error.lisp -> misc-error.lisp
3846 ** early-error.lisp -> early-target-error.lisp
3847 ** late-error.lisp -> late-target-error.lisp
3848 ** host-defboot.lisp -> early-defboot.lisp
3849 ** code/misc.lisp -> code/target-misc.lisp
3850 ** code/host-misc.lisp -> code/misc.lisp
3851 ** code/numbers.lisp -> code/target-numbers.lisp
3852 ** code/early-numbers.lisp -> numbers.lisp
3853 ** early-host-type.lisp -> early-type.lisp
3854 ** late-host-type.lisp -> late-type.lisp
3855 ** host-typep.lisp -> typep.lisp
3856 ** load.lisp -> target-load.lisp
3857 ** host-load.lisp -> load.lisp
3858 ** host-disassem.lisp -> disassem.lisp
3859 ** host-insts.lisp -> insts.lisp
3860 ** byte-comp.lisp -> target-byte-comp.lisp
3861 ** host-byte-comp.lisp -> byte-comp.lisp
3862 ** host-signal.lisp -> signal.lisp
3863 ** host-defstruct.lisp -> defstruct.lisp
3864 ** late-target-type.lisp -> deftypes-for-target.lisp
3865 Furthermore, several other previously target-only files foo.lisp (e.g.
3866 hash-table.lisp and random.lisp) have been split into a target-and-host
3867 foo.lisp file and a target-only target-foo.lisp file, with their key type
3868 definitions in the target-and-host part, so that the cross-compiler will
3869 know more about target types.
3870 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3871 gone away. In their place are various *BACKEND-FOO* variables
3872 corresponding to the slots of the old structure.
3873 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3874 into the target SBCL has been fixed.
3875 * The chill.lisp system for loading cold code into a running SBCL
3877 * Support for the CMU CL "scavenger hook" extension has been removed.
3878 (It was undocumented and unused in the CMU CL sources that SBCL was
3879 derived from, and stale in sbcl-0.6.1.)
3880 * Various errors in the cross-compiler type system were detected
3881 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3882 (enabling various consistency checks). Many of them were fixed,
3883 but some hard problems remain, so the compiler is back to
3884 running without *TYPE-SYSTEM-INITIALIZED* for now.
3885 * As part of the cross-compiler type system cleanup, I implemented
3886 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3887 * I have started adding UNCROSS calls throughout the type system
3888 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3889 on cross-compiler input files will be able to go away, and various
3891 * CONSTANTP now returns true for quoted forms (as explicitly required
3894 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3896 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3897 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3898 (MAKE-LIST -1). In the longer term, it ought to become true
3899 that declarations are assertions even at SAFETY 1. For now, it's not
3900 quite true even at SAFETY 3, but it's at least more nearly true..
3901 (Note that this change seems to increases the size of the system by
3902 O(5%) and to decrease the speed of the compiler by 20% or more.)
3903 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3904 for the problem of printing dozens of lines of distracting information
3905 about low-level system machinery as part of the top stack frame
3906 on entry to the debugger when an undefined function was called.
3907 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3908 is not reset to COMMON-LISP-USER.
3909 * Compilation of stuff related to dyncount.lisp has been made conditional
3910 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3911 smaller. The various dyncount-related symbols have been moved into
3912 a new "SB-DYNCOUNT" package.
3913 * tty-inspect.lisp has been renamed to inspect.lisp.
3914 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3915 feature has gone away. (When we eventually port to other flavors of
3916 libc and/or Unix, we'll try to make the differences between flavors
3917 invisible at the user level.)
3918 * Various other *FEATURES* tags, and/or their associated conditionals,
3919 have been removed if obsolescent, or given better documentation, or
3920 sometimes given more-mnemonic names.
3922 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3924 * tidied up "make.sh" script
3925 * tidied up system directory structure
3926 * better "clean.sh" behavior
3927 * added doc/FOR-CMUCL-DEVELOPERS
3928 * many many small tweaks to output format, e.g. removing possibly-confusing
3929 trailing #\. character in DESCRIBE-INSTANCE
3930 * (EQUALP #\A 'A) no longer signals an error.
3931 * new hashing code, including EQUALP hashing
3932 * tidied up Lisp initialization and toplevel
3933 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3934 * command line argument processing
3935 * added POSIX-GETENV function to deal with Unix-ish environment variables
3936 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3937 terminating SBCL on EOF
3938 * non-verbose GC by default
3939 * There is no more "sbcl" shell script; the sbcl file is now the C
3940 runtime executable (just like CMU CL).
3941 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3943 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3944 symbol and package information as Lisp native symbol and package objects
3945 instead of strings naming symbols and strings naming packages. This way,
3946 whenever packages are renamed (as in warm init), debug information is
3947 transformed along with everything else.
3948 * tweaked the optimization policy declarations which control the building
3949 of SBCL itself. Now, among other things, the system no longer saves
3950 source location debugging information. (This helps two problems at once
3951 by reducing SBCL size and by keeping SBCL from trying to look for its
3952 sources -- which may not exist -- when reporting errors.)
3953 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3954 debugging and testing purposes
3955 * cleaned up printing, making the printer call PRINT-OBJECT for
3956 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3957 methods, giving nearly-ANSI behavior
3958 * converted almost all special variables to use *FOO* naming convention
3959 * deleted PARSE-TIME functionality, since it can be done portably
3960 * moved some files out of cold init into warm init
3961 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3963 * regularized formatting of source files
3964 * added an install.sh script
3965 * fixed ridiculous memory usage of cross-compiler by making
3966 compiler/alloc.lisp not try to do pooling unless it can hook
3967 itself into the GC of the cross-compilation host. Now the system
3968 builds nicely on my old laptop.
3969 * added :SB-ALLOC in target-features.lisp-expr
3970 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3971 was not implemented)
3972 * re-did condition handling and note reporting in the compiler. Notes
3973 are no longer handled by signalling conditions. Style warnings
3974 and warnings are handled more correctly and reported in such a way
3975 that it's easy to find one or the other in your output (so that you
3976 can e.g. figure out which of many problems caused COMPILE-FILE to
3978 * changed the severity of several compiler warnings from full WARNING
3979 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3980 compiler note reporting so that it doesn't use the condition system
3981 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3982 COMPILE-FILE command)
3983 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3984 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3985 no longer supported, since their implementation was deeply intertwingled
3986 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3987 * removed bogus "support" for compiler macros named (SETF FOO), and
3988 removed the compiler macro for SETF INFO (but only after making a fool
3989 of myself on the cmucl-imp mailing list by posting a bogus patch for
3990 DEFINE-COMPILER-MACRO..)
3991 * Compiled files containing forms which have side effects on the Lisp
3992 reader (such as DEFPACKAGE forms) are now handled more correctly.
3993 (Compiler queuing of top level lambdas has been suppressed by setting
3994 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3995 * deleted various currently-unused source files, e.g. gengc.lisp. They
3996 may be added back at some point e.g. when porting to other architectures,
3997 but until they are it's distracting to distribute them and to try to
3999 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4000 there were so many of them they're just distractions, and UNCROSS is
4001 known to be able to handle the current sources
4002 * moved PROFILE functionality into TRACE, so that it will be clear
4003 how the wrapping and unwrapping of functions when you profile them
4004 interacts with the wrapping and unwrapping of functions when you
4005 trace them. (Actually, the functionality isn't there yet, but at least
4006 the interface specification is there. Hopefully, the functionality will
4007 arrive with some future maintenance release.)
4008 * removed host-oops.lisp
4009 * changed signature of QUIT function to allow UNIX-CODE argument
4010 * fixed READ-SEQUENCE bug
4011 * tweaked verbose GC output so that it looks more like the progress
4012 output that ANSI specifies for functions like LOAD
4013 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4014 * added <http://sbcl.sourceforge.com> to the banner information printed by