1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.19 (1.0.0?) relative to sbcl-0.9.18:
3 * improvement: floating point modes in effect are now saved in
4 core, and restored on startup.
5 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
6 startup, not time since first call to GET-INTERNAL-REAL-TIME.
7 * bug fix: compiler bug triggered by a (non-standard) VALUES
8 declaration in a LET* was fixed. (reported by Kaersten Poeck)
9 * bug fix: file compiler no longer confuses validated and already
10 dumped structurres. (reported by Kaersten Poeck)
11 * improvements to the Windows port:
12 ** floating point exceptions are now reported correctly.
13 ** stack exhaustion detection works partially.
14 ** more accurate GET-INTERNAL-REAL-TIME.
15 ** better RUN-PROGRAM behaviour in case of errors in the child
17 ** PROBE-FILE now simplifies pathnames correctly.
18 ** DIRECTORY now works correctly with :WILD-INFERIORS.
19 ** (DECODE-UNIVERSAL-TIME 0) works.
21 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
22 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
23 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
25 * bug fix: two potential GC deadlocks affecting threaded builds.
26 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
27 returns T (reported by Anton Kazennikov)
28 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
29 with non-variable places
30 * bug fix: remove a race condition in the setting of
31 funcallable-instance functions, this should make threaded CLOS
32 code more stable against memory faults.
33 * bug fix: corruption of specials when unbinding is interrupted by an
34 asynchronous unwind (reported by Hannu Koivisto)
35 * improvement: the debugger will now also display local variables that
36 are only used once, for code compiled with a DEBUG optimization quality
39 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
40 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
41 * incompatible change: External-format support for FFI calls. The
42 SB-ALIEN:C-STRING no longer implies an ASCII
43 external-format. Instead the string is subject to external-format
44 conversion. Additionally return values of type SB-ALIEN:C-STRING
45 are by default of type (SIMPLE-ARRAY CHARACTER), not
46 SIMPLE-BASE-STRING. For an alien type that behaves like the old
47 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
48 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
49 * incompatible change: SB-EXT package no longer contains the
50 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
51 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
53 * incompatible change: the single-stepper is no longer available
54 on Alpha, Mips and Sparc platforms.
55 * minor incompatible change: the direct superclasses of
56 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
57 not (STANDARD-OBJECT FUNCTION). This makes the
58 class-precedence-lists of GENERIC-FUNCTION and
59 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
61 * minor incompatible change: the default stream external format on
62 non-unicode SBCL is no longer detected from the locale, but is always
64 * new feature: Add a version of evaluator that uses an interpreter instead
65 of the compiler. EVAL still uses the compiler by default, to switch it
66 to use the interpreter, set the value of the variable
67 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
68 * minor incompatible change: the single-stepper REPL has been merged
69 with the normal debugger (see the "Stepping" heading of the debugger help
70 for more details). The debugger command STEP will no longer switch
71 to the single-stepper REPL.
72 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
74 * bug fix: loading of default sysinit file works. (thanks to Leonid
76 * bug fix: better detection of circularities in the file-compiler.
77 (reported by Marco Monteiro)
78 * bug fix: the CL pathname functions now work with files that have
79 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
80 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
81 systems with Unix98 pty semantics.
82 * bug fix: ASDF-INSTALL will now work with bsd tar.
83 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
85 * bug fix: timers expiring in dead threads no longer cause a
86 type-error (reported by Paul "Nonny Mouse").
87 * bug fix: thanks to more lightweight single-stepper instrumentation,
88 code compiled with (DEBUG 3) will compile and execute significantly faster,
89 and will have more accurate type-inferencing than before
90 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
91 whose bindings are modified
92 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
93 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
94 ** replace dummy implementations of CL:MACHINE-INSTANCE and
95 CL:SOFTWARE-VERSION with proper version
97 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
98 * feature: implemented the READER-METHOD-CLASS and
99 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
100 as specified by AMOP.
101 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
103 * optimization: faster LOGCOUNT implementation on x86 and x86-64
104 (thanks to Lutz Euler)
105 * optimization: hashing of general arrays and vectors has been
106 improved. (reported by Any Fingerhut)
107 * enhancement: SB-INTROSPECT is now able to find definitions of
108 profiled functions. (thanks to Troels Henriksen)
109 * enhancement: compiler-macro expansion applies now to FUNCALL forms
110 as well, allowing compiler-macros for SETF-functions to expand.
111 * enhancement: step-instrumentation no longer wraps known
112 single-value functions with multiple-value context, allowing
113 better type inference.
114 * fixed bug #337: use of MAKE-METHOD in method combination now works
115 even in the presence of user-defined method classes. (reported by
116 Bruno Haible and Pascal Costanza)
117 * fixed bug #339(c): if there are applicable methods not part of any
118 long-form method-combination group, call INVALID-METHOD-ERROR.
119 (reported by Bruno Haible)
120 * fixed bug #361: the :FUNCTION initarg in the protocol for
121 initialization of methods can now be used to override
122 internally-produced optimized functions. (reported by Bruno
124 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
125 system-provided lambda expression no longer cause warnings about
126 unbound #:|pv-table| symbols.
127 * bug fix: improved the handling of type declarations and the
128 detection of violations for keyword arguments with non-constant
130 * bug fix: potentially erronous calls to PATHNAME and
131 MERGE-PATHNAMES were being flushed in some cases.
132 (reported by Richard Kreuter)
133 * bug fix: compiled calls to TYPEP were mishandling obsolete
134 instances. (reported by James Bielman and Attila Lendvai)
135 * bug fix: format strings with ~> without matching ~< no longer
136 trigger an AVER, but signal an understandable error instead.
137 (reported by Antonio Martinez)
138 * bug fix: specifying an output-file without a pathname-type for
139 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
140 of input-file instead of "fasl". (reported by Robert Dodier)
141 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
142 (reported by James Y Knight).
143 * bug fix: compiler-macros-function did not consider the environment
144 argument for shadowing by local functions.
145 * bug fix: compiler-macros expansion was inhibited by local INLINE
147 * bug fix: inline expansions of known functions were subject to
148 step-instrumentation in high debug policies, leading to problems
150 * bug fix: compiler failed to differentiate between different CONS
152 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
153 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
154 * bug fix: #368: incorrect use of expressed vs. upgraded array
156 * bug fix: #306a: more precise unions of array types.
157 * thread-safety improvements:
158 ** CONDITION-WAIT could return early on Linux, if the thread was
159 interrupted and subsequently continued with SIGCONT.
160 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
162 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
163 * added support for the ucs-2 external format. (contributed by Ivan
165 * minor incompatible change: pretty printing of objects of type
166 (cons symbol) is, in the default pprint-dispatch-table, now
167 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
169 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
170 called later in a class's lifetime, possibly as late as when the
171 first instance of the class is created. Previously,
172 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
173 class became finalizeable.
174 * fixed bug: FILE-POSITION sometimes returned inconsistent results
175 for multibyte external-format streams. (thanks to "vbzoli")
176 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
177 with :ALLOCATION :CLASS inherited from superclasses of the
179 * fixed bug: anonymous classes can now be created using the :NAME
180 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
181 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
182 * fixed bug: core-files saved with :EXECUTABLE T can again be
183 executed when SBCL_HOME isn't set. (reported by James Knight)
184 * fixed bug: toplevel LOCALLY forms with declarations could
185 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
186 * fixed bug: printing from several different threads using different
187 values of *print-case* could cause invalid output, due to
188 some internal special variables of the printer not being bound
189 thread-locally (reported by Max Mikhanosha)
190 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
191 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
192 (reported by Pascal Costanza)
193 * fixed bug: FINALIZE-INHERITANCE is called from
194 REINITIALIZE-INSTANCE on classes when the class has previously
195 been finalized, as required by AMOP.
196 * minor code generation optimizations:
197 ** better register allocation in CLOS dispatching functions
198 ** overflow detection when coercing signed bytes to fixnums on x86-64
199 ** is now implemented with one IMUL instruction instead of three shifts
200 ** more efficient bit-vector access on x86 and x86-64
201 ** more efficient access to raw structure slots on x86 and x86-64
202 * fixed some bugs revealed by Paul Dietz' test suite:
203 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
206 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
207 * feature: thread support on Solaris/x86, and experimental thread support
209 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
211 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
212 default method for SLOT-UNBOUND.
213 * minor incompatible change: prevent the user from specializing the
214 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
215 somewhat counter to the intent of the protocol, I (CSR) think, and
216 additionally it just doesn't work in SBCL as currently
217 implemented, thanks to optimizations (that are always valid for
218 the other three SLOT-VALUEish functions, but not for the setter).
219 * minor incompatible change: the :SB-LDB feature is now enabled by
220 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
221 the low-level debugger.
222 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
223 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
224 down to an SB-SYS:FD-STREAM, allowing direct communication in
225 more cases. Temporary files and pipes are still used as a fallback
227 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
228 on the non-reentrant C functions localtime(3) and gmtime(3).
229 * bug fix: global optimization policy was not visible in LOCALLY and
231 * bug fix: class objects can be used as specializers in methods.
232 (reported by Pascal Costanza)
233 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
234 with a type error. (reported by blitz_ on #lisp)
235 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
236 failed with a type error. (reported by Pascal Bourguignon)
237 * bug fix: merging pathnames against defaults with :DIRECTORY
238 starting with '(:RELATIVE :BACK) should preserve the :BACK.
239 (reported by James Y Knight)
240 * bug fix: saving large (>2GB) cores on x86-64 now works
241 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
242 constant index and a (simple-array (signed-byte 32)) array
243 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
244 error (patch by Robert J. Macomber)
245 * feature: TIME now displays the amount of run-time spent in GC
246 * bug fix: The debugger now does a better job of respecting
247 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
248 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
249 workaround for bug 403.)
250 * bug fix: floating point exception handling now works on FreeBSD
251 (thanks to NIIMI Satoshi)
252 * fixed some bugs revealed by Paul Dietz' test suite:
253 ** MISC.641: LET-conversion were not supposed to work in late
255 * improvements to the Win32/x86 port:
256 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
257 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
259 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
260 * new feature: source path information is generated for macro-expansion
261 errors for use in IDE's like Slime (thanks to Helmut Eller)
262 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
263 * bug fix: compiler does not loop forever on an invalid type in
265 * improvement: compilation of most CLOS applications is significantly
267 * optimization: added a limited bytecode compiler for simple toplevel
268 forms, speeding up compilation and FASL loading
269 * bug fix: the statistical profiler now properly distinguishes anonymous
272 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
273 * minor incompatible change: in sbcl-0.9.11 (but not earlier
274 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
275 into subclasses of the finalized class. Now user calls to
276 FINALIZE-INHERITANCE finalize just the one class, and calls by the
277 system return before any subclasses are finalized.
278 * minor incompatible change: The reader no longer ignores errors
279 regarding non-existent packages in #+ and #- feature tests.
280 * new feature: command line options --no-sysinit, --no-userinit to
281 inhibit loading the corresponding init files
282 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
283 for sending data through UDP sockets (thanks to François-René Rideau)
284 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
285 error is signalled instead (thanks to François-René Rideau)
286 * minor incompatible change: Error signalling behaviour of lexical
287 operations violating package locks has changed slightly. Refer to
288 documentation on package locks for details.
289 * enhancement: EVAL can process IF-expressions without resorting to the
291 * bug fix: Some operations involving SIN, COS, or TAN weren't being
292 constant-folded properly due to missing out-of-line functions.
293 (reported by Mika Pihlajamäki)
294 * bug fix: LISTEN sometimes returned T even in cases where no data was
295 immediately available from the stream
296 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
297 were reversed. (reported by Levente Mészáros)
298 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
299 when given a vector displaced to an adjustable array. (reported
301 * bug fix: derivation of float boundaries from numbers outside the
302 appropriate float range (reported by John Wiseman)
303 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
305 * bug fix: precomputing cache entries for generic functions with
306 some subclasses of specializers as yet invalid does not attempt to
307 fill a cache line with a negative offset. (reported by Levente
309 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
310 allow condition class objects as arguments to DOCUMENTATION and
311 (SETF DOCUMENTATION); only find and set documentation for
312 structure names for the STRUCTURE doc-type. (suggested by Gary
314 * improvements to the Win32/x86 port:
315 ** (user-homedir-pathname) and default initialization file
316 locations now know about the user's "Documents and Settings"
317 directory (thanks to Yaroslav Kavenchuk)
318 ** run-program is implemented (thanks to Mike Thomas)
319 ** sockets support (thanks to Timothy Ritchey)
320 ** better backtrace support (thanks to Alastair Bridgewater)
321 ** sb-grovel supported
322 ** asdf-install and sb-posix work somewhat
323 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
324 * improvements to the Solaris/x86 port:
325 ** works on Solaris 11/Solaris Express
326 ** floating-point exception handling support
327 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
328 * fixed some bugs revealed by Paul Dietz' test suite:
329 ** REMOVE-METHOD returns its generic function argument even when
330 the method is not one of the generic functions' methods.
331 ** objects declared as MEMBER types can be admissible arguments to
333 ** printing characters should simply be printed by the FORMAT ~:C
335 ** compiler failure when compiling functions with hairy constant
336 defaults for optional parameters.
337 ** compiler produces wrong code when MAYBE-INLINE-expanding a
338 function, which is already optimized.
340 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
341 * new platform: experimental support for SBCL x86/Darwin, including
342 MacOS X 10.4.5 on Intel.
343 * new feature: Unicode character names are now known to the system
344 (through CHAR-NAME and NAME-CHAR).
345 * new feature: the filesystem location of SBCL's core file is
346 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
347 * minor incompatible change: the contrib modules SB-POSIX and
348 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
349 this change is to make it easier to distribute
350 location-independent binaries.
351 * bug fix: as implied by AMOP, standardized classes no longer have
352 slots named by external symbols of public packages. (reported by
354 * optimization: calling functions via constant symbols -- as in
355 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
356 via the function object as in (FUNCALL #'FOO).
357 * optimization: on x86 and x86-64, the calling convention for the
358 case of dealing with an unknown number of values has been altered
359 to be friendlier to the prediction heuristics implemented,
360 particularly with reference to CALL and RET pairing. (thanks to
361 Alastair Bridgewater)
362 * optimization: on x86, the code for access to thread-local symbol
363 values and binding/unbinding of thread-local symbols is smaller.
364 (thanks to Alastair Bridgewater)
365 * enhancement: CONSTANTP is now able to determine constantness of
366 more complex forms, including calls to constant-foldable standardized
367 functions and some special forms beyond QUOTE.
368 * fixed bug: occasional GC crashes on Solaris/x86
369 * optimization: x86-64 supports stack allocation of results of simple
370 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
371 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
372 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
373 (thanks to James Knight)
374 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
375 by the standard (thanks to Stephen Compall)
377 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
378 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
379 be used for bundling the runtime and the core file into one
380 executable binary. This feature is not currently supported on all SBCL
381 platforms. (thanks to James Bielman and NIIMI Satoshi)
382 * new feature: a generational or ephemeral garbage collector is now
383 the default on the PowerPC platform (both Linux and Darwin). The
384 old Cheney (stop and copy) collector is a build-time option.
385 (thanks to Cyrus Harmon, after Raymond Toy)
386 * minor incompatible change: the method by which SBCL finds its
387 contributed modules has changed; it no longer relies on symbolic
388 links from an $SBCL_HOME/systems directory, but searches directly
389 in the subdirectories of $SBCL_HOME.
390 * enhancement: the dynamic heap size on the Linux/PPC platform is
391 markedly larger, even using the older Cheney garbage collector.
392 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
393 certain complicated string types. (reported by Gary King)
394 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
395 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
396 * fixed bug: DOCUMENTATION on structures no longer signals an
397 error if no structure type of the right name exists. (reported by
399 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
400 applicable non-standard primary or :AROUND method on
401 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
402 no longer cause unbound variable errors. (reported by Kevin Reid)
403 * fixed bug: metacircle resolution in cases where methods have slots
404 added before the slots from STANDARD-METHOD. (reported by Jean
406 * fixed bug: the Power PC assembly code for calling into lisp
407 sometimes computed its return address wrongly (depending on the
408 15th bit of the address, fixed at link time). This used to
409 manifest itself in a segmentation violation while building PCL.
410 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
411 many others over the years)
412 * fixed bug: ROOM no longer reports silly numbers for stack usage if
413 the stack pointer (treated as a signed integer) is negative.
414 (thanks to Peter van Eynde)
416 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
417 * new platform: experimental support for the Windows operating
418 system has been added. (thanks to Alastair Bridgewater)
419 * fixed several bugs in and robustified the PPC FFI (including
420 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
421 * bug fix: allow non-simple string symbol names (reported by Paul
423 * bug fix: interrupt handling on NetBSD (thanks to Richard M
425 * bug fix: saving a core corrupted callbacks on x86/x86-64
426 * bug fix: closed a loophole in metacircularity detection and
427 grounding in the PCL implementation of CLOS.
428 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
429 (reported by tomppa on #lisp)
430 * bug fix: clear the direction flag on Lisp -> C transitions, as
431 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
432 (reported by Andrej Grozin and Hendrik Maryns)
433 * optimization: major improvements to GC efficiency on GENCGC platforms
434 * optimization: faster implementation of EQUAL
435 * optimization: emit more efficient opcodes for some common
436 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
437 * optimization: several other minor code-generation improvements on x86-64
438 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
440 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
441 * minor incompatible change: (SETF CLASS-NAME) and (SETF
442 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
443 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
444 by ANSI as a generic function, and for consistency (SETF
445 GENERIC-FUNCTION-NAME) is treated likewise.
446 * fixed bug #233.b: make constraint propagation notice when a variable
447 value is changed after it is referenced but before it is used
448 * fixed bug #296: no more arbitrary behaviour from filenames with
449 odd characters as --load arguments. (reported by Adam Warner)
450 * fixed bug #390: :CHARACTER-SET pathname components now work as
451 expected. (reported by Tim Daly Jr)
452 * fixed bug #391: complicated :TYPE intersections in slot
453 definitions no longer cause an error in PCL internals.
454 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
455 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
456 returning the number of octets which would be written to the
457 file-stream. (thanks to Robert J. Macomber)
458 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
459 arguments to the internal stream functions. (thanks to David
461 * bug fix: stack exhaustion is now handled on the x86/SunOS
462 platform. (thanks to Daisuke Homma)
463 * bug fix: more accurate ROOM results on GENCGC platforms
464 * optimization: improved type inference for arithmetic-for
465 index variables in LOOP
466 * optimization: faster floating-point SQRT on x86-64
467 * fixed some bugs revealed by Paul Dietz' test suite:
468 ** DOCUMENTATION returns NIL instead of "" for method combinations
469 that don't have a docstring
471 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
472 * minor incompatible change: (SETF CLASS-NAME) and (SETF
473 GENERIC-FUNCTION-NAME) are no longer generic functions, and
474 therefore are not customizeable by user code (as seems to be at
475 least permitted and maybe required by AMOP). As a consolation,
476 however, the SBCL implementation of these functions now calls
477 REINITIALIZE-INSTANCE as specified by AMOP.
478 * bug fix: slot-definition documentation is propagated as per ANSI
479 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
480 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
481 argument, as do eslotd initialization methods. (from Pascal
482 Costanza's "Closer" project)
483 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
484 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
486 * bug fix: it is now possible to have more than one subclass of
487 STANDARD-GENERIC-FUNCTION without causing stack overflow.
488 (reported by Bruno Haible, Pascal Costanza and others)
489 * bug fix: the dependent update protocol now works for generic
490 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
492 * bug fix: condition-class instances corresponding to
493 DEFINE-CONDITION forms are now created eagerly. (reported by
494 Kalle Olavi Niemitalo on comp.lang.lisp)
495 * bug fix: floating point printing is more accurate in some
496 circumstances. (thanks to Simon Alexander)
497 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
498 merged with *DEFAULT-PATHNAME-DEFAULTS*.
499 * bug fix: callbacks on OS X now preserve stack-alignment.
500 (thanks to Cyrus Harmon)
501 * enhancement: the x86-64 disassembler is much better at
502 disassembling SSE instructions. (thanks to Lutz Euler)
503 * enhancement: CHECK-TYPE on a variable now allows type-inference
504 (thanks to Vincent Arkesteijn)
505 * optimization: improved performance of EUC-JP external format.
506 (thanks to NIIMI Satoshi)
507 * optimization: performance improvements to IO on file streams of
508 :ELEMENT-TYPE CHARACTER
509 * optimization: much faster memory allocation on x86-64
510 * optimization: faster garbage collections (latency and throughput)
512 * optimization: faster bignum subtraction and division on x86 and x86-64
513 (thanks to Lutz Euler)
514 * optimization: more accurate type-inference for loops that use a
515 floating point index variable or a negative step.
517 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
518 * bug fix: add a workaround to SBCL looping infinitely at startup on
519 Linux kernels with apparently buggy implementations of personality().
520 (thanks to Svein Ove Aas)
521 * bug fix: Unicode symbols are correctly printed in LDB backtraces
522 (thanks to David Lichteblau)
523 * bug fix: local bindings shadow global symbol macros for
524 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
525 * optimization: non-open coded uses of numeric comparison operators
526 (e.g. >) no longer cons when called with more than one parameter
527 on platforms supporting dynamic-extent allocation.
528 * enhancement: saving cores with foreign code loaded is now supported
529 on MIPS/Linux in addition to the previously supported platforms.
530 * bug fix: FIND-RESTART now tests for activity, not applicability when given
531 a restart object as identifier. (reported by Helmut Eller for CMUCL)
532 * bug fix: division by zero in sb-sprof when no samples were collected
533 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
534 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
535 (thanks to Alexander Kjeldaas)
536 * new feature: ignoring the return values of destructive functions
537 like NREVERSE signals a compile-time style-warning.
538 (thanks to Kevin Reid)
539 * new feature: the alignment of alien structure fields can be
540 explicitly specified. (thanks to Cyrus Harmon)
541 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
542 and dump core on SIGQUIT
544 ** incompatible change: threads do not inherit values of specials
545 from their parents (see manual)
546 ** bug fix: threads stacks belonging to dead threads are freed by the
547 next exiting thread, no need to gc to collect thread stacks anymore
548 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
549 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
550 ** bug fix: dynamic variable and thread start related gc lossage
551 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
553 * fixed some bugs revealed by Paul Dietz' test suite:
554 ** SUBTYPEP is slightly more accurate on heinously complicated
555 CONS types where some of the members have uncertain (in the
556 NIL, NIL sense) type relationships to each other.
558 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
559 no longer purified unless :PURIFY T is explicitly specified.
560 ** Non-purified cores are significantly smaller than before
562 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
563 * new feature: timers based on Zach Beane's excellent timer package
564 * added support for the following external formats: koi8-u,
565 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
566 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
567 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
568 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
569 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
570 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
571 * incompatible change: a threaded SBCL will no longer revert to
572 non-threaded mode on non-NPTL systems, but refuse to start entirely.
573 * bug fix: interrupts are disabled until startup is complete; no
574 more sigsegvs when receiving a signal to soon
575 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
577 * bug fix: add a workaround for the memory randomization features in
578 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
579 workaround will only be in effect on systems with the proc filesystem
581 * bug fix: printing objects of type HASH-TABLE signals a
582 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
584 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
585 have been running for over 50 days. (reported by Gilbert Baumann)
586 * bug fix: the logic for getting names of functions gets less
587 confused when confronded with alternate-metaclass
588 funcallable-instances. (reported by Cyrus Harmon)
589 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
590 consistent, even on internal alternate-metaclass objects.
591 * bug fix: flush closure information collected by physical
592 environment analysis prepass before the main pass. (bug reported
594 * bug fix: compiler pack phase does not modify a hashtable, which is
595 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
596 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
597 may be expanded. (reported by Alexander <malishev> on c.l.l)
598 * bug fix: no more style-warnings when compiling code using LAST
599 under high-SPEED low-SPACE settings. (reported by David Wallin
601 * fixed some incompatibilities between SBCL's MOP and the MOP
603 ** the METAOBJECT class is now implemented;
604 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
605 STANDARD-OBJECT, as required;
606 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
607 now compatible (as required by VALIDATE-SUPERCLASS); there
608 remains a consistency requirement over the presence or absence
609 of the FUNCTION class in the superclass list of finalized
610 classes; see the manual for more details;
611 ** the :AROUND method for COMPUTE-SLOTS on
612 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
613 requested slot ordering.
615 ** bug fix: parent thread now can be gc'ed even with a live
617 ** bug fix: binding a special with PROGV to no value is not
618 the same as that symbol not having been bound (thanks to
620 ** bug fix: binding specials is thread safe (thanks to
622 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
623 and SB-SPROF do not die with 'no handler for signal XX in
624 interrupt_handle_now(..)' anymore
625 ** bug fix: WITH-TIMEOUT works with multiple running threads
626 * fixed some bugs revealed by Paul Dietz' test suite:
627 ** ENSURE-GENERIC-FUNCTION should take a method class object for
628 the :method-class keyword argument.
630 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
631 * new port: the Solaris operating system on x86 processors is now
632 mostly supported, though some rough edges in the environment
633 remain. (thanks to Daisuke Homma)
634 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
635 space for its heap. (thanks to Thiemo Seufer)
636 * optimization: SBCL on MIPS platforms now supports dynamic-extent
637 closures. (thanks to Thiemo Seufer)
638 * minor incompatible change: eof selects abort in the debugger.
639 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
640 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
642 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
643 deprecated, and will go away in a future revision of SBCL.
644 * minor incompatible change: GC-ON and GC-OFF are no longer
645 implemented with a counter, it does not matter how many times gc
646 is switched on or off
647 * bug fix: discriminating functions for generic function classes
648 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
649 make invalid assumptions about method precedence order. (reported
651 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
652 platforms. (thanks to Thiemo Seufer)
653 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
654 * bug fix: correct pathname printing with printer escaping is on.
655 (thanks to Kevin Reid)
656 * bug fix: complex VOP definitions in "user-space" no longer trigger
657 package locks. (reported by Zach Beane)
658 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
659 for funcallable-instance functions, and is no different from
661 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
662 characters in the current readtable. (reported by Nicholas Neuss)
663 * bug fix: SUBTYPEP on various CONS types returns more a more
664 accurate acknowledgment of its certainty.
665 * optimizations: REMOVE-DUPLICATES now runs in linear time on
666 lists in some cases. This partially fixes bug 384.
667 * flush all standard streams before prompting in the REPL and the
669 * bug fix: signal handling and triggering gc do not conflict
670 directly anymore, in particular a high frequency sb-sprof does
671 not prevent gc from running
672 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
673 approximation for timezone and DST information between the
674 universal time epoch and the smallest negative 32-bit time_t.
675 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
676 year 1899 so long as the encoded time is non-negative.
677 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
678 the correct answer for non-constant rotate amounts when performing
679 an inline 32-bit rotation.
681 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
682 there is only one thread in the session
683 ** bug fix: memory leak for streams created in one thread and
684 written to in another
685 ** bug fix: lockup when compiled with gcc4
686 ** bug fix: race that allows the gc to be triggered when gc is
688 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
689 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
691 ** bug fix: finished threads are gc'ed properly
692 * fixed some bugs revealed by Paul Dietz' test suite:
693 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
694 has arguments with a different set of applicable methods from
695 the orignal arguments.
696 ** The type error thrown by MAP now has a correct expected-type
698 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
699 name a compiled function.
700 ** Three MISC tests where a large bignum was improperly coerced to
701 a float that couldn't represent that bignum during type
702 derivation were fixed.
703 ** SUBTYPEP can now handle types involving the negation of a
704 list-form FUNCTION type.
705 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
706 as (COMPLEX (AND RATIO (NOT FIXNUM))).
707 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
709 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
710 * New feature: Experimental support for bivalent streams: streams
711 opened with :element-type :default now allow character and binary
712 (unsigned-byte 8) I/O
713 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
714 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
716 * bug fix: no more highly sporadic "couldn't check whether ~S is
717 readable" when reading a stream and an interrupt hits in the middle
718 of a select system call
719 * compiler better recognizes complex arrays (reported by Eduardo
721 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
722 for variables on SBCL built with linkage-tables. (reported by Luis
724 * various error reporting improvements.
725 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
726 (Thanks to Thiemo Seufer)
727 * bug fix: nested reader invokations work correctly
728 * bug fix: it is possible to have simultaneous references to foreign
729 code and foreign data with the same name.
731 ** added x86-64 support
732 ** incompatible change: the threading api now works with thread
733 objects instead of thread ids
734 ** bug fix: threads are protected from signals and interruption when
735 starting up or going down
736 ** bug fix: a race where an exiting thread could lose its stack to gc
737 ** bug fix: don't halt on infinite error in threads if possible
738 ** fixed numerous gc deadlocks introduced in the pthread merge
739 ** bug fix: fixed thread safety issues in read and print
740 ** bug fix: debugger doesn't hang on session lock if interrupted at
741 an inappropriate moment
742 ** bug fix: run-program is now thread safe(r)
743 ** bug fix: inner with-recursive-lock no longer releases the mutex
744 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
745 * fixed some bugs revealed by Paul Dietz' test suite:
746 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
747 EXPECTED-TYPE slots filled.
748 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
750 ** MAKE-CONDITION accepts classes as type-designators.
751 ** COMPILE may never return NIL.
752 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
753 range before calling Unix time functions
755 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
756 * numerous signal handling fixes to increase stability
757 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
758 * minor incompatible change: we now correctly canonize default
759 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
760 as per the MOP, rather than the historical (INITARG INITFUNCTION
761 INITFORM). (reported by Bruno Haible)
762 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
763 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
764 for more information.
765 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
766 pathname is a directory pathname.
767 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
768 * dynamic space size on PPC has been increased to 768Mb. (thanks to
770 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
771 :METACLASS argument in addition to a class name. (reported by
772 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
773 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
774 (reported by Bruno Haible; patch from Gerd Moellmann)
775 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
777 * bug fix: more cleanups to the floating point exception handling on
778 x86-64 (thanks to James Knight)
779 * bug fix: the compiler does not try to constant fold calls of
780 COERCE and BIT-* functions when they return freshly constructed
781 objects. (reported by Nikodemus Siivola and Paul Dietz)
782 * optimization: improved the MIPS versions of generic (in the
783 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
784 * optimization: direct conversion of (unsigned-byte 32) to floats on
785 the PowerPC platform.
786 * optimization: structure instances with raw slots now use less
787 memory, and probably show better memory locality. (thanks to
789 * optimization: DYNAMIC-EXTENT declarations for lists and closures
790 are treated as requests for stack allocation on the x86-64,
791 Alpha-32, PPC and SPARC platforms.
792 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
793 the wrong file descriptor; implementation of SOCKET-OPEN-P.
794 (thanks to Tony Martinez)
796 ** gcing a dead thread can no longer lead to lockups
797 ** threads block signals until they are set up properly
798 ** errno is no longer shared by threads
799 ** interrupt-thread restores the eflags register on x86
800 ** fixed some lockups due to gc/thread interaction
801 * fixed some bugs revealed by Paul Dietz' test suite:
802 ** invalid dotted lists no longer raise a read error when
804 ** signal an error if a symbol that names a declaration is used
805 as the name of a type, or vice versa
806 ** allow using the (declare (typespec var*)) abbreviation for
807 (declare (type typespec var*)) with all type specifiers
808 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
809 ** always return NIL from PPRINT-POP when OBJECT is NIL
810 ** don't signal errors when pretty-printing malformed LABELS,
811 FLET or MACROLET forms
812 ** declarations in a DOLIST body are also in scope for the
814 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
816 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
819 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
820 * fixed cross-compiler leakages that prevented building a 32-bit
821 target with a 64-bit host compiler.
822 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
823 opened with :IF-EXISTS :OVERWRITE.
824 * fixed bug 281: error for an invalid qualifier in a short-form method
825 combination method is not signalled until the faulty method is called.
826 * bug fix: iteration variable type inferrer failed to deal with open
827 intervals. (reported by Alan Shields)
828 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
829 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
830 or more runtime options were provided to the sbcl binary.
831 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
832 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
833 * optimization: there's now a fast-path for fixnum arguments in the
834 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
835 * optimization: the code generated on x86-64 is more compact thanks
836 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
837 * bug fix: floating-point exception handling now partly works on
838 x86-64. (Thanks to James Knight)
839 * improvement to the MIPS backend from Thiemo Seufer: C-style
840 64-bit long long arguments and return values to alien functions
842 * fixed some bugs revealed by Paul Dietz' test suite:
843 ** the type-error signalled from WARN has a filled-in DATUM slot.
844 ** the type-error required when a stream is not associated with
845 a file has the stream as its datum.
846 ** type-errors on single-floats on x86-64 no longer have
847 :INVALID-OBJECT as the datum
848 ** the type-errors signalled for invalid function names now have
849 a correct expected type
850 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
851 for structures defined with a :TYPE. Documentation strings for
852 typed structures are no longer immediately discarded
853 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
854 broadcast streams and synonym streams. FILE-LENGTH now also works
855 on broadcast streams.
857 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
858 * incompatible change: the --noprogrammer option, deprecated since
859 version 0.7.5, has been removed. Please use the equivalent
860 --disable-debugger option instead.
861 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
863 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
864 inoperational for a while now) has been completely removed.
865 * null lexical environments are now printed as #<NULL-LEXENV>,
866 significantly reducing the amount of clutter in typical backtraces.
867 * documentation on weak pointers, finalization, and after GC hooks
868 has been added to the manual.
869 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
870 as other specialized array types, is much faster. SUBSEQ and
871 COPY-SEQ on such arrays have also been sped up.
872 * optimization: EQL is now more efficient when at least other argument
873 is known to be of type (OR FIXNUM (NOT NUMBER)).
874 * fixed bug: the runtime is now less vulnerable to changes in the
875 size of the SBCL object on OS X, and virtual memory is reserved for
876 all spaces that need to be at a fixed address.
877 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
878 * fixed bug: finalizers and after GC hooks that cause consing are now
880 * fixed bug: compiler error messages and summaries are now printed to
881 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
882 * fixed inference of the upper bound of an iteration variable.
883 (reported by Rajat Datta).
884 * fixed bug 211e: calling local functions with duplicated constant
885 keyword argument no longer causes a bogus style warning about an
887 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
888 ftype declaration to be disregarded. (reported by Dave Roberts)
889 * fixed bug 373: caused by erronous compilation of references to alien
890 variables in the runtime on ppc/darwin.
891 * fixed bug 376: CONJUGATE type deriver.
892 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
893 (reported by Baughn on #lisp)
894 * a cross-compiler bug on non-x86 platforms has been identified and
895 fixed. (thanks to Bruno Haible)
896 * improvements to the MIPS runtime code for increased stability.
897 (thanks to Thiemo Seufer)
898 * increased the maximimum compact environment size to allow
899 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
900 * partial workaround for bug 135: don't name the function we're
901 calling for hairy cases of EVAL, so as not to accumulate one environment
902 entry per call to EVAL. (reported by Kevin Reid)
903 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
904 * optimization: type testing for non-vector arrays should be faster.
905 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
907 * bug fix: setting 31st element of a bit vector to zero did not work
909 * bug fix: redefining a class definition which failed due to a
910 previous accessor / function clash now works (but see BUGS entry
911 #380 for more problems in this area). (thanks to Zach Beane)
912 * the long form of DEFINE-METHOD-COMBINATION disables method group
913 checking when given a single method group with pattern *. (thanks
915 * on x86 compiler supports stack allocation of results of simple
916 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
917 * support for building and running on Mac OS X 10.4 "Tiger" has been added
918 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
919 * fixed some bugs related to Unicode integration:
920 ** the restarts for recovering from input and output encoding
921 errors only appear when there is in fact such an error to
923 * fixed some bugs revealed by Paul Dietz' test suite:
924 ** MISC.549 and similar: late transformation of unsafe type
925 assertions into derived types caused unexpected code
927 ** SCALE-FLOAT type deriver is less wrong.
928 ** type derivers for EXP, LOG and similar functions compute result
929 types for complex arguments better.
930 ** (MISC.563) CONJUGATE type deriver works for very restricted
932 ** out-of-line type testers for character strings are available.
933 ** EQUAL compiler transform understands specialness of objects
935 ** accessing double-floats stored on the stack now works on x86-64.
936 ** debugger internals could sometimes create invalid lispobjs,
937 resulting in GC crashes.
938 ** MISC.548: type check weakening can convert required type into
940 ** initialization forms for bindings are not in scope of free special
943 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
944 * incompatible change: thread support for non-NPTL systems has
945 been removed - locking is buggy and unreliable. A threaded
946 SBCL build will now warn at startup and refuse to create
947 new threads, unless futex support is detected in the kernel
948 * incompatible change: the top level REPL now has only an ABORT
949 restart associated with it, not TOPLEVEL and ABORT as it used to.
950 TOP and TOPLEVEL are now available as debugger commands for
951 returning to the top level.
952 * incompatible change: forms evaluated in the REPL now use the
953 global optimization policy.
954 * incompatible change: user- and system-initialization files are
955 no longer processed with LOAD, but by READ and EVAL; hence the
956 global optimization policy, startup package, readtable, etc,
958 * The .fasl file format number has been incremented because of
959 various incompatible changes.
960 * internal entry point details and argument counts no longer appear
961 in backtraces unless explicitly requested by setting
962 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
963 * built-in and standard functions no longer have names like "top
964 level local call to FOO".
965 * fixed bug 32: functions defined in non-null lexical environments
966 now have more legible printed representation
967 * fixed bug 33: functions defined in non-null lexical environemnts
968 are now more amenable to inspection by INSPECT.
969 * workaround for bug 354: XEPs no longer appear in backtraces unless
970 explicitly requested.
971 * fixed bug: receiving the signal which results from attempting to
972 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
973 SIGBUS on other *BSDs) on architectures where the C stack is also
974 the Lisp stack (x86 and x86-64) from foreign code no longer leads
975 to debugger confusion or wild instability. (reported by Cheuksan
977 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
978 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
979 (reported by Lutz Euler)
980 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
981 compiled code. (reported by Paul Dietz)
982 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
983 specialized parameter to a method to something that is not TYPEP
984 the specializer is now possible.
985 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
986 face of package deletion.
987 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
988 pathname. (reported by David Tolpin on c.l.l)
989 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
990 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
991 STRING-STREAM and FILE-STREAM, respectively.
992 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
993 than previously; in addition, its test suite is now run on build.
994 (thanks to Luke Gorrie)
995 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
996 * optimization: added a immediate representation for single-floats
998 * fixed some bugs revealed by Paul Dietz' test suite:
999 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1000 ** attempting to create a package with a colliding nickname causes
1001 correctable errors to be signalled.
1002 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1003 ** several x86-64 backend bugs related to sign-extension of immediate
1006 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1007 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1008 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1010 * fixed loading of multiply forward-referenced layouts.
1011 (thanks to Cheuksan Wang)
1012 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1013 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1014 (see COMPILE-FILE documentation for details of :PRINT option)
1015 * fixed bugs 19 and 317: fixed-format floating point printing is
1016 more accurate. This also fixes a bug reported by Adam Warner
1017 related to the ~@F format directive.
1018 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1020 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1021 dispatch table if the from-char is a dispatch macro character.
1022 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1023 by the wordsize no longer ignore the last word. (reported by Lutz
1025 * fixed bug in type checking of dynamic-extent variables. (reported
1027 * optimization: sequence traversal functions use their freedom to
1028 coerce function designators to functions.
1029 * optimization: code with many calls to CLOS methods specialized on
1030 CLOS classes has had redundant type checks removed.
1031 * optimization: added declarations to speed up operations that access
1032 the internal character database (for example STRING-UPCASE)
1033 * optimization: comparison operations between floats and sufficiently small
1034 fixnums no longer create extra rationals
1035 * fixed some bugs related to Unicode integration:
1036 ** portions of multibyte characters at the end of buffers for
1037 character-based file input are correctly transferred to the
1038 start of the buffer at the next read.
1039 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1040 passing it through to OPEN.
1041 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1042 argument given to internal calls to OPEN.
1043 ** the built-in comment readers (introduced by character sequences
1044 ";" and "#|") are more forgiving to encoding errors; they will
1045 STYLE-WARN and then attempt to resync the stream at a character
1046 boundary. (thanks to Teemu Kalvas)
1047 * fixed some bugs in the x86-64 port:
1048 ** Negative short int return values from c-calls are sign-extended
1050 ** The stack is aligned to 16-bytes for c-calls, as required by
1051 the ABI. (thanks to Cheuksan Wang)
1052 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1053 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1054 for enums. (thanks to Vincent Arkesteijn)
1055 ** Multiple small optimizations and bugfixes for floating point
1057 * fixed some bugs revealed by Paul Dietz' test suite:
1058 ** Space, Tab, Linefeed, Return and Page have the invalid
1059 secondary constituent character trait.
1060 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1062 ** WITH-INPUT-FROM-STRING should only update the index place on
1064 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1065 works more reliably.
1066 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1067 with one-dimensional arrays with fill pointers.
1068 ** TYPE-OF failed on a complex with an integer realpart and a
1070 ** compiler failure during type inference for the code of form
1071 (IF (EQL X (THE ...)) ...) (MISC.535).
1073 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1074 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1075 platform. The port supports 61-bit fixnums, large memory spaces
1076 and reloading shared object files.
1077 * enhancement: saving cores with foreign code loaded is now
1078 supported on ppc/Darwin in addition to the previously supported
1080 * enhancement: the statistical profiler now walks deeper into the
1081 call stack for more meaningful call-graphs and accrued time
1082 reports (x86/x86-64 only). It also now reports time spent in
1084 * enhancement: it is now possible to trace most individual methods
1085 of a generic function in addition to tracing the generic function
1087 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1089 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1090 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1091 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1092 produces an error. (thanks to Vincent Arkesteijn)
1093 * bug fix: NAMESTRING on pathnames with :WILD components in their
1094 directories works correctly. (thanks to Artem V. Andreev)
1095 * fixed bug 125: compiler preserves identity of closures. (reported
1097 * bug fixed: functions with &REST arguments sometimes failed with
1098 "Undefined function" when compiled with (DEBUG 3). (reported by
1100 * bug fix: overflow during compiling of setting element of a bit
1101 vector with constant index and value. (reported by Timmy Douglas)
1102 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1103 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1104 * on x86 compiler partially supports stack allocation of dynamic-extent
1106 * GO and RETURN-FROM do not check the extent of their exit points
1107 when compiled with SAFETY 0.
1108 * fixed some bugs related to Unicode integration:
1109 ** encoding and decoding errors are now much more robustly
1110 handled; it should now be possible to recover even from invalid
1111 input or output to the terminal. (thanks to Teemu Kalvas)
1112 ** provided a first cut at implementing STRING-TO-OCTETS and
1113 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1114 ** altered the SB-MD5 contributed module slightly, changing the
1115 interface just enough for it to be supportable for builds where
1116 lisp characters are not eight bits.
1117 * fixed some bugs revealed by Paul Dietz' test suite:
1118 ** the FORMATTER-generated functions for ~V[ conditionals require
1119 the correct number of arguments.
1120 ** READ-FROM-STRING returns the mandated second value when applied
1121 to displaced strings.
1122 ** the #\Rubout and #\Backspace characters are treated as invalid
1123 constituent characters by the tokenizer.
1125 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1126 * new feature: reloading changed shared object files with
1127 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1128 * new feature: references to foreign variables and functions
1129 can now be compiled and loaded before the corresponding shared
1130 object file is loaded, as long as the foreign definitions are
1131 available at runtime.
1132 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1133 supported. (thanks to Dan Debertin)
1134 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1135 just keywords. (thanks to Vincent Arkesteijn)
1136 * fixed bug #331: structure-class instances corresponding to
1137 DEFSTRUCT forms are now created eagerly.
1138 * fixed bug #345: backtraces from calls to undefined functions work
1139 on x86 as well. Related bug #61 is now also partially fixed on x86
1140 (backtraces from throws to unknown catch tags.)
1141 * bug fix: lambda-list parsing is now stricter vrt. order and number
1142 of lambda-list keywords.
1143 * bug fix: as specified by AMOP, an error is signalled if a
1144 class-option appears multiple times in a DEFCLASS form. (reported
1146 * bug fix: RANDOM can be compiled when the compiler derives the type
1147 of its numeric argument as a disjoint set of small integers.
1148 (reported by Paul Dietz)
1149 * bug fix: starting a core saved with shared objects loaded when
1150 those objects are not available no longer causes threaded SBCL to
1151 hang. (reported by Sean Ross)
1152 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1154 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1155 argument, which must always be NIL. (reported by Kalle Niemitalo)
1156 * bug fix: printing 1.0d+23 no longer results in an error.
1157 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1158 * bug fix: structure slot setters preserve evaluation order. (thanks
1160 * fixed some bugs related to Unicode integration:
1161 ** RUN-PROGRAM can allow its child to take input from a Lisp
1162 stream. (reported by Stefan Scholl)
1163 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1164 locales. (reported by Ken Causey)
1165 * fixed some bugs revealed by Paul Dietz' test suite:
1166 ** INCF, DECF and REMF evaluate their place form as specified in
1168 ** FORMATTER expands ~{ ~} iteration directives with V or #
1169 parameters correctly.
1170 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1171 consequent uses no arguments correctly.
1172 ** the system has a partial understanding of the (COMPLEX RATIO)
1174 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1177 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1178 * new feature: a build-time option (controlled by the :SB-UNICODE
1179 keyword feature, enabled by default) for building the system with
1180 support for the entire 21-bit character space defined by the
1182 * new feature: the system now has rudimentary external-format
1183 support; the primary user-visible change at this time is that
1184 characters with the high bit set (such as lower-case-e-acute) will
1185 print correctly to a terminal in a UTF-8 environment.
1186 * minor incompatible change: BASE-CHAR no longer names a class;
1187 however, CHARACTER continues to do so, as required by ANSI.
1188 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1189 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1190 has been moved to the SB-EXT package (temporarily re-exported from
1192 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1193 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1195 * fixed bug #351: better error-handlind and reporting for malformed
1197 * fixed bug #350: bignum-printing is now more memory-efficient,
1198 allowing printing of very large bignums, eg. (expt 2 10000000).
1199 (reported by Bruno Haible)
1200 * fixed bug #302: better primitive-type selection for intersection
1202 * fixed bug #308: non-graphic characters now all have names, as
1203 required. (reported by Bruno Haible)
1204 * bug fix: redefining a class with different superclasses now correctly
1205 removes it from the direct-subclasses of its previous superclasses.
1206 (reported by David Morse)
1207 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1208 new value now works. (reported by Bruno Haible)
1209 * bug fix: correct canonicalization of multiple non-standard slot
1210 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1211 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1212 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1213 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1214 now exists, an signals an error.
1215 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1216 during macroexpansion for non-iteration variables as well. (reported
1217 by Bruno Haible for CMUCL)
1218 * bug fix: Cyclic structures and unprintable objects in compiler
1219 messages no longer cause errors. (reported by Bruno Haible)
1220 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1221 and READ-FROM-STRING all now return a primary value of NIL if
1222 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1223 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1224 (reported by Bruno Haible for CMUCL)
1225 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1226 do not designate a condition. (reported by Bruno Haible for
1228 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1229 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1230 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1231 permitting resolution in favour of any of the conflicting symbols.
1232 (reported by Bruno Haible for CMUCL)
1233 * bug fix: EQUAL compiler optimizations is less aggressive on
1234 strings which can potentially compare true despite having distinct
1235 specialized array element types.
1236 * bug fix: unit enumerations can be defined without dividing by
1237 zero. (thanks to Vincent Arkesteijn)
1238 * FORMAT compile-time argument count checking has been enhanced.
1239 (report from Bruno Haible for CMUCL)
1240 * a partial workaround for the bug 262: the compiler does not try to
1241 inline-expand a local function doing RETURN-FROM from a deleted
1242 BLOCK. (thanks to Peter Denno for the bug report and to David
1243 Wragg for the simple test case)
1244 * fixed some bugs revealed by Paul Dietz' test suite:
1245 ** RENAME-PACKAGE allows all package designators as new package
1247 ** constraint propagation and lambda variable substitution are
1248 more cautious in dealing with partially deleted code.
1249 ** compiler sometimes lost reoptimization passes.
1250 ** CERROR, given a condition as condition designator, passes its
1251 remaining arguments to the continue format control without
1253 ** Case-altering FORMAT directives work correctly on non-ASCII
1255 ** The REAL type specifier handles bounds outside the floating
1256 point ranges without signalling FLOATING-POINT-OVERFLOW.
1257 ** Functions with IR1-transformations can create intercomponent
1258 references to global functions.
1259 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1261 ** FORMAT ~R treats a nil value for its first parameter correctly.
1263 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1264 * enhancement: saving cores with foreign code loaded is now
1265 supported on x86/NetBSD and sparc/Linux in addition to the previously
1266 supported platforms.
1267 * bug fix: on some platforms repeated installations caused multiple
1268 copies of HTML documentation to be installed -- should not happen
1269 any more. (reported by Stefan Scholl)
1270 * bug fix: parsing self-recursive alien record types multiple times
1271 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1272 original patch by Helmut Eller for CMUCL)
1273 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1274 (thanks to Richard Kreuter)
1275 * bug fix: defining classes whose accessors are methods on existing
1276 generic functions in other (locked) packages no longer signals
1277 bogus package lock violations. (reported by François-René Rideau)
1278 * bug fix: special variables as DEFMETHOD parameters no longer have
1279 associated bogus type declarations. (reported by David Wragg and
1281 * bug fix: read-write consistency on streams of element-type
1282 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1283 * bug fix: redefiniton of the only method of a generic function with
1284 no DEFGENERIC no longer emits a full WARNING. In addition,
1285 redefinition of generic functions with no DEFGENERIC to an
1286 incompatible lambda list now signals an error. (thanks to Zach
1288 * bug fix: DEFGENERIC now works even when there's a function of the
1289 same name in an enclosing lexical environment. (thanks to Zach
1291 * fixed compiler failure, caused by instrumenting code during
1292 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1293 * optimization: added loop analysis and improved register allocation
1294 * fixed some bugs revealed by Paul Dietz' test suite:
1295 ** POSITION on displaced vectors with non-zero displacement
1296 returns the right answer.
1297 ** (SIMPLE-STRING) is a valid type specifier for sequence
1299 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1301 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1302 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1304 ** The justification version of the FORMAT ~< directive treats
1305 non-zero minpad parameter correctly.
1307 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1308 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1309 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1310 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1311 the supported interface.
1312 * new feature: Single-stepping of code compiled with DEBUG 2 or
1313 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1314 * new feature: saving cores with foreign code loaded is now
1315 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1316 Timothy Moore's work for CMUCL)
1317 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1318 parameters to * instead of NIL if no initform is supplied.
1319 (reported by Johan Bockgaard)
1320 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1321 correctly when FUNCALL appears as the car of the form. Note:
1322 despite this FUNCALL forms are not currently subject to
1323 compiler-macro expansion. (port of Raymond Toy's fix for the
1324 same from CMUCL, reported by Johan Bockgaard)
1325 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1326 (thanks for Teemu Kalvas)
1327 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1328 case instead of terminating the iteration (thanks for Julian
1329 Squires, Sean Champ and Raymond Toy)
1330 * bug fix: incorrect expansion of defgeneric that caused a style
1331 warning. (thanks for Zach Beane)
1332 * bug fix: slot accessor effective method computation works properly
1333 for classes with multiple non-standard applicable methods on
1334 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1335 * on x86 compiler supports stack allocation of results of LIST and
1336 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1337 CMUCL implementation by Gerd Moellmann)
1338 * fixed some bugs revealed by Paul Dietz' test suite:
1339 ** FORMAT strings with both the ~<~:;~> form of the justification
1340 directive and pretty-printing directives cause an error.
1342 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1343 * incompatible change: the internal functions
1344 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1345 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1346 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1347 instead of the old functions.
1348 * new feature: on platforms where "dladdr" is available foreign
1349 function names now appear in backtraces. (based on Helmut Eller's
1351 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1352 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1354 * documentation: ASDF manual is now distributed with SBCL.
1355 * bug fix: SBCL can now load its contributed modules with REQUIRE
1356 even if the system-provided entries have been removed from
1357 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1359 * bug fix: backtraces involving undefined functions or assembly
1360 routines are more informative. (thanks to Brian Downing)
1361 * bug fix: mutually referent alien structures now work correctly.
1362 (reported by Rick Taube)
1363 * bug fix: structures defined by WITH-ALIEN can be referred to
1364 within other definitions in the same WITH-ALIEN.
1365 * bug fix: division operators (MOD, TRUNCATE and the like) with
1366 constant zero divisors and integer dividends no longer generate
1368 * bug fix: provide default methods for INPUT-STREAM-P and
1369 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1370 * bug fix: improve the detection and resolution of MOP metacycles
1371 (where effective method computation depends on the generic function
1372 having its effective method computed). (reported by Bruno Haible)
1373 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1374 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1375 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1376 * bug fix: pathnames with relative directory components can be
1377 represented relative to default pathnames.
1378 * optimization: in taking the GCD of bignums, reduce the two bignums
1379 to approximately the same size (using Euclid's algorithm) before
1380 applying the more sophisticated binary GCD. (thanks to Juho
1382 * optimization: COUNT on bitvectors now operates word-at-a-time.
1383 * optimization: ASH with a positive, but not necessarily constant,
1384 (leftwards) shift, when in a modular context, is compiled to a
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1388 if the corresponding argument is NIL.
1389 ** FORMAT directives accepting a minpad parameter treat negative
1391 ** literal commas as character parameters to FORMAT directives are
1393 ** literal spaces directly after ~<Newline> directives within a
1394 format-logical-block (~:< ~@:>) do not induce :FILL-style
1395 conditional newlines.
1396 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1397 as specified: it no longer includes conditional newlines).
1398 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1400 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1401 :PER-LINE-PREFIX argument does not evaluate to a string.
1403 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1404 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1405 the manual for details; for now, package locks can be disabled by
1406 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1407 if no major problems are found then it is likely that they will be
1408 compiled in unconditionally.
1409 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1410 unsupported operators on all platforms. To load a shared library
1411 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1412 object file, link it into a shared library outside of SBCL and
1413 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1415 * bug workaround: ROOM T and the GC were not working together
1416 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1417 reporting facility aren't preserved. That reporting has been
1418 disabled, so now until and unless someone figures out how to make it
1419 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1420 * minor incompatible change: as threatened around sbcl-0.8.0, the
1421 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1422 The symbols are also exported from SB-PCL for backwards
1423 compatibility, but more so than before SB-PCL should be treated as
1424 an implementation-internal package.
1425 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1427 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1428 It is known as of this release not to work on the Alpha, however.
1429 * fixed bug #167: errors signalled due to illegal syntax in method
1430 bodies are now more legible.
1431 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1432 designators and can hence be used with TYPEP.
1433 * fixed bug #333: CHECK-TYPE now ensures that the type error
1434 signalled, if any, has the right object to be accessed by
1435 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1436 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1437 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1439 * fixed bug #334: programmatic addition of slots using specialized
1440 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1441 and :ALLOCATION :CLASS slots.
1442 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1443 float exponents. (rereported by Peter Seibel)
1444 * fixed a bug: #\Space (and other whitespace characters) are no
1445 longer considered to be macro characters in standard syntax by
1446 GET-MACRO-CHARACTER.
1447 * fixed bug: initialization of condition class metaobjects no longer
1448 causes an instance of the condition to be created. (reported by
1450 * fixed bug: it is now possible to have slots such that
1451 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1452 description is neither :INSTANCE nor :CLASS.
1453 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1454 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1455 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1456 non-standard SLOT-DEFINITION-ALLOCATION values.
1457 * fixed bugs: various race conditions handling exiting threads.
1458 CL-PPCRE's thread torture test now passes.
1459 * fixed arguably-a-bug: GC time proportional to number of threads in
1460 system even when most of them are idle
1461 * optimization: improved performance of BIT and SBIT on bit-vectors.
1462 * .fasl file incompatibility: The fasl file version number has
1463 been incremented because of changes associated with package locks.
1465 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1466 * minor incompatible change: the system no longer provides
1467 optimization hints (conditions which are TYPEP
1468 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1469 mode; these hints are still emitted when higher SPEED optimization
1471 * new contrib module: a sampling profiler (profiling by statistical
1472 sampling, rather than by instrumenting functions) is available as
1473 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1474 * the behaviour of the standard function ED is now customizeable by
1475 third parties through a hook variable: see ED's documentation
1476 string for information on the protocol.
1477 * the compiler no longer emits efficiency notes for (FUNCALL X)
1478 when the type of X is uncertain under default optimization
1480 * fixed bug 276: mutating a binding of a specialized parameter to a
1481 method to something that is not TYPEP the specializer is now
1483 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1484 exhibit the required behaviour on the x86 platform. (thanks to
1485 Peter van Eynde, Eric Marsden and Bruno Haible)
1486 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1487 even for difficult arguments. (reported by Peter Graves)
1488 * fixed bug 141a: the backquote printer now descends quoted
1490 * fixed another bug in backquote printing: no more destructive
1491 modification of the form's list structure. (reported by Brian
1493 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1494 move between its address being taken and the call to
1495 interrupt_thread, fixing a crashing race condition.
1496 * the SB-POSIX contrib implementation has been adjusted so that it
1497 no longer exhibits ridiculously poor performance when constructing
1498 instances corresponding to C structs.
1500 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1501 * minor incompatible change: the sb-grovel contrib now treats C
1502 structures as alien (in the sense of SB-ALIEN) objects rather than
1503 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1504 has implications for memory management of client code
1505 (sb-grovel-returned objects must now be manually managed) and for
1506 type safety (alien objects now have full types).
1507 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1508 used to control emission of compiler diagnostics, rather than the
1509 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1510 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1511 quality should be considered deprecated.
1512 * install.sh now installs the user manual as well
1513 * (not quite a new documentable feature, but worth considering in
1514 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1515 beginnings of a semantically meaningful condition hierarchy is
1516 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1518 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1519 designator as the defaults argument.
1520 * fixed bug: Displaced arrays whose displaced-to array has become
1521 too small now cause ARRAY-DIMENSION to signal an error, providing
1522 for safer bounds-checking. (reported by Bruno Haible)
1523 * fixed bug: DEFCLASS slot definitions with identical :READER and
1524 :WRITER names now signal a reasonable error. (reported by Thomas
1526 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1528 * fixed bug: Invalid slot specification errors now print correctly.
1529 (thanks to Zach Beane)
1530 * fixed bug 320: Shared to local slot value transfers in class
1531 redefinitions now happen corrently from superclasses as
1532 well. (reported by Bruno Haible)
1533 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1535 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1536 improper lists correctly. (reported by Bruno Haible)
1537 * fixed bug 313: source-transform for <fun-name> was erroneously
1538 applied to a call of a value of a variable with name <fun-name>.
1539 (reported by Antonio Menezes Leitao)
1540 * fixed bug 307: The obsolete instance protocol ensures that
1541 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1542 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1543 remove dead unknown-values globs from the middle of the stack.
1544 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1545 easier to resume long computations after using *BREAK-ON-SIGNALS*
1546 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1547 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1548 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1549 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1550 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1551 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1553 * fixed a (fairly theoretical) bug in string printing: if
1554 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1555 does not have array-element-type equal to the most general string
1557 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1558 function name. (thanks to Nikodemus Siivola)
1559 * fixed bug: compile-time format string checker failed on
1560 non-closed ~{. (reported by Thomas F Burdick)
1561 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1562 don't warn on legal loop code involving more than one aggregate
1563 boolean. (thanks to Nikodemus Siivola)
1564 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1565 in keyword position to MAKE-PACKAGE was not being reported as
1567 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1568 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1570 * fixed bugs 280 and 312: the checking for multiple definitions in a
1571 file is less likely to become confused by uses of inline
1573 * fixed bug: the #S reader macro performs the keyword coercion
1574 specified for slot names. (reported by Kalle Niemitalo)
1575 * fixed bug: lambda lists may contain symbols whose names start with
1576 & but are not lambda-list-keywords; their occurrence triggers a
1578 * fixed bug 321: define-method-combination argument lambda lists do
1579 not cause invalid code to be generated when &optional or &aux
1580 variables have default values. (reported by Bruno Haible)
1581 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1582 have CLOS classes; this appears also to have fixed reported
1583 instability in stack exhaustion detection.
1584 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1585 MUFFLE-WARNING when no associated restart is present is now
1587 * optimization: rearranged the expansion of various defining macros
1588 so that each expands into only one top-level form in a
1589 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1591 * optimization: used a previously little-used slot in symbols to
1592 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1594 * fixed some bugs revealed by Paul Dietz' test suite:
1595 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1597 ** Fixed an optimization bug related to inheritance of initforms
1598 from local to shared slots.
1599 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1600 ** CAST optimizer forgot to flush argument derived type.
1601 ** print/read consistency on floats is now orders of magnitude
1602 more likely. (thanks also to Bruno Haible for a similar report
1604 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1605 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1606 for objects of type REAL. Make it so.
1607 ** SXHASH is specified (infelicitously) to respect similarity,
1608 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1609 it so. (thanks to Markus Ziegler)
1610 ** on the Alpha, the compiler succeeds in compiling functions
1611 returning a known number of arguments greater than 63.
1612 ** fixed handling of invalid NIL arguments in keyword position
1614 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1616 ** provided a readably-printable representation for RANDOM-STATE
1618 ** ensured that pathnames, where they have a namestring, always
1619 print using #P"..." syntax.
1621 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1622 * Support for the forthcoming 2.0 version of the NetBSD kernel
1623 running on x86 hardware has been added. (thanks to Perry
1624 E. Metzger most immediately, and others for their past work)
1625 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1626 current release version; SBCL's OpenBSD support had been broken
1627 since about the time of OpenBSD's switch to ELF binary format.)
1628 * [placeholder for DX summary]
1629 ** user code with &REST lists declared dynamic-extent, under high
1630 speed or space and low safety and debug optimization policy.
1631 * The manual has been converted to Texinfo format and the debugger
1632 chapter from the cmucl manual has been added.
1633 * A facility has been added to extract documentation strings from
1634 sbcl and store them as Texinfo-formatted snippets for inclusion in
1635 the manual (via Texinfo's @include directive)
1636 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1637 the test case to Dave Roberts)
1638 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1639 pointers. (reported by Sean Ross)
1640 * bug fix: PROFILE output is printed nicely even for large numerical
1641 values. (thanks to Zach Beane)
1642 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1643 greater than 32 handle EOF correctly.
1644 * bug fix: on X86 an immediate argument of the IMUL instruction is
1645 correctly printed in disassembly. (thanks to Lutz Euler)
1646 * bug fix: class slots in redefined classes preserve their old
1647 values. (thanks to Bruno Haible and Nikodemus Siivola)
1648 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1649 succeeds. (reported by Marco Baringer)
1650 * fixed some bugs revealed by Paul Dietz' test suite:
1651 ** READ-SEQUENCE now works on ECHO-STREAMs.
1652 ** RATIONALIZE works more according to its specification. (thanks
1655 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1656 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1657 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1658 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1659 should matter to you only if you rebind the printer control
1660 variables and then find you want different bindings in the
1661 debugger than in the ordinary execution of your program.)
1662 * The runtime build system has been tweaked to support building
1663 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1664 assembler and linker. (thanks to Nikodemus Siivola)
1665 * Unbound, undefined, undeclared variables now trigger full
1666 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1667 more often programmer error than deliberate exploitation of undefined
1669 * optimization: the hash algorithm for strings has changed to one
1670 that is less vulnerable to spurious collisions. (thanks to Juho
1672 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1673 needless bounds checking. (thanks to Juho Snellman)
1674 * optimization: implemented multiplication as a modular
1675 (UNSIGNED-BYTE 32) operation on the PPC backend.
1676 * fixed some bugs revealed by Paul Dietz' test suite:
1677 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1679 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1680 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1681 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1683 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1685 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1687 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1688 their output stream on EOF from read.
1689 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1690 have been read to end-of-file.
1691 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1693 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1694 description of determination of which consecutive characters
1696 ** Printing the "Space" character with escaping on now yields "#\\ ",
1697 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1698 ** Reading floating-point numbers with *READ-BASE* set to a number
1699 less than 10 works correctly.
1700 ** Reading floating-point numbers with *READ-BASE* set to a number
1701 more than 10 works correctly.
1702 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1703 the readtable currently in effect.
1705 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1706 * minor incompatible change: parsing of namestrings on a physical
1707 (Unix) host has changed; numbers after the final #\. in a
1708 namestring are no longer interpreted as a version field. This is
1709 intented to be largely invisible to the user, except that the
1710 meaning of the namestring "*.*.*" has changed: it now refers to a
1711 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1712 should usually be replaced by
1713 (make-pathname :name :wild :type :wild :version :wild)
1714 with the added benefit that this is more likely to be portable.
1715 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1716 option to OPEN now signals an error if the file being opened
1717 exists; this may have an impact on existing code.
1718 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1719 signals received, on the PowerPC platforms (both Linux and
1720 Darwin). (thanks to Pierre Mai for pointing out the location of
1722 * several fixes on OS X: The system now builds and runs cleanly on
1723 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1724 (thanks to Brian Mastenbrook)
1725 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1726 second-resolution: integer multiples of 1/3600 between -24 and 24.
1727 (thanks to Vincent Arkesteijn)
1728 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1729 type when called with 1 argument; PEEK-CHAR checked type of
1730 PEEK-TYPE only after having read first character from a
1731 stream. (reported by Peter Graves)
1732 * bug fix: the garbage collector now has much better locality
1733 behaviour, and in particular no longer treats all memory as being
1734 exhausted when many small objects point to each other in a deeply
1736 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1737 recognized as being TYPEP their class.
1738 * bug fix: the PUSHNEW documentation string has been corrected.
1739 (thanks to Vincent Arkesteijn)
1740 * bug fix: defaulting of the value for the last of an atypically
1741 large number of multiple values being bound was not being
1742 performed correctly on the Alpha or PPC platforms
1743 * optimization: implemented multiplication as a modular
1744 (UNSIGNED-BYTE 32) operation on the x86 backend.
1745 * optimization: SEARCH on simple-base-strings can now be open-coded.
1746 (see also contrib/compiler-extras.lisp for inspiration for
1747 teaching the compiler about the Boyer-Moore algorithm).
1748 * value, returned by MAX (and MIN) called with several EQUALP, but
1749 not EQL, arguments now does not depend on compiler settings.
1750 * fixed some bugs revealed by Paul Dietz' test suite:
1751 ** in stack analysis liveness information is propagated from
1752 non-local entry points.
1753 ** pathwise CAST removing failed when the CAST node did not start
1755 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1756 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1758 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1759 host is already defined.
1760 ** RENAME-FILE works on streams instead of signalling an internal
1762 ** PEEK-CHAR uses the current readtable when determining whether
1763 or not a character is whitespace.
1764 ** MERGE-PATHNAMES handles the case when the pathname does not
1765 specify a name while the default-pathname specifies a version.
1766 ** Pathnames now stand a better chance of respecting print/read
1768 ** Attempting to use standardized file system operators with a
1769 pathname with invalid :DIRECTORY components signals a
1771 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1773 ** DIRECTORY on logical pathnames is more correct.
1774 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1775 signal a TYPE-ERROR if their argument is not a stream.
1776 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1777 designator argument does not designate a stream.
1778 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1779 examining the synonym.
1780 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1782 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1783 element-type larger than ([UN]SIGNED-BYTE 32).
1785 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1786 * When built with the :SB-FUTEX feature, threaded builds now take
1787 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1788 for faster/more reliable mutex and condition variable support.
1789 * Incompatible change (but one you probably shouldn't have been using
1790 anyway): the interface and code for arbitrating between multiple
1791 threads in the same user session has been redesigned.
1792 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1793 error when called without an explicit environment argument.
1794 (thanks to Nikodemus Siivola)
1795 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1796 confused about their position. (thanks to Adam Warner and Gerd
1798 * bug fix: Pretty printing backquoted forms with unquotations in the
1799 argument list position of various code constructs such as LAMBDA
1800 now works correctly. (reported by Paul Dietz)
1801 * bug fix: Pretty printing unquotations no longer loses all
1802 stream position information.
1803 * optimization: performance of string output streams is now less
1804 poor for multiple small sequence writes.
1805 * optimization: performance of CSUBTYPEP in the presence of complex
1806 expressions involving CONS and NOT many times has been improved.
1807 (reported by Paul Dietz)
1808 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1810 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1811 streams for the REPL, for the debugger, and so on. (thanks to
1813 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1814 * fixed some bugs revealed by Paul Dietz' test suite:
1815 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1816 is ignored if it is too small, rather than propagating through
1817 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1818 ** extremely complex negations of CONS types were not being
1819 sufficiently canonicalized, leading to inconsistencies in
1821 ** VALUES tranformer lost derived type.
1823 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1824 * fixed a bootstrapping bug: the build process no longer assumes
1825 that the various BOOLE-related constants have the same value in
1826 host and target lisps. (noted by Paul Dietz' test suite on an
1827 SBCL binary built from CLISP)
1828 * The system can now be dynamically linked on the MIPS platform,
1829 which enables dynamic loading of foreign code from Lisp. (thanks
1830 to Ralf Baechle for discussions on the MIPS ABI)
1831 * The system now records debugging information for its own source
1832 files in a filesystem-position-independent manner, relative to
1833 the "SYS" logical host.
1834 * fixed a compiler bug: MV-LET convertion did not check references
1835 to the "max args" entry point. (reported by Brian Downing)
1836 * tweaked disassembly notes to be less confident about proclaiming
1837 some instruction as an LRA. (thanks to Brian Downing)
1838 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1839 now each have their own history, command character, and other
1840 characteristics. (thanks to David Lichteblau)
1841 * fixed some bugs revealed by Paul Dietz' test suite:
1842 ** compiler failure in compiling LOGAND expressions including a
1844 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1845 shift greater than 32.
1846 ** FLUSH-DEST did not mark blocks for type check regeneration.
1847 ** HANDLER-CASE failed to accept declarations in handler clauses
1848 in some circumstances.
1850 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1851 * New code in contrib/sb-introspect (still probably not entirely
1852 stable yet) provides some support for smart Lisp development
1853 environments like SLIME.
1854 * The conditions signalled for errors occurring when loading .fasl
1855 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1856 in a way which should help ASDF recover gracefully.
1857 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1858 stuff has been cleaned up. If you code contrib/ stuff, this might
1859 affect you, and you can look at contrib/README, contrib/STANDARDS,
1860 and/or the 0.8.4.27 diff to check.
1861 * In full calls the compiler now does not generate checks for declared
1862 argument types for all arguments.
1863 * various threading fixes
1864 ** and some experimental patches which didn't make it into
1865 the main tree for this release, but which are shipped in
1866 contrib/experimental-thread.patch as a possible fix for some
1867 failures (deadlock, spinning...) in GC-intensive multithreaded
1869 * fixed PPC build problem (source code incompatibility of different
1870 library versions): added offsetof() hackery which attempts to divine
1871 where glibc maintainers put uc_mcontext today
1872 * fixed bug 282: compiler does not trust type assertions while passing
1873 arguments to a full call.
1874 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1875 &OPTIONAL VALUES type parameter.
1876 * fix bug 214: algorithm for noting rejected templates is now more
1877 similar to that of template seletion. (also reported by rydis on
1879 * fixed bug 141b: printing backquoted information readably and prettily
1880 inserts a space where necessary.
1881 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1882 (CHARACTER 10) are now reported as errors, rather than propagated
1883 as unknown types. (reported by piso on #lisp)
1884 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1885 with values NIL and :ERROR. (thanks to Milan Zamazal)
1886 * fixed bug 191c: CLOS now does proper keyword argument checking as
1887 described in CLHS 7.6.5 and 7.6.5.1.
1888 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1889 longer raise an error; further, using a list as a for-as-arithmetic
1890 counter now raises a meaningful error.
1891 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1892 now understood by sequence creation functions such as MAKE-SEQUENCE
1894 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1895 a STRING-INPUT-STREAM.
1896 * compiler enhancement: SIGNUM is now better able to derive the type
1898 * type declarations inside WITH-SLOTS are checked. (reported by
1900 * fixed some bugs revealed by Paul Dietz' test suite:
1901 ** incorrect optimization of TRUNCATE for a positive first
1902 argument and negative second.
1903 ** compiler failure in let-convertion during flushing dead code.
1904 ** compiler failure while deriving type of TRUNCATE on an
1905 interval, containing 0.
1906 ** ASH of a negative bignum by a negative bignum count now returns
1908 ** intersection of CONS types now canonicalizes properly, fixing
1909 inconsistencies in SUBTYPEP.
1911 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1912 * incompatible change: The --disable-debugger command line
1913 option now clobbers the debugger at a more fundamental
1914 level, by redefining #'INVOKE-DEBUGGER instead of by
1915 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1916 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1917 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1918 enter the debugger REPL and then suffer recursive errors
1919 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1920 while under the new system, BREAK in --disable-debugger mode
1921 terminates the system just as an unhandled error would.
1922 * fixed compiler performance when processing loops with a step >1;
1923 * bug fix: DOCUMENTATION now retrieves generic function
1924 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1925 support has been systematized, and now supports the methods
1926 specified by ANSI, along with a default method and a method for
1927 slot documentation. (reported by Nathan Froyd)
1928 * bug fix: effective methods associated with a generic function are
1929 no longer cached over a change of that generic function's method
1930 combination. (reported by Andreas Fuchs)
1931 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1932 names a non-existent directory. (thanks to Andreas Fuchs)
1933 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1934 the PPC platform, lead to stack corruption; this has been fixed.
1935 (reported by Rainer Joswig)
1936 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1937 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1938 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1939 arithmetic in general, is now much more reliable on the PPC
1941 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1942 platform now returns the right answer.
1943 * optimization: restored some effective method precomputation in
1944 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1945 precomputation is now tunable.
1946 * optimization: compiler-internal data structure use has been
1947 reviewed, and changes have been made that should improve the
1948 performance of the compiler by about 20%.
1949 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1950 simple-base-strings and simple-bit-vectors is improved.
1951 * optimization: the optimization of 32-bit logical and arithmetic
1952 functions introduced in version 0.8.3 on the x86 has been
1953 implemented on the mips, ppc and sparc platforms; an
1954 implementation of the same facility, but for 64-bit arithmetic,
1955 has been added for the alpha.
1956 * microoptimization: the compiler is better able to make use of the
1957 x86 LEA instruction for multiplication by constants.
1958 * bug fix: in some situations compiler did not report usage of
1959 generic arithmetic in (SPEED 3) policy.
1960 * bug 145b fix: compiler used wrong type specifier while converting
1961 MEMBER-types to numeric.
1962 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1964 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1965 index argument now works properly on the Alpha platform.
1966 * bug fix: floating point exception treatment on the Alpha platform
1968 * bug fix: FILE-POSITION works much better on string input and
1969 output streams. (thanks to Nikodemus Siivola)
1970 * bug fix: many threading/garbage collection symptoms sorted.
1971 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1972 might be pseudo-atomic.
1973 * internal change: Stopping for GC is now done with signals not
1974 ptrace. GC is now done in whichever thread wanted it, instead of
1976 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1978 * bug fix: result form in DO is not contained in the implicit
1980 * incompatible change: ICR structure is changed; the value part of
1981 CONTINUATION is now called LVAR; corresponding functions are
1982 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1983 * added type deriver for ISQRT (thanks to Robert E. Brown).
1984 * bug fix: better support for loading from the command line when an
1985 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1987 * fixed some bugs revealed by Paul Dietz' test suite:
1988 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1989 ** ROUND and FROUND now give the right answer when given very
1990 small float arguments.
1991 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1993 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1994 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1995 ** LCM should return a non-negative integer.
1996 ** PARSE-INTEGER returned the index of a terminator instead of the
1997 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1998 ** PARSE-INTEGER returned an incorrect index being applied to a
2000 ** LCM with two arguments of 0 returns 0 rather than signalling
2002 ** unsigned addition of a 32-bit constant with the high bit set no
2003 longer causes an internal compiler error on the x86.
2004 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2005 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2006 with negative last argument.
2007 ** byte specifiers with zero size and position no longer cause
2008 an error during type derivation.
2009 ** bignum multiplication on the Alpha platform now returns the
2011 * porting: The system now builds on SuSE AMD64, although it still
2012 generates a 32-bit binary.
2013 * .fasl file incompatibility: The fasl file version number has
2014 been incremented (because of the changes to internal compiler
2015 data structures referred to above).
2017 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2018 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2019 more accurately, on the Darwin kernel running on PowerPC hardware.
2020 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2021 * Compiler code deletion notes now signal a condition of type
2022 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2023 an associated MUFFLE-WARNING restart.
2024 * The compiler now performs limited argument count validation of
2025 constant format strings in FORMAT, and where appropriate in ERROR,
2026 CERROR and WARN. (thanks to Gerd Moellmann)
2027 * New ASDF-INSTALL contrib can be used for automatic download and
2028 installation of third-party Lisp code from CCLAN or other sites
2030 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2031 which forces another thread to execute a function supplied by the
2033 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2034 now accept and act upon their :ELEMENT-TYPE keyword argument.
2035 (reported by Martin Atzmueller, Edi Weitz)
2036 * bug fix: FILE-POSITION now accepts position designators up to
2037 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2038 is the greater. (thanks to Patrik Nordebo)
2039 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2040 Kalle Olavi Niemitalo)
2041 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2042 printable. (reported by Eric Marsden)
2043 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2044 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2045 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2046 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2047 (thanks to Stig Sandoe)
2048 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2049 UNREAD-CHAR now correctly includes the unread character in the
2050 target sequence. (thanks to Gerd Moellmann)
2051 * bug fix in threaded builds: the system can now be suspended and
2052 resumed by shell job control with minimal disruption.
2053 * bug fixes in times and timezones >2038AD
2054 * better handling of "where is GNU make?" problem in build scripts
2055 (thanks to Nikodemus Siivola)
2056 * new optimization: inside a named function any reference to a
2057 function with the same name is considered to be a self-reference;
2058 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2059 optimization quality.
2060 * new optimization on x86: logical functions and + now have
2061 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2062 used when the result is truncated to 32 bits.
2063 * VALUES declaration is partially enabled.
2064 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2065 * fixed some bugs revealed by Paul Dietz' test suite:
2066 ** The system now obeys the constraint imposed by
2067 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2068 types form a lattice under type intersection.
2069 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2070 ** ASSOC now ignores NIL elements in an alist.
2071 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2072 and (1+ MOST-POSITIVE-FIXNUM) answers.
2073 ** The addition of a method with invalid qualifiers to a generic
2074 function does not cause an error to be signalled immediately;
2075 a warning is signalled, and the error is generated only on
2076 calling the generic function.
2077 * changed .fasl file version number, in part to add type codes for
2078 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2079 obscure ANSI requirements
2081 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2082 * fixed bug 148: failure to inline-expand a local function left
2083 garbage, confusing the compiler.
2084 * fixed bugs 3cd: structure slot readers perform type check if the
2085 slot can have an invalid value (i.e. it is either not initialized
2086 or can be written with a less specific slot writer).
2087 * bug fix: the compiler now traps array references to elements off
2088 the end of an array; previously, the bounds checking in some
2089 circumstances could go off-by-one.
2090 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2092 * type declarations for array element types now obey the description
2093 on the CLHS page "Declaration TYPE", as per discussions on
2094 sbcl-help around 2003-05-08. This means that a declaration
2095 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2096 declaration, all references to BAR will be asserted or assumed
2097 (with THE, so dependent on compiler policy) to involve objects of
2098 type FOO. Note that no such declaration is implied in
2099 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2100 * declared types of functions from the "Conditions"
2101 chapter. (reported by Paul Dietz)
2102 * bug fix: CERROR accepts a function as its first argument.
2103 * bug fix: NTH an NTHCDR accept a bignum as index
2104 arguments. (reported by Adam Warner)
2105 * optimization: character compare routines now optimize comparing
2106 against a constant character. (reported by Gilbert Baumann)
2107 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2108 argument now works properly on the MIPS platform.
2109 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2110 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2111 declared to have a functional type, does not cause an error
2113 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2114 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2115 * fixed bug: if last continuation of a deleted block has a
2116 destination, this destination should be deleted too. (reported by
2118 * fixed a bug in the bootstrap process: the host compiler's values
2119 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2120 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2121 test case from Patrik Nordebo)
2122 * improved the ability of the disassembler on the PPC platform to
2123 provide helpful disassembly notes.
2124 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2125 the class in more cases than previously.
2126 * bug fix: FILE-POSITION now understands :START and :END for
2127 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2128 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2129 8) data. (thanks to David Lichteblau)
2130 * bug fix: it is possible to add a method to a generic function
2131 without lambda list.
2132 * bug fix: reader failed to signal END-OF-FILE inside an
2133 object representation. (reported by Nikodemus Siivola)
2134 * fixed some bugs revealed by Paul Dietz' test suite:
2135 ** LAST and [N]BUTLAST should accept a bignum.
2136 ** condition slot accessors are methods.
2137 ** (VECTOR NIL) is a subtype of STRING.
2139 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2140 * minor incompatible change: some nonsensical specialized lambda
2141 lists (used in DEFMETHOD) which were previously ignored now signal
2143 * minor incompatible change: the system is now aware of the types of
2144 variables in the COMMON-LISP package, and will signal errors for
2145 most violations of these type constraints (where previously they
2146 were silently accepted).
2147 * minor incompatible change: COMPILE-FILE now uses the freedom
2148 afforded (ANSI 3.2.2.3) to use derived function types for
2149 functions defined in the same file. This also permits the system
2150 to warn on static type mismatches and function
2151 redefinition. (Currently it does not work with high DEBUG level.)
2152 * minor incompatible change: VALUES declaration is disabled.
2153 * When issuing notes, the compiler now signals a condition of type
2154 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2155 restart for use in user handlers. It is expected that the
2156 COMPILER-NOTE condition will eventually become a condition
2157 supertype to a hierarchy of note types, which will then be
2158 handleable in a similar fashion. However, at the moment, no such
2159 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2160 but it's an implementation detail, not a classification for the
2162 * Changes in type checking closed the following bugs:
2163 ** type checking of unused values (192b, 194d, 203);
2164 ** template selection based on unsafe type assertions (192c, 236);
2165 ** type checking in branches (194bc).
2166 * A short form of VALUES type specifier has ANSI meaning (it has
2167 increased the number of situations when SBCL cannot perform type
2169 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2170 keywords or constants is permissible.
2171 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2172 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2173 classes. (thanks to Antonio Martinez)
2174 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2175 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2176 argument to a pretty-print stream. Instead, it leaves any such
2177 operations to DESCRIBE-OBJECT methods.
2178 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2180 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2181 lambda lists. (thanks to David Lichteblau)
2182 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2183 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2184 (thanks to Gerd Moellmann)
2185 * fixed bug 246: increased compilation speed of long
2186 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2188 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2190 * DEFCONSTANT now throws a condition of type
2191 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2192 constant to a non-EQL value; CONTINUE and ABORT restarts
2193 respectively change and preserve the value.
2194 * fixed bug 63: The code walker, part of the implementation of CLOS,
2195 is now better at handling symbol macros.
2196 * bug fix: There is no longer an internal implementation type named
2197 CL:LENGTH. (reported by Raymond Toy)
2198 * bug fix: In macro-like defining macros/special operators the
2199 implicit block does not enclose the lambda list.
2200 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2201 enclosed in parentheses) are not suitable as type specifiers, and
2202 their use properly signals an error now.
2203 * bug fix: An argument count mismatch for a type specifier in code
2204 being compiled no longer causes an unhandled error at compile
2205 time, but signals a compile-time warning.
2206 * fixed simple vector readable printing
2207 * bug fix: DESCRIBE takes more care over whether the class
2208 precedence list slot of a class is bound before accessing it.
2209 (reported by Markus Krummenacker)
2210 * bug fix: FORMATTER can successfully compile pretty-printer format
2211 strings which use variants of the ~* directive inside.
2212 * bug fix: SEARCH now applies its TEST predicate to the elements of
2213 the arguments in the correct order. (thanks to Wolfhard Buss)
2214 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2215 in type inference. (thanks to Robert E. Brown)
2216 * bug fix: Reading in symbols with an explicit package name of ""
2217 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2218 * fixed some bugs revealed by Paul Dietz' test suite:
2219 ** NIL is now allowed as a structure slot name.
2220 ** Arbitrary numbers, not just REALs, are allowed in certain
2221 circumstances in LOOP for-as-arithmetic clauses.
2222 ** Multiple class redefinitions before slot access no longer
2223 causes a type error.
2224 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2225 association between the name and a class.
2226 ** Generic functions with non-standard method-combination and over
2227 five methods all of which return constants no longer return NIL
2228 after the first few invocations. (thanks to Gerd Moellmann)
2229 ** CALL-NEXT-METHOD with no arguments now passes the original
2230 values of the arguments, even in the presence of assignment.
2231 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2233 ** DISASSEMBLE works with closures and funcallable instances.
2234 ** ADD-METHOD now returns the generic function, not the new method.
2235 ** FIND-METHOD signals an error if the lengths of the specializers
2236 is incompatible with the generic function, even if the ERRORP
2238 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2239 which its argument is a member.
2240 ** DEFCLASS only redefines the class named by its class-name
2241 argument if that name is the proper name of the class;
2242 otherwise, it creates a new class.
2243 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2244 of the UNBOUND-SLOT condition to the name of the slot.
2245 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2246 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2247 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2248 SLOT-MAKUNBOUND in the specified fashion.
2250 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2251 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2252 cross-compilation host. As a consequence, we can now bootstrap our
2253 way up to SBCL starting with a bare gcc toolchain and human-readable
2254 source code (first the source to CLISP, then the source to SBCL).
2255 * A contributed module containing a partial implementation of the
2256 simple-streams interface has been included. (thanks to Rudi
2258 * A contributed module implementing the RFC1321 Message Digest
2259 Algorithm, known as MD5, has been included.
2260 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2261 no longer has any effect, as the code controlled by this feature
2262 has been deleted. (As far as we know, no-one has ever built using
2263 this feature, and its semantics were confused in any case).
2264 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2265 work on conditions (as required by the ANSI specification),
2266 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2267 expected behaviour on conditions. Users should note, however,
2268 that such behaviour is not required by the ANSI specification,
2269 and so use of this behaviour may render their code unportable.
2270 * fixed some bugs revealed by Paul Dietz' test suite:
2271 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2273 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2274 ** SXHASH on condition objects no longer returns NIL.
2275 ** :ALLOCATION :CLASS slots are better treated; their values are
2276 updated on class redefinition, and initforms inherited from
2277 superclasses are applied.
2278 ** REMOVE-METHOD returns its generic function argument even when
2279 no method was removed.
2280 ** SHARED-INITIALIZE now initializes the values of the requested
2281 slots, including those with :ALLOCATION :CLASS.
2282 ** ALLOCATE-INSTANCE now works on structure classes defined via
2283 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2285 ** SLOT-EXISTS-P now works on conditions, as well as structures
2287 ** MAKE-LOAD-FORM now has the required methods on
2288 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2289 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2290 keyword, and now implements the SLOT-NAMES argument.
2291 ** methods with &OPTIONAL arguments no longer allow too many
2292 arguments to be passed in the call without error.
2293 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2294 option is consistent with the required arguments of the generic
2295 function lambda list.
2296 * bug fix: REQUIRE accepts a string designator. (Thanks to
2298 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2299 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2300 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2301 * bug fix: The compiler now checks for duplicated variables in macro
2303 * bug fix: SETQ on globals returns the correct value.
2304 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2305 not a COMPILER-ERROR (followed by some other strange error on
2306 choosing the CONTINUE restart).
2307 * bug fix: make.sh and friends are now more consistent in the way that
2308 they look for GNU "make".
2310 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2311 * experimental native threads support (on x86 Linux >=2.4 only).
2312 This is not compiled in by default: you need to add :SB-THREAD to
2313 the target features. See the "Beyond ANSI" chapter of the manual
2315 * fix for longstanding nonANSIism: The old distinction between
2316 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2317 The return value from CL:FIND-CLASS is now a CLOS class, and
2318 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2319 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2320 * An interface to the MetaObject Protocol, as described in Kiczales,
2321 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2322 MIT Press, 1991, is available from the SB-MOP package.
2323 * incompatible change: the SB-PCL package should now be considered
2324 a private implementation detail, and no longer a semi-private MOP
2326 * minor incompatible change: due to rearrangement for threads, the
2327 control stack and binding stack are now allocated at arbitrary
2328 addresses instead of being hardcoded per-port. Users affected by
2329 this probably have to be doing advanced things with shared
2330 libraries, and will know who they are.
2331 * minor incompatible change: Previously, all --eval forms used were
2332 processed with READ before any of them were processed with EVAL.
2333 Now each --eval form is processed with both READ and EVAL before
2334 the next --eval form is processed. (Thus package operations like
2335 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2336 work as the user might reasonably expect.)
2337 * minor incompatible change: *STANDARD-INPUT* is now only an
2338 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2340 * minor incompatible change: Y-OR-N-P is now character-oriented,
2341 not line oriented. Also, YES-OR-NO-P now works without errors.
2342 (thanks to Antonio Martinez)
2343 * sb-aclrepl module improvements: an integrated inspector, added
2344 repl features, and a bug fix to :trace command.
2345 * Known functions, which cannot be open coded by the backend, are
2346 considered to be able to check types of their arguments. (fixing
2347 a bug report by Nathan J. Froyd)
2348 * fixed a bug in computing method discriminating functions: It is
2349 now possible to define methods specialized on classes which have
2350 forward-referenced superclasses. (thanks to Gerd Moellmann)
2351 * fixed evaluation order in optional entries (reported by Gilbert
2353 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2354 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2355 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2356 required-by-AMOP NAME argument, as well as CLASS and
2357 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2358 * fixed bug 20: DEFMETHOD can define methods using names that are
2359 not the proper names of classes to designate class specializers.
2360 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2361 Unix file descriptors, instead of blowing up. (thanks to Antonio
2363 * Garbage collection refactoring: user-visible change is that a
2364 call to the GC function during WITHOUT-GCING will not do garbage
2365 collection until the end of the WITHOUT-GCING. If you were doing
2366 this you were probably losing anyway.
2367 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2368 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2369 TYPEP the latter but not the former.
2370 * The compiler issues a full WARNING for calls to undefined functions
2371 with names from the CL package.
2372 * MAP-INTO for a vector destination is open coded. (reported by
2373 Brian Downing on c.l.l)
2374 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2375 documentation string.
2376 * fixed some bugs revealed by Paul Dietz' test suite:
2377 ** COPY-ALIST now signals an error if its argument is a dotted
2379 ** Condition slots are now accessed more correctly in the presence
2380 of multiple initargs for a given slot.
2381 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2382 exclude restarts of the same name associated with a different
2384 ** DEFCLASS of forward-referenced classes with another
2385 forward-referenced class in the superclasses list no longer
2387 ** Condition slots are now initialized once each, not multiple
2388 times. (thanks to Gerd Moellmann)
2389 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2391 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2392 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2393 arguments contain duplicated elements.
2394 ** RESTART-CASE understands local macros.
2395 ** RESTART-CASE associates exactly its own restarts with a condition.
2396 ** ENDP in safe mode checks its argument to be of type LIST.
2397 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2398 in question is unbound.
2399 ** Optimized MAKE-INSTANCE functions no longer cause internal
2400 assertion failures in the presence of duplicate initargs.
2401 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2402 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2403 propagated even in the ctor optimized implementation of
2405 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2407 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2408 * a better implementation of SXHASH on (simple) bit vectors,
2409 measured both in execution speed and in distribution of results
2410 over the positive fixnums, has been installed. Likewise, a better
2411 implementation of EQUAL for simple bit vectors is now available.
2412 * fixed CEILING optimization for a divisor of form 2^k.
2413 * fixed bug 240 (emitting extra style warnings "using the lexical
2414 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2415 by Antonio Martinez)
2416 * fixed SXHASH, giving different results for NIL depending on type
2417 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2418 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2419 variable. (found by Rolf Wester)
2420 * SBCL does not ignore type declarations for special
2421 variables. (reported by rif on c.l.l 2003-03-05)
2422 * some bug fixes in contrib/sb-aclrepl/
2423 * fixed some bugs revealed by Paul Dietz' test suite:
2424 ** a bug in the CONS type specifier, whereby the CAR and CDR
2425 types got intertwined, has been fixed;
2426 ** the type system is now able to reason about the interaction
2427 between INTEGER and RATIO types more completely;
2428 ** APPEND, [N]REVERSE and NRECONC check that those their
2429 arguments, which must be proper lists, are really so;
2430 ** An array specialized to be unable to hold elements has been
2431 implemented, as required -- yes, really -- by ANSI;
2432 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2433 on malformed property lists;
2435 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2436 * incompatible packaging change: in line with Unix convention,
2437 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2438 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2439 * REQUIRE and PROVIDE are now optionally capable of doing something
2440 useful. See the documentation string for REQUIRE.
2441 * infrastructure for a managed SBCL contrib system: contributed
2442 modules in this release include:
2443 ** the ASDF system definition facility;
2444 ** an interface to the BSD Sockets API;
2445 ** an ACL-like convenience interface to the repl;
2446 (thanks to Kevin Rosenberg)
2447 ** an implementation of ROTATE-BYTE, with efficient implementation
2449 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2450 gives a closer approximation to the right answer than previously.
2451 (thanks to Raymond Toy)
2452 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2453 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2454 optional environment arguments, as required by ANSI.
2455 * fixed bugs in other functions taking environment objects, allowing
2456 calls with an explicit NIL environment argument to be compiled
2458 * fixed bug 228: primary return values from
2459 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2460 COMPILE or FUNCTION.
2461 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2462 :TYPE will no longer signal errors on innocuous objects.
2463 * fixed bug 231b: SETQ is better at respecting type declarations in
2464 the lexical environment.
2465 * fixed a bug in DEFCLASS: classes named by symbols with no or
2466 unprintable packages can now be defined.
2467 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2468 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2469 * fixed compiler failure related to checking types of functions.
2470 (reported by Robert E. Brown)
2471 * the compiler is now much more consistent in its error-checking
2472 treatment of bounding index arguments to sequence functions: in
2473 (SAFETY 3) code, errors will be signalled in almost all cases if
2474 invalid sequence bounding indices are passed to functions defined
2475 by ANSI to operate on sequences.
2476 * fixed a bug in the build procedure: documentation of SBCL-specific
2477 packages is now preserved and available in the final Lisp image.
2478 * lifted FDEFINITION lookup out of loops in the implementation of
2479 many list operations. (thanks to Robert E. Brown)
2480 * fixed a bug in the reader: the #n# reader macro now works for
2481 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2482 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2483 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2484 E. Brown and Tony Martinez respectively)
2485 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2486 SLOT-BOUNDP: in particular, it is now able to optimize them much
2487 better, and is now not vulnerable to having packages renamed.
2488 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2489 always called when a slot is not present in an instance. (thanks
2491 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2492 longer signal an error on unreading a character following EOF on
2493 the previous constituent stream. (thanks to Tony Martinez)
2494 * fixed some bugs revealed by Paul Dietz' test suite:
2495 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2496 not just nonnegative fixnums;
2497 ** the logical bit-array operators such as BIT-AND now accept an
2498 explicit NIL for their "opt-arg" argument (to indicate a
2499 freshly-consed result bit-array);
2500 ** ELT now signals an error on an invalid sequence index in safe
2502 ** the type system is now cleverer about negations of numeric
2503 types, and consequently understands the BIGNUM and RATIO types
2505 ** the type system is now cleverer about the interaction between
2506 INTEGER and RATIO types: while bugs still remain, many more
2507 cases are accurately computed;
2508 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2509 if it is in the last clause;
2510 ** CONSTANTLY now correctly returns a side-effect-free function in
2512 ** DECLARE is no longer treated as a special-operator; in
2513 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2514 * incremented fasl file version number due to the change in the
2515 DEFSTRUCT-SLOT-DESCRIPTION structure.
2517 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2518 * minor incompatible change: code processed by the "interpreter" or
2519 EVAL now has a compilation optimization policy of (DEBUG 2)
2520 (changed from (DEBUG 1)) to improve debuggability of interactive
2521 development, and to allow the use of the debug RETURN command in
2523 * an experimental implementation of the RETURN command for the
2524 debugger has been included. (thanks to Frederik Kuivinen)
2525 * fixed bug 62: constraints were not propagated into a loop.
2526 * fixed bug in embedded calls of SORT (reported and investigated by
2528 * fixed some bugs revealed by Paul Dietz' test suite:
2529 ** printing and reading of arrays with some dimensions having
2530 length 0 (thanks to Gerd Moellmann);
2531 ** BOA constructor with &AUX argument without a default value does
2532 not cause a type error;
2533 ** CONSTANTP now returns true for all self-evaluating objects.
2535 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2536 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2537 accessors that are related by inheritance, as specified in the
2538 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2539 Valtteri Vuorikoski)
2540 * The compiler is now able to inline functions that were defined in
2541 a complex lexical environment (e.g. inside a MACROLET).
2542 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2544 * Support for the upcoming FreeBSD-5.0 release has been included.
2545 (thanks to Dag-Erling Smorgrav)
2546 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2547 effect when it is not in a toplevel context.
2548 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2549 stand a better chance of being correct. (thanks to Gerd
2551 * fixed bug in COERCE, which now signals an error on coercing a
2552 rational to a bounded real type which excludes the expected
2554 * The compiler is now able to derive types more accurately from the
2555 COERCE and COMPILE functions.
2556 * fixed bug 223: functional binding is considered to be constant
2557 only for symbols in the CL package.
2558 * fixed bug 231: SETQ did not check the type of a variable being set
2559 (reported by Robert E. Brown)
2560 * A new optimization for MAKE-INSTANCE has been included, fixing
2561 various bugs (including relating to :ALLOCATION :CLASS slots and
2562 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2564 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2565 ** As required by ANSI, LOOP now disallows anonymous collection
2566 clauses such as COLLECT I in conjunction with aggregate boolean
2567 clauses such as THEREIS (= I 1);
2568 ** LOOP now signals an error when any variable is reused in the
2569 same loop (including the potentially useful construct analogous
2570 to WITH A = 1 WITH A = (1+ A);
2571 ** IT is only a special loop symbol within the first clause of a
2572 conditional loop clause;
2573 ** LOOP with a typed iteration variable over a hashtable now
2574 signals a type error iff it should.
2575 * fixed some other bugs revealed by Paul Dietz' test suite:
2576 ** FILE-STREAM now names the class previously known as FD-STREAM;
2577 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2578 argument) no longer signals an error;
2579 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2580 of the slot symbol, rather than using the current package
2581 ((:CONC-NAME "") continues to intern the slot's name in the
2583 * incremented fasl file version number, because of the incompatible
2584 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2585 of the new implementation of DEFINE-COMPILER-MACRO.
2587 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2588 * Support for building SBCL for MIPS platforms running in
2589 little-endian mode has now been checked in, and basic
2590 functionality on said platforms verified.
2591 * minor incompatible change: PCL now records the pathname of a file
2592 in which methods and the like are defined, rather than its
2594 * minor incompatible change: TRUENAME now considers the truename of
2595 a file naming a directory to be the pathname with :DIRECTORY
2596 component indicating that directory.
2597 * minor incompatible change: a NAMED clause in the extended form of
2598 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2599 reason for the previous behaviour is unclear.
2600 * more systematization and improvement of CLOS and MOP conformance
2601 in PCL (thanks to Gerd Moellman and Pierre Mai):
2602 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2604 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2605 primary methods with no specializers;
2606 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2608 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2609 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2610 FORWARD-REFERENCED-CLASSes; error reporting on
2611 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2613 ** SXHASH on CLOS instances now uses a slot internal to the
2614 instance to return different numbers on distinct instances,
2615 while preserving the same return value through invocations of
2617 ** DEFMETHOD signals errors when methods with longer incongruent
2618 lambda lists are added to generic functions;
2619 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2620 CLASS, as specified in AMOP;
2621 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2622 on the order returned by the primary method for classes of
2623 class STANDARD-CLASS;
2624 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2625 * fixed some bugs shown by Paul Dietz' test suite:
2626 ** DOLIST puts its body in TAGBODY;
2627 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2629 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2630 value producing form;
2631 ** if more variables are given to PROGV than values, extra
2632 variables are bound and made to have no value;
2633 ** NSUBSTITUTE on list arguments gets the right answer with
2635 ** ELT signals an error of type TYPE-ERROR when the index argument
2636 is not a valid sequence index;
2637 ** LOOP signals (at macroexpansion time) an error of type
2638 PROGRAM-ERROR when duplicate variable names are found;
2639 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2640 ** FUNCALL of special-operators now cause an error of type
2642 ** PSETQ now works as required in the presence of side-effecting
2643 symbol-macro places;
2644 ** NCONC accepts any object as its last argument;
2645 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2647 ** loop-for-as-package does not require a package to be explicitely
2649 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2651 * fixed bug 166: compiler preserves "there is a way to go"
2652 invariant when deleting code.
2653 * fixed bug 172: macro lambda lists with required arguments after
2654 &REST arguments now cause an error to be signalled. (thanks to
2656 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2657 bug. (thanks to Matthew Danish)
2658 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2660 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2661 arguments are lexically rebound. (thanks to Gerd Moellmann and
2663 * fixed bug 194: error messages are now more informative when there
2664 is no primary method applicable in a call to a generic
2665 function. (thanks to Gerd Moellmann)
2666 * fixed bug in command line argument checking (thanks to Julian
2668 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2669 (thanks to Matthew Danish)
2670 * incremented fasl file version number, because of the
2671 SXHASH-related changes in the layout of CLOS data structures
2673 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2674 * minor incompatible change: The runtime (the Unix executable named
2675 "sbcl") is now much pickier about the .core files it will load.
2676 Essentially it now requires .core files to descend from the same
2677 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2678 as the runtime does. (The intent is to prevent the crashes which
2679 can occur, and which can even be reported as mysterious failures,
2680 when people patch the sources or change the build parameters
2681 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2682 sbcl and .core files.)
2683 * fixed bug: VALUES-LIST is no longer optimized away.
2684 * fixed bug 142: The FFI conversion of C string values to Lisp
2685 string values no longer conses excessively. (thanks to Nathan
2686 Froyd porting Raymond Toy's fix to CMU CL)
2687 * began to systematize and improve MOP conformance in PCL (thanks to
2688 Nathan Froyd, Gerd Moellman and Pierre Mai):
2689 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2691 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2692 ** FINALIZE-INHERITANCE is now called on class finalization;
2693 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2694 argument precedence order.
2695 * fixed bug 202: The compiler no longer fails on functions whose
2696 derived types contradict their declared type.
2697 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2698 so it can be non-toplevel.
2699 * The fasl file version number has changed (because of the new
2700 implementation of DEFMACRO).
2701 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2702 safe code, that any length requirement by their type-specifier
2703 argument is valid. The exceptions to this are described in bug
2705 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2706 functions now check the types of their inputs as required by ANSI.
2707 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2708 introduced symbol is DECLAREd to be SPECIAL.
2709 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2710 * fixed bug: :COUNT argument to sequence functions may be negative
2711 * fixed bug: body of DO-SYMBOLS may contain declarations
2712 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2713 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2714 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2715 symbol macro only once
2716 * fixed printing of call frame when argument list is unavailable
2717 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2718 * compiler no longer signals WARNING on unknown keyword
2721 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2722 * A beta-quality port to the mips architecture running Linux,
2723 based on the old CMUCL backend, has been made. It has been tested
2724 on a big-endian kernel, and works sufficiently well to be able to
2725 rebuild itself; it has not been tested in little-endian mode.
2726 * fixed an inconsistency between gencgc.c and purify.c which made
2727 dumping/loading .core files unreliable
2728 * fixed bug 120a: The compiler now deals correctly with IFs where
2729 the consequent is the same as the alternative, instead of
2730 misderiving the return type. (thanks to Alexey Dejneka)
2731 * fixed bug 113: Logical pathnames are now dumpable (the logical
2732 host is resolved at load-time, throwing an error if it is not
2734 * fixed bug 174: FORMAT's error message is slightly clearer when a
2735 non-printing character is used in a format directive.
2736 * fixed several bugs in compiler checking of type declarations, i.e.
2737 violations of the Python "declarations are assertions" principle
2738 (thanks to Alexey Dejneka)
2739 * fixed several bugs in PCL's error checking (thanks to Gerd
2741 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2743 * fixed bug in compilation of functions as first class values
2744 (thanks to Antonio Martinez-Shotton)
2745 * The compiler's handling TYPE-ERRORs which it can prove will
2746 inevitably happen at runtime has been cleaned up and corrected
2747 in several ways. (thanks to Alexey Dejneka)
2748 * improved argument type checking for various basic arithmetic
2749 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2750 lavished on them in the past that they can be compiled in many
2751 ways in different special cases
2752 * fixed bug 181: compiler checks validity of user supplied type
2754 * cleaned up code flushing in optimization: Function calls which
2755 should signal errors for safety purposes (e.g. which ANSI says
2756 should signal errors when their arguments are of incorrect type)
2757 are no longer optimized away.
2758 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2759 * incremented fasl file version number, because changes in the
2760 implementation of sequence functions like COERCE caused
2761 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2762 in old inline expansions) to become undefined. (Actually these
2763 changes were later undone, so we might very well be binary
2764 compatible with 0.7.7 after all, but leaving the version number
2765 incremented seemed like the simplest and most conservative
2768 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2769 * An alpha-quality port to the parisc architecture running Linux,
2770 based on the old CMUCL backend, has been made. This, even more so
2771 than the other backends, should be considered still a work in
2772 progress; known problems include that the Linux kernel in 64-bit
2773 mode does not propagate the correct sigcontext structure to
2774 userspace, and consequently SBCL on a parisc64 kernel will not
2776 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2777 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2778 suppressed.) Also now that the compiler is looking at declarations
2779 in the environment, it checks optimization declarations as well,
2780 and suppresses inlining when (> DEBUG SPEED).
2781 * More fixes have been made to treatment of floating point exception
2782 treatment and other Unix signals. In particular, floating point
2783 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2784 * The detection and handling of control stack exhaustion (infinite
2785 or very deeply nested recursion) has changed. Stack exhaustion
2786 detection is now done by write-protecting pages at the OS level
2787 and applies at all optimization settings; when found, a
2788 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2789 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2790 be caught using IGNORE-ERRORS.
2791 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2792 correctly when transforming recursive LABELS functions to LETs.
2793 (thanks to Alexey Dejneka)
2794 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2795 that are names of constants or global variables.
2796 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2797 alien routines with docstrings.
2798 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2799 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2801 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2802 (thanks to Eric Marsden)
2803 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2804 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2805 * Bug fix: The SPARC backend can now compile functions involving
2806 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2807 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2808 to a functions expecting a PATHNAME-DESIGNATOR.
2809 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2810 lambda lists. (thanks to Alexey Dejneka)
2811 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2812 functions so that EXPORTing the name of the class doesn't cause
2813 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2814 to Antonio Martinez for reporting this)
2815 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2816 constant-folded without failing an assertion. (thanks to Einar
2817 Floystad Dorum for reporting this)
2818 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2819 types are now dealt with more correctly by the compiler.
2820 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2821 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2822 behaviour analogous to RENAME-FILE. This puts its behaviour more
2823 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2825 * The fasl file version number has changed again. (because of the
2826 bug fix involving the names of PCL MAKE-INSTANCE functions)
2828 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2829 * bug fix: Floating point exceptions are treated much more
2830 consistently on the x86/Linux and PPC/Linux platforms.
2831 * Array initialization with :INITIAL-ELEMENT is now much faster for
2832 cases when the compiler cannot open code the array creation, but
2833 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2834 array accesses have also seen a speed increase.
2835 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2836 specified. (This was at the root of some bad interactions between
2837 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2839 * bug fix: Internal error arguments for undefined functions are now
2840 computed correctly on the PPC/Linux platform.
2841 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2842 Raymond Toy's patch for CMU CL)
2843 * Support for the Solaris 9 operating environment has been included
2844 (thanks to Daniel Merritt)
2845 * A very ugly but hopefully complete draft of the missing FFI chapter
2846 of the manual has been created by reformatting the corresponding
2847 CMU CL manual chapter into (currently very ugly and incoherent)
2848 DocBook and bringing it up to date for SBCL behavior. Thus, the
2849 manual is now essentially complete, at least by my extreme
2850 once-and-only-once standards, whereby it's acceptable to refer to
2851 the doc strings of SB-EXT functions as the primary documentation.
2852 * The fasl file version number has changed again, due to cleanup of
2853 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2854 is no longer a static symbol.)
2856 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2857 * SBCL now builds with OpenMCL (version 0.12) as the
2858 cross-compilation host; also, more progress has been made toward
2859 bootstrapping under CLISP.
2860 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2862 * bug 158 fixed: The compiler can now deal with integer loop
2863 increments different from 1; fixing this turned out also to fix
2865 * bug 169 fixed: no more bogus warnings about using lexical bindings
2866 despite the presence of perfectly good SPECIAL declarations (thanks
2867 to David Lichteblau)
2868 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2869 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2870 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2872 * bug 180 fixed: Method combination specifications no longer ignore
2873 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2874 * bug fix: Structure type predicate functions now check their argument
2875 count as they should.
2876 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2877 correctly. (thanks to Pierre Mai)
2878 * minor incompatible change: The --noprogrammer option is deprecated
2879 in favor of the new --disable-debugger option, which is very similar.
2880 (The major difference is that it takes effect at a slightly different
2881 time at startup, causing handling of errors in --sysinit and
2882 --userinit files will be affected differently.) The
2883 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2884 been added to allow this functionality to be controlled from ordinary
2885 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2886 maintainers, who might want to run non-interactive scripts to
2887 build SBCL cores which will later be used interactively.)
2888 * minor incompatible change: The LOAD function no longer, when given
2889 a wild pathname to load, loads all files matching that pathname.
2890 Instead, an error of type FILE-ERROR is signalled.
2892 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2893 * bug 147 fixed: The compiler preserves its block link/count
2894 invariants more correctly now so that it doesn't crash. (thanks
2896 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2898 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2899 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2900 no-value-for-this-character as NIL (as specified by ANSI).
2901 * HOST-NAMESTRING on physical pathnames now returns a string that is
2902 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2903 (thanks to Christophe Rhodes)
2904 * The Alpha port handles icache flushing more correctly. (thanks to
2906 * More progress has been made toward bootstrapping under CLISP. (thanks
2907 to Christophe Rhodes)
2908 * The fasl file format has changed again, because dynamic loading
2909 on OpenBSD (which has non-ELF object files) motivated some cleanups
2910 in the way that foreign symbols are transformed and passed around.
2911 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2912 is no longer treated as whitespace by the reader, but instead as
2913 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2914 |A
\7fB|, instead of A as it used to.
2916 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2917 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2918 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2919 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2920 this as of 0.7.1.45, but was left out of the previous news section
2921 (thanks to Dan Barlow)
2922 * SBCL now runs on the Solaris operating system on SPARC architectures
2923 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2924 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2925 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2926 * SPARC backend cleanups, allowing builds of cores optimized for V8
2927 and V9 SPARCS, and also emission of code targeted to a particular
2928 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2930 * SBCL is closer to bootstrapping under CLISP, thanks to various
2931 fixes by Christophe Rhodes.
2932 * The fasl file format has changed again, to allow the compiler's
2933 INFO database to support symbol macros.
2934 * The user manual (in doc/) is formatted into HTML more nicely.
2935 (thanks to coreythomas)
2936 * The system is smarter about SUBTYPEP relationships, especially
2937 those involving NOT types (including types such as ATOM which are
2938 represented internally using NOT types). Thus SUBTYPEP is less
2939 likely to return (VALUES NIL NIL) in general, and in particular
2940 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2942 * The fasl file format has changed again, because the internal
2943 representation of types now includes a new slot to support the new
2944 SUBTYPEP-of-NOT-types logic.
2945 * (not a change in the main branch of SBCL, but a related prototype
2946 which can hopefully be merged into the main branch of SBCL in the
2947 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2948 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2949 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2950 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2951 dispatch macro character combinations. (thanks to Alexey Dejneka)
2952 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2953 that superficially look like logical namestrings correctly.
2954 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2956 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2957 default supplied-p) for &optional and &key arguments. (thanks to
2960 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2961 * incompatible change: The compiler is now less aggressive about
2962 tail call optimization, doing it only when (> SPACE DEBUG) or
2963 (> SPEED DEBUG). (This is an incompatible change because there are
2964 programs which relied on the old CMU-CL-style behavior to optimize
2965 away their unbounded recursion which will now die of stack overflow.)
2966 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2967 for non-GENCGC systems has been increased to 20M (since that
2968 seems much closer to the likely performance optimum for modern
2969 systems than the old 4M value was)
2970 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2971 and *DEBUG-PRINT-LEVEL*
2972 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2973 of CMU CL's support for SPARC, and various endianness and other
2974 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2975 * new syntactic sugar for the Unix command line: --load foo.bar is now
2976 an alternate notation for --eval '(load "foo.bar")'.
2978 ** The system now detects stack overflow and handles it gracefully,
2979 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2980 optimization settings. (This is a good thing in general, and
2981 its introduction in this version should be particularly timely
2982 for anyone whose code fails because of suppression of tail
2984 ** The system now hunts for the C variable "environ" in a more
2985 devious way, to avoid segfaults when the C library version differs
2986 between compile time and run time. (thanks to Christophe Rhodes)
2987 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2988 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2989 ** The compiler no longer issues bogus style warnings for undefined
2990 classes in the same source file as the DEFCLASSes which defined
2991 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2993 ** fixes in CONDITION class precedence list for undefined function
2994 errors (thanks to Alexei Dejneka)
2995 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2996 correctly. (thanks to Dan Barlow)
2997 ** portability fixes aiming at bootstrapping under CLISP (thanks
2998 to Dave McDonald and Christophe Rhodes)
2999 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3000 ** fixes in type translation and and type inference (thanks to
3002 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3003 ** various fixes in the new ports (thanks to Dan Barlow)
3004 * several changes related to debugging:
3005 ** suppression of tail recursion, as noted above
3006 ** stack overflow detection, as noted above
3007 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3008 is now the default. (For some time encapsulation has been more
3009 reliable than the breakpoint-based :ENCAPSULATE NIL
3010 implementation, at least on X86 systems; and I just noticed that
3011 encapsulation also seems closer to the spirit of the ANSI
3014 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3016 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3017 up properly again. (There was a packaging bug in 0.7.0 which
3018 left their definitions in SB-SYS::LOAD-FOREIGN and
3019 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3020 vital for most things which interface to C-level interfaces,
3021 like extensions working with sockets or databases or
3022 Perl-compatible regexes or whatever, and the need to fix
3023 this bug is the main reason that 0.7.1 was released so
3025 ** DEFGENERIC is now choosier about the methods it redefines, so that
3026 reLOADing a previously-LOADed file containing DEFGENERICs does
3027 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3028 cycle now works as it should. (thanks to Alexey Dejneka)
3029 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3030 fixed by Christophe Rhodes. (He actually submitted this patch
3031 months ago, and I delayed until after 0.7.0.)
3032 ** Bug 111 (internal compiler confusion about runtime checks on
3033 FUNCTION types) was fixed by Alexey Dejneka.
3034 * Some internal cleanups (getting rid of variables which aren't
3035 needed now that the byte interpreter is gone) caused the fasl
3036 file format number to change again.
3038 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3039 * major incompatible change: The default fasl file extension, i.e. the
3040 default extension for files produced by COMPILE-FILE, has changed
3041 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3043 ** There are many changes in the implementation of the compiler.
3044 SBCL is now essentially a compiler-only implementation of ANSI
3045 Common Lisp. EVAL still "interprets" a few special cases, but
3046 almost all the interesting cases are handled by creating
3047 a LAMBDA expression, calling COMPILE on it, then calling
3048 FUNCALL on the result.
3049 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3050 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3051 Since the code is newer, there might still be some new bugs
3052 (though not as many as before Martin Atzmueller's fixes:-). But
3053 the new code is substantially simpler and clearer, and hopefully
3054 any remaining bugs will be simpler, less fundamental, and more
3055 fixable then the bugs in the old code.
3056 ** The revised compiler is still a little unsteady on its feet.
3058 *** The debugging information it produces (particularly the names
3059 of FUNCTION objects) is sometimes much less useful than what
3060 the old compiler produced.
3061 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3062 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3063 within a MACROLET) has been temporarily weakened.
3064 ** There are new compiler optimizations for various functions:
3065 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3066 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3067 *** the math functions TRUNCATE, FLOOR, and CEILING
3068 *** the function-of-all-trades COERCE
3069 Mostly these should be transparent, but there's one
3070 potentially-annoying problem (bug 117): when the compiler
3071 inline-expands a function and does type analysis on the result,
3072 it can create control paths which have type mismatches, and
3073 when it can't prove that those control paths aren't taken,
3074 it will issue WARNINGs about the type mismatches. This is
3075 a particular problem in practice for the new sequence functions.
3076 It's not clear how this should be fixed, and for now, a
3077 workaround is given in the entry for 117 in the BUGS file.
3078 ** (Because of the interaction between the two previous items --
3079 occasional inlining problems and new inline expansions -- some
3080 of the new sequence function optimizations won't really kick in
3081 completely until debugging information, and then inlining, are
3082 straightened out in some future version.)
3083 * minor incompatible changes:
3084 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3085 DIRECTORY no longer implicitly promotes NIL slots of its
3086 pathname argument to :WILD. In particular, when you ask for the
3087 contents of a directory (which you used to be able to do without
3088 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3089 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3090 ** changes in behavior that ANSI explicitly defines to be
3091 implementation dependent:
3092 *** The new compiler-only implementation still conforms with ANSI,
3093 but acts a little different than before. Besides the obvious
3094 changes in performance tradeoffs (that the cost per form passed
3095 to EVAL has gone up, and the cost per form executed by EVAL
3096 has gone down), the behavior of the system changes a little
3097 because there are no longer any interpreted function objects.
3098 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3099 e.g. doing COMPILE on the output of interactive DEFUN is
3101 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3103 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3104 has changed from (:CL) to NIL.
3105 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3106 e.g. #\Newline and #\Tab, don't have names specified in the
3107 ANSI Common Lisp standard, is now based on their ASCII symbolic
3108 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3109 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3110 are no longer used for output.
3111 ** changes in internal implementation constants:
3112 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3113 4 million. (If your application spends a lot of time GCing and
3114 you have a lot of RAM, you might want to experiment with
3115 increasing it even more.)
3116 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3117 However, almost all old code should still continue to work without
3118 immediate update, as SB-C-CALL is now a (deprecated) nickname
3120 ** Old operator names in the style DEF-FOO are now deprecated in
3121 favor of new corresponding names DEFINE-FOO, for consistency with
3122 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3123 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3124 internal symbols, but a few supported extensions like
3125 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3126 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3127 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3128 etc. as you get deeper into recursive calls to the debugger
3129 command loop, instead of the old "5]", "5]]", "5]]]"
3130 sequence. (I was motivated to do this when squabbles between
3131 ILISP and SBCL left me very deeply nested in the debugger. In the
3132 short term, this change will probably provoke more ILISP/SBCL
3133 squabbles, but hopefully it will be an improvement in the long run.)
3134 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3135 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3136 function, since declaiming return types involving aliens is
3137 (1) annoyingly messy to do by hand and (2) vital to efficient
3138 compilation of code which calls such functions.
3139 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3140 longer reexported by the SB-EXT package. They're solely useful
3141 for alien code, so it seems more logical that you should get
3142 them from the SB-ALIEN package, not in SB-EXT.
3143 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3144 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3145 features. Instead, the code that they used to control is always
3146 built into the system.
3147 * many other bug fixes
3148 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3149 advantage of the new EVAL-WHEN stuff and to clean them up in
3150 general, and they are now more ANSI-compliant in a number of
3151 ways. Martin Atzmueller is responsible for a lot of this.
3152 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3154 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3155 *** correct ERROR type for various file operations
3156 *** some fixes for Lisp streams
3157 *** DEFMETHOD syntax checking
3158 *** changing old weird representation of debug information as
3159 strings (which, among their other deficiencies, don't transform
3160 correctly when you rename packages, and don't change their
3161 print representation when you change things like *PACKAGE*
3162 and *PRINT-LENGTH*) to symbols and lists of symbols
3163 He also made several improvements and fixed several bugs in DESCRIBE.
3164 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3166 *** misbehavior of WRITE-STRING/WRITE-LINE
3167 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3168 and several other LOOP problems as well
3169 *** DIRECTORY when similar filenames are present
3170 *** DEFGENERIC with :METHOD options
3171 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3172 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3173 *** argument ordering in FIND with :TEST option
3174 *** mishandled package designator argument in APROPOS-LIST
3175 *** various problems in the backquote readmacro
3176 *** a bug in APROPOS
3177 *** probably some others that I'm not describing very well here,
3178 since the CVS log documents them by reference to sbcl-devel
3179 messages, and the SourceForge archives aren't working well.:-(
3180 ** Dan Barlow improved the Alpha port (and is making progress on the
3181 PPC port, for those of you who think different).
3182 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3183 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3184 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3185 continues to work on the SPARC port (for those of you in a position
3186 to look down upon our little PC-compatible boxes from a great height).
3187 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3188 and uses that copy, rather than the current dynamic value, when
3189 it's trying to decide whether to truncate output. Thus e.g.
3190 (let ((*print-lines* 50))
3191 (pprint-logical-block (stream nil)
3193 (let ((*print-lines* 8))
3194 (print (aref possiblybigthings i) stream)))))
3195 should now truncate the logical block only at 50 lines, instead of
3196 often truncating it at 8 lines, as it did before.
3197 * The doc/cmucl/ directory, containing old CMU CL documentation
3198 from the time of the fork, is no longer part of the base system.
3199 SourceForge has shut down its anonymous FTP service, and with it
3200 my original plan for distributing the old CMU CL documentation
3201 there. For now, if you need these files you can download an old
3202 SBCL source release and extract them from it.
3203 * The fasl file version number changed again, for dozens of reasons,
3204 some of which are apparent above.
3206 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3207 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3208 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3209 patch, so that DISASSEMBLE output is much nicer.
3210 * The code in the SB-PROFILE package now seems reasonably stable.
3211 I still haven't decided what the final interface should look like
3212 (I'd like PROFILE to interact cleanly with TRACE, since both
3213 facilities use function encapsulation) but if you have a need
3214 for profiling now, you can probably use it successfully with
3215 the current CMU-CL-style interface.
3216 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3217 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3218 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3219 link now returns the dangling link itself, and for similar
3220 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3221 link itself. (In these cases the old code signalled an error and
3222 looped endlessly, respectively.) Thus, DIRECTORY now works even
3223 in the presence of dangling and cyclic symbolic links.
3224 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3225 is now a supported extension again, since the consensus on
3226 sbcl-devel was that it can be useful for ordinary development
3227 work, not just for debugging SBCL itself.
3228 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3229 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3230 that currently-defined functions might be redefined later with
3231 different return types.
3232 * Hash tables can be printed readably, as inspired by CMU CL code
3233 of Eric Marsden and SBCL code of Martin Atzmueller.
3234 * better error handling in CLOS method combination, thanks to
3235 Martin Atzmueller porting Pierre Mai's CMU CL patches
3236 * more overflow fixes for >16Mbyte I/O buffers
3237 * A bug in READ has been fixed, so that now a single Ctrl-D
3238 character suffices to cause end-of-file on character streams.
3239 In particular, now you only need one Ctrl-D at the command
3240 line (not two) to exit SBCL.
3241 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3243 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3244 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3245 does the right thing.
3246 * The implementation of some type tests, especially for CONDITION
3247 types, is now tidier and maybe faster, due to CMU CL code
3248 originally by Douglas Crosher, ported by Martin Atzmueller.
3249 * Some math functions have been fixed, and there are new
3250 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3251 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3252 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3253 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3254 likely to be built into sbcl-0.7.0.)
3255 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3256 CL, ported to SBCL by Martin Atzmueller)
3257 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3258 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3259 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3260 * There's a new slam.sh hack to shorten the edit/compile/debug
3261 cycle for low-level changes to SBCL itself, and a new
3262 :SB-AFTER-XC-CORE target feature to control the generation of
3263 the after-xc.core file needed by slam.sh.
3264 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3265 COMPILE-FILE is no longer supported, so that now every function
3266 gets an entry point, so that block compilation looks a little
3267 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3268 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3269 returns the number of bytes consed since the system started,
3270 rather than the number consed since the first time the function
3271 was called. (The new definition parallels ANSI functions like
3272 CL:GET-INTERNAL-RUN-TIME.)
3273 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3274 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3275 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3276 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3277 * Fasl file version numbers are now independent of the target CPU,
3278 since historically most system changes which required version
3279 number changes have affected all CPUs equally. Similarly,
3280 the byte fasl file version is now equal to the ordinary
3283 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3284 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3285 is no longer recognized. I apologize for this, because it was
3286 listed in SB-EXT as a supported extension, but I found that
3287 its existing behavior was poorly specified, as well as incorrectly
3288 specified, and it looked like too much of a mess to straighten it
3289 out. I have enough on my hands trying to get ANSI stuff to work..
3290 * many patches ported from CMU CL by Martin Atzmueller, with
3291 half a dozen bug fixes in pretty-printing and the debugger, and
3292 half a dozen others elsewhere
3293 * fixed bug 13: Floating point infinities are now supported again.
3294 They might still be a little bit flaky, but thanks to bug reports
3295 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3296 as flaky as they were.
3297 * The --noprogrammer command line option is now supported. (Its
3298 behavior is slightly different in detail from what the old man
3299 page claimed it would do, but it's still appropriate under the
3300 same circumstances that the man page talks about.)
3301 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3302 are now supported, and enabled by default. Thus, the compiler can
3303 handle many floating point and complex operations much less
3304 inefficiently. (Thus e.g. you can implement a complex FFT
3306 * The compiler now detects type mismatches between DECLAIM FTYPE
3307 and DEFUN better, and implements CHECK-TYPE more correctly, and
3308 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3309 thanks to patches from Martin Atzmueller.
3310 * various fixes to make the cross-compiler more portable to
3311 ANSI-conforming-but-different cross-compilation hosts (notably
3312 Lispworks for Windows, following bug reports from Arthur Lemmens)
3313 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3314 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3315 fixed thanks to Pierre Mai's CMU CL patches.
3316 * fixes to keep the system from overflowing internal counters when
3317 it tries to use i/o buffers larger than 16M bytes
3318 * fixed bug 45a: Various internal functions required to support
3319 complex special functions have been merged from CMU CL sources.
3320 (When I was first setting up SBCL, I misunderstood a compile-time
3321 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3322 * improved support for type intersection and union, fixing bug 12
3323 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3324 more obscure bugs as well
3325 * some steps toward byte-compiling non-performance-critical
3326 parts of the system, courtesy of patches from Martin Atzmueller
3327 * Christophe Rhodes has made some debian packages of sbcl at
3328 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3329 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3330 stable, but are nonetheless usable. When he's ready, I'd be happy
3331 to add them to the SourceForge "File Releases" section. (And if
3332 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3333 * new fasl file format version number (because of changes in
3334 internal representation of (OR ..) types to accommodate the new
3335 support for (AND ..) types, among other things)
3337 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3338 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3340 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3342 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3343 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3344 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3345 * DESCRIBE now works on structure objects again.
3346 * Most function call argument type mismatches are now handled as
3347 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3348 know whether the function will be redefined before the call is
3349 executed. (The compiler could flag local calls with full WARNINGs,
3350 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3351 it doesn't keep track of enough information to know whether calls
3352 are local in this sense.)
3353 * Compiler output is now more verbose, with messages truncated
3354 later than before. (There should be some supported way for users
3355 to override the default verbosity, but I haven't decided how to
3356 provide it yet, so this behavior is still controlled by the internal
3357 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3358 src/compiler/ir1util.lisp.)
3359 * Fasl file format version numbers have increased again, because
3360 support for the Gray streams extension changes the layout of the
3361 system's STREAM objects.
3362 * The Gray subclassable streams extension now works, thanks to a
3363 patch from Martin Atzmueller.
3364 * The full LOAD-FOREIGN extension (not just the primitive
3365 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3366 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3367 but like most other programs, it defaults to copying the Unix
3368 environment from the original process instead of starting the
3369 new process in an empty environment.
3370 * Extensions which manipulate the Unix environment now support
3371 an :ENVIRONMENT keyword option which doesn't smash case or
3372 do other bad things. The CMU-CL-style :ENV option is retained
3373 for porting convenience.
3374 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3375 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3377 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3379 * A patch from Martin Atzmueller seems to have solved the SIGINT
3380 problem, and as far as we know, signal-handling now works cleanly.
3381 (If you find any new bugs, please report them!)
3382 * The system no longer defaults Lisp source file names to types
3383 ".l", ".cl", or ".lsp", but only to ".lisp".
3384 * The compiler no longer uses special default file extensions for
3385 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3386 seems to expect a single default extension for all compiled code,
3387 and there's no compelling reason to try to stretch the standard
3388 to allow two different extensions.) Instead, byte-compiled files
3389 default to the same extension as native-compiled files.
3390 * Fasl file format version numbers have increased again, because
3391 a rearrangement of internal implementation packages made some
3392 dumped symbols in old fasl files unreadable in new cores.
3393 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3394 many fewer weird special cases.
3395 * Bug #17 (differing COMPILE-FILE behavior between logical and
3396 physical pathnames) has been fixed, and some related misbehavior too,
3397 thanks to a patch from Martin Atzmueller.
3398 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3399 by Tim Moore, ported to SBCL by Martin Atzmueller.
3400 * Martin Atzmueller fixed several filesystem-related problems,
3401 including bug #36, in part by porting CMU CL patches, which were
3402 written in part by Paul Werkowski.
3403 * More compiler warnings in src/runtime/ are gone, thanks to
3404 more patches from Martin Atzmueller.
3405 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3408 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3410 * DESCRIBE now works on CONDITION objects.
3411 * The debugger now handles errors which arise when trying to print
3412 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3414 * The build system now uses an additional file, customize-target-features.lisp,
3415 to allow local modifications to the target *FEATURES* list. (The point of
3416 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3417 debugging features enabled, without having to worry about propagating it
3418 into everyone's system when I do a "cvs update".) When no
3419 customize-target-features.lisp file exists, the target *FEATURES* list
3420 should be constructed the same way as before.
3421 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3422 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3423 warn about behavior which is undefined under the ANSI spec. Note: This
3424 is specified by ANSI, but it's not very popular with programmers.
3425 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3426 macro in the SBCL sources for an example of a workaround which you
3427 might use to make portable ANSI-standard code which does what you want.
3428 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3429 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3430 when it's used as a non-toplevel form. (This is required in order
3431 to implement the DEFCONSTANT-EQX macro.)
3432 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3433 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3434 funny things, due to bugs in the implementation of EVAL-WHEN
3435 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3436 non-toplevel DEFCONSTANTs are uncommon.)
3437 * The core file version number and fasl file version number have been
3438 incremented, because the old noncompliant DEFCONSTANT behavior involved
3439 calling functions which no longer exist, and because I also took the
3440 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3441 * fixed bug 1 (error handling before read-eval-print loop starts), and
3442 redid debugger restarts and related debugger commands somewhat while
3444 ** The QUIT debugger command is gone, since it did something
3445 rather different than the SB-EXT:QUIT command, and since it never
3446 worked properly outside the main toplevel read/eval/print loop.
3447 Invoking the new TOPLEVEL restart provides the same functionality.
3448 ** The GO debugger command is also gone, since you can just invoke
3449 the CONTINUE restart directly instead.
3450 ** The TOP debugger command is also gone, since it's redundant with the
3451 FRAME 0 command, and since it interfered with abbreviations for the
3453 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3455 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3456 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3457 some porting work by Martin Atzmueller)
3458 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3459 with a boatload of patches which he ported to SBCL. Now that those
3461 ** The system tries to make sure that its low-priority messages
3462 are prefixed by semicolons, to help people who like to use
3463 syntax highlighting in their ILISP buffer. (This patch
3464 was originally due to Raymond Toy.)
3465 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3466 patches originally written by Raymond Toy.
3467 ** The compiler understands coercion between single-value and
3468 multiple-VALUES type expressions better, getting rid of some very
3469 weird behavior, thanks to patches originally by Robert MacLachlan
3470 and Douglas Crosher.
3471 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3472 lambda lists, thanks to a patch originally by Pierre Mai.
3473 ** The system no longer bogusly warns about "abbreviated type
3475 ** The compiler gets less confused by inlining and RETURN-FROM,
3476 thanks to some patches originally by Tim Moore.
3477 ** The system no longer hangs when dumping circular lists to fasl
3478 files, thanks to a patch originally from Douglas Crosher.
3479 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3480 undefined function error.
3481 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3482 in structure slot initforms) for now, documented workaround instead:-|
3483 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3484 * fixed bug 5: added stubs for various Gray stream functions called
3485 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3486 installed, at least appropriate type errors are generated
3487 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3488 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3489 consistently in DEFMETHOD forms.
3490 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3491 it doesn't seem to affect SBCL after all
3492 * The C runtime system now builds with better optimization and many
3493 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3495 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3497 * The system is now under CVS at SourceForge (instead of the
3498 CVS repository on my home machine).
3499 * The new signal handling code has been tweaked to treat register
3500 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3501 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3502 downstream, e.g. in debug-int.lisp, has implicit dependencies
3503 on the unsignedness of integer representation of machine words,
3504 and that caused the system to bomb out with infinite regress
3505 when trying to recover from type errors involving signed values,
3506 e.g. (BUTLAST '(1 2 3) -1).)
3507 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3508 (This was one of the bugs Peter Van Eynde reported back in July.)
3509 * The system now uses code inspired by Colin Walters' O(N)
3510 implementation of MAP (from the cmucl-imp@cons.org mailing
3511 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3512 inline the MAP operation, and there is more than one
3513 sequence argument to the MAP call (so that it can't just
3514 do ETYPECASE once and for all based on the type of the
3515 single sequence argument). (The old non-inline implementation
3516 of the general M-argument sequence-of-length-N case required
3517 O(M*N*N) time when any of the sequence arguments were LISTs.)
3518 * The QUIT :UNIX-CODE keyword argument has been renamed to
3519 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3520 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3522 * Raymond Wiker's patches to port dynamic loading from Linux to
3523 FreeBSD have been added.
3524 * The BUGS file is now more nearly up to date, thanks in large part
3525 to Martin Atzmueller's review of it.
3526 * The debugger now flushes standard output streams before it begins
3527 its output ("debugger invoked" and so forth).
3528 * The core version number and fasl file version number have both
3529 been incremented, because of incompatible changes in the layout
3531 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3532 used to not be called for a saved Lisp image.)
3533 * Martin Atzmueller's version of a patch to fix a compiler crash,
3534 as posted on sbcl-devel 13 September 2000, has been installed.
3535 * Instead of installing Martin Atzmueller's patch for the
3536 compiler transform for SUBSEQ, I deleted the compiler transform,
3537 and transforms for some similar consing operations.
3538 * A bug in signal handling which kept TRACE from working on OpenBSD
3540 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3541 away by constant folding
3542 * The system now defines its address space constants in one place
3543 (in the Lisp sources), and propagates them automatically elsewhere
3544 (through GENESIS and the sbcl.h file). Therefore, patching the
3545 address map is less unnecessarily tedious and error-prone. The
3546 Lisp names of address space constants have also been systematized.
3547 * CVS tags like dollar-Header-dollar have been removed from
3548 the sources, because they have never saved me trouble and
3549 they've been source of trouble working with patches and other
3550 diff-related operations.
3551 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3552 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3554 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3556 * The system has been ported to OpenBSD.
3557 * The system now compiles with a simple "sh make.sh" on the systems
3558 that it's supported on. I.e., now you no longer need to tweak
3559 text in the target-features.lisp-expr and symlinks in src/runtime/
3560 by hand, the make.sh takes care of it for you.
3561 * The system is no longer so grossly inefficient when compiling code
3562 involving vectors implemented as general (not simple) vectors (VECTOR T),
3563 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3564 which dares to use the various sequence functions on non-simple
3565 vectors, takes less of a performance hit.
3566 * There is now a primitive type predicate VECTOR-T-P
3567 to test for the (VECTOR T) type, so that e.g.
3568 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3569 can now be compiled with some semblance of efficiency. (The old code
3570 turned the type declaration into a full call to %TYPEP at runtime!)
3571 * AREF on (VECTOR T) is still not fast, since it's still compiled
3572 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3573 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3574 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3575 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3576 without making full calls to SUBTYPEP at runtime.
3577 (Some analogous efficiency issues for non-simple vectors specialized to
3578 element types other than T, or for non-simple multidimensional arrays,
3579 have not been addressed. They could almost certainly be handled the
3580 same way if anyone is motivated to do so.)
3581 * The changes in array handling break binary compatibility, so
3582 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3583 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3584 instead of (VALUES T T).
3585 * By following the instructions that Dan Barlow posted to sbcl-devel
3586 on 2 July 2000, I was able to enable primitive dynamic object
3587 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3588 functionality is not implemented (since it calls ld to resolve
3589 library references automatically, requiring RUN-PROGRAM for its
3590 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3591 try to resolve library references) is now supported.
3592 * The system now flushes the standard output streams when it terminates,
3593 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3594 them at several other probably-convenient times, e.g. in each pass of
3595 the toplevel read-eval-print loop, and after evaluating a form given
3596 as an "--eval" command-line option. (These changes were motivated by a
3597 discussion of stream flushing issues on cmucl-imp in August 2000.)
3598 * The source transform for TYPEP of array types no longer assumes
3599 that an array whose element type is a not-yet-defined type
3600 is implemented as an array of T, but instead punts, so that the
3601 type will be interpreted at runtime.
3602 * There is now some support for cross-compiling in make.sh: each of
3603 the phases of make.sh has its own script. (This should be transparent
3604 to people doing ordinary, non-cross-compile builds.)
3605 * Since my laptop doesn't have hundreds of megabytes of memory like
3606 my desktop machine, I became more motivated to do some items on
3607 my to-do list in order to reduce the size of the system a little:
3608 ** Arrange for various needed-only-at-cold-init things to be
3609 uninterned after cold init. To support this, those things have
3610 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3611 symbols with such names are now uninterned after cold init).
3612 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3613 fasl files for cold load.
3614 ** Remove the old compiler structure pooling code (which used to
3615 be conditional on the target feature :SB-ALLOC) completely.
3616 ** Redo the representation of some data in cold init to be more compact.
3617 (I also looked into supporting byte compiled code at bootstrap time,
3618 which would probably reduce the size of the system a lot, but that
3619 looked too complicated, so I punted for now.)
3620 * The maximum signal nesting depth in the src/runtime/ support code has
3621 been reduced from 4096 to 256. (I don't know any reason for the very
3622 large old value. If the new smaller value turns out to break something,
3623 I'll probably just bump it back up.)
3624 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3626 * Many, many bugs reported by Peter Van Eynde have been added to
3627 the BUGS list; some have even been fixed.
3628 * While enabling dynamic object file loading, I tried to make the
3629 code easier to understand, renaming various functions and variables
3630 with less ambiguous names, and changing some function calling
3631 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3632 * While trying to figure out how to do the OpenBSD port, I tried to
3633 clean up some of the code in src/runtime/. In particular, I dropped
3634 support for non-POSIX signal handling, added various comments,
3635 tweaked the code to reduce the number of compilation warnings, and
3636 renamed some files to increase consistency.
3637 * To support the new automatic configuration functionality in make.sh,
3638 the source file target-features.lisp-expr has been replaced with the
3639 source file base-target-features.lisp-expr and the machine-generated
3640 file local-target-features.lisp-expr.
3641 * fixed a stupid quoting error in make.sh so that using CMU CL
3642 "lisp -batch" as cross-compilation host works again
3644 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3646 * DESCRIBE no longer tries to call itself recursively to describe
3647 bound/fbound values, so that it no longer fails on symbols which are
3648 bound to themselves (like keywords, T, and NIL).
3649 * DESCRIBE now works on generic functions.
3650 * The printer now prints less-screwed-up representations of closures
3651 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3652 * A private symbol is used instead of the :EMPTY keyword previously
3653 used to mark empty slots in hash tables. Thus
3654 (DEFVAR *HT* (MAKE-HASH-TABLE))
3655 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3656 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3657 now does what ANSI says that it should. (You can still get
3658 similar noncompliant behavior if bang on the hash table
3659 implementation with all the symbols you get back from
3660 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3661 This breaks binary compatibility, since tests for equality to
3662 :EMPTY are wired into things like the macroexpansion of
3663 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3665 * There's now a minimal placeholder implementation for CL:STEP,
3666 as required by ANSI.
3667 * An obscure bug in the interaction of the normal compiler, the byte
3668 compiler, inlining, and structure predicates has been patched
3669 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3670 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3672 * Missing ordinary arguments in a macro call are now detected even
3673 when the macro lambda list contains &KEY or &REST.
3674 * The debugger no longer complains about encountering the top of the
3675 stack when you type "FRAME 0" to explicitly instruct it to go to
3676 the top of the stack. And it now prints the frame you request even
3677 if it's the current frame (instead of saying "You are here.").
3678 * As specified by ANSI, the system now always prints keywords
3679 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3681 * The default initial SIZE of HASH-TABLEs is now smaller.
3682 * Type information from CLOS class dispatch is now propagated
3683 into DEFMETHOD bodies, so that e.g.
3684 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3686 is now basically equivalent to
3687 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3688 (DECLARE (TYPE SINGLE-FLOAT X))
3690 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3691 operation, without having to do run-time type dispatch.
3692 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3693 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3694 * Since I don't seem to be making big file reorganizations very often
3695 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3696 is overflowing my ability to conveniently back them up), I've finally
3697 checked the system into CVS. (The CVS repository is on my home system,
3698 not at SourceForge -- putting it on SourceForge might come later.)
3699 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3700 high-level GC-NOTIFY-FOO functions send their output. (There's
3701 still very little control of where low-level verbose GC functions
3702 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3703 less than it used to -- the GC-NOTIFY-FOO functions are now under
3704 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3705 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3706 in only one place in the source code, and propagates it automatically
3707 everywhere that it's needed. Thus e.g. when I bump the version from
3708 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3709 * The C source files now include boilerplate legalese and documentation
3710 at the head of each file (just as the Lisp source files already did).
3711 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3712 to his page will be replaced with a link to his new CLiki service.
3714 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3716 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3717 * The build process now looks for GNU make under the default name "gmake",
3718 instead of "make" as it used to. If GNU make is not available as "gmake"
3719 on your system, you can change this default behavior by setting the
3720 GNUMAKE environment variable.
3721 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3722 can build without error under CMU CL.
3724 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3726 * There is now a partial SBCL user manual (with some new text and some
3727 text cribbed from the CMU CL manual).
3728 * The beginnings of a profiler have been added (starting with the
3729 CMU CL profiler and simplifying and cleaning up). Eventually the
3730 main interface should be through the TRACE macro, but for now,
3731 it's still accessed through vaguely CMU-CL-style functions and macros
3732 exported from the package SB-PROFILE.
3733 * Some problems left over from porting CMU CL to the new
3734 cross-compilation bootstrap process have been cleaned up:
3735 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3736 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3737 part when getting rid of the compiler *BACKEND* stuff.)
3738 ** The value of *NULL-TYPE* was screwed up, because it was
3739 being initialized before the type system knew the final
3740 definition of the 'NULL type. This screwed up several key
3741 optimizations in the compiler, causing inefficiency in all sorts
3742 of places. (I found it because I wanted to understand why
3743 GET-INTERNAL-RUN-TIME was consing.)
3744 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3745 PROCLAIM FTYPE information. Unfortunately this broke binary
3746 compatibility again, since now the forms output by DEFGENERIC
3747 to refer to functions which didn't exist in 0.6.3.
3748 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3749 can use the new (as of 0.6.3) transform for SOME into MAP into
3751 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3752 values to (INTEGER 0 1000000), so that the time code will no longer
3753 occasionally get blown up by Linux returning 1000000 microseconds
3754 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3755 its output conform to the ANSI spec. (Alas, this makes its output
3756 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3757 * A full call to MAP NIL with a single sequence argument no longer conses.
3758 * fixes to problems pointed out by Martin Atzmueller:
3759 * The manual page no longer talks about multiprocessing as though
3760 it were currently supported.
3761 * The ILISP support patches have been removed from the distribution,
3762 because as of version 5.10.1, ILISP now supports SBCL without us
3763 having to maintain patches.
3764 * added a modified version of Raymond Toy's recent CMU CL patch for
3765 EQUALP comparison of HASH-TABLE
3767 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3769 * The system still can't cross-compile itself with
3770 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3771 entails), but at least it can compile more of itself that way
3772 than it used to be able to, and various buglets which were uncovered
3773 by trying to cross-compile itself that way have now been fixed.
3774 * This release breaks binary compatibility again. This time
3775 at least I've incremented the FASL file format version to 2, so that the
3776 problem can be detected reliably instead of just causing weird errors.
3777 * various new style warnings:
3778 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3779 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3780 ** using the lexical binding of a variable named in the *FOO* style
3781 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3782 as specified by ANSI.
3783 * *RANDOM-STATE* is no longer automatically initialized from
3784 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3785 default behavior of the system is to repeat its behavior every time
3786 it's run. If you'd like to change this behavior, you can always
3787 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3788 old convention there was no comparably easy way to get the system to
3789 repeat its behavior every time it was run.
3790 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3791 FTYPE declarations has been removed, in favor of their ANSI
3792 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3793 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3794 the types of their sequence arguments can be determined at compile time.
3795 This is done through a new open code expansion for MAP which eliminates
3796 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3797 argument types can be determined at compile time.
3798 * The optimizer now transforms COERCE into an identity operation when it
3799 can prove that the coerced object is already of the correct type. (This
3800 can be a win for machine generated code, including the output of other
3801 optimization transforms, such as the MAP transform above.)
3802 * Credit information has been moved from source file headers into CREDITS.
3803 * Source file headers have been made more standard.
3804 * The CASE macro now compiles without complaining even when it has
3807 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3809 * (Note that the way that the PCL macroexpansions were rewritten
3810 to accommodate the change in DEFGENERIC below breaks binary
3811 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3812 not run under sbcl-0.6.2. Once we get out of alpha releases,
3813 i.e. hit release 1.0.0, we'll probably try to maintain binary
3814 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3815 and sbcl-1.4.4. Until then, however, it might be fairly common
3816 for maintenance releases to break binary compatibility.)
3817 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3819 * The reporting of unhandled signals has been changed to print some
3820 explanatory text as well as the report form. (Previously only
3821 the report form was printed.)
3822 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3823 it defines, so that the compiler no longer issues undefined function
3824 warnings for compiled-but-not-yet-loaded generic functions.
3825 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3826 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3827 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3828 section of the ANSI spec are supported.
3829 * The "" nickname for the "KEYWORD" package has been removed.
3830 The reader still handles symbol tokens which begin with a package marker
3831 as keywords, but it doesn't expose its mechanism for doing so in the
3832 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3833 * The system now issues STYLE-WARNINGs for contradictory TYPE
3834 proclamations. (Warnings for contradictory FTYPE proclamations would
3835 be nice too, but those can't be done usefully unless the type system
3836 is made smarter about FUNCTION types.)
3837 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3838 systematized, so that "*target-*.lisp is supposed to exist only on the
3839 target and imply that there's a related file which exists on the
3840 host, and *host-*.lisp is supposed to exist only on the host and imply
3841 that there's a related file which exists on the target. This involves a
3842 lot of renaming. Hopefully the acute confusion caused by the renaming
3843 will be justified by the reduction in chronic confusion..
3844 ** runtime-type.lisp -> early-target-type.lisp
3845 ** target-type.lisp -> late-target-type.lisp
3846 ** early-host-format.lisp -> early-format.lisp
3847 ** late-host-format.lisp -> late-format.lisp
3848 ** host-error.lisp -> misc-error.lisp
3849 ** early-error.lisp -> early-target-error.lisp
3850 ** late-error.lisp -> late-target-error.lisp
3851 ** host-defboot.lisp -> early-defboot.lisp
3852 ** code/misc.lisp -> code/target-misc.lisp
3853 ** code/host-misc.lisp -> code/misc.lisp
3854 ** code/numbers.lisp -> code/target-numbers.lisp
3855 ** code/early-numbers.lisp -> numbers.lisp
3856 ** early-host-type.lisp -> early-type.lisp
3857 ** late-host-type.lisp -> late-type.lisp
3858 ** host-typep.lisp -> typep.lisp
3859 ** load.lisp -> target-load.lisp
3860 ** host-load.lisp -> load.lisp
3861 ** host-disassem.lisp -> disassem.lisp
3862 ** host-insts.lisp -> insts.lisp
3863 ** byte-comp.lisp -> target-byte-comp.lisp
3864 ** host-byte-comp.lisp -> byte-comp.lisp
3865 ** host-signal.lisp -> signal.lisp
3866 ** host-defstruct.lisp -> defstruct.lisp
3867 ** late-target-type.lisp -> deftypes-for-target.lisp
3868 Furthermore, several other previously target-only files foo.lisp (e.g.
3869 hash-table.lisp and random.lisp) have been split into a target-and-host
3870 foo.lisp file and a target-only target-foo.lisp file, with their key type
3871 definitions in the target-and-host part, so that the cross-compiler will
3872 know more about target types.
3873 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3874 gone away. In their place are various *BACKEND-FOO* variables
3875 corresponding to the slots of the old structure.
3876 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3877 into the target SBCL has been fixed.
3878 * The chill.lisp system for loading cold code into a running SBCL
3880 * Support for the CMU CL "scavenger hook" extension has been removed.
3881 (It was undocumented and unused in the CMU CL sources that SBCL was
3882 derived from, and stale in sbcl-0.6.1.)
3883 * Various errors in the cross-compiler type system were detected
3884 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3885 (enabling various consistency checks). Many of them were fixed,
3886 but some hard problems remain, so the compiler is back to
3887 running without *TYPE-SYSTEM-INITIALIZED* for now.
3888 * As part of the cross-compiler type system cleanup, I implemented
3889 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3890 * I have started adding UNCROSS calls throughout the type system
3891 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3892 on cross-compiler input files will be able to go away, and various
3894 * CONSTANTP now returns true for quoted forms (as explicitly required
3897 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3899 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3900 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3901 (MAKE-LIST -1). In the longer term, it ought to become true
3902 that declarations are assertions even at SAFETY 1. For now, it's not
3903 quite true even at SAFETY 3, but it's at least more nearly true..
3904 (Note that this change seems to increases the size of the system by
3905 O(5%) and to decrease the speed of the compiler by 20% or more.)
3906 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3907 for the problem of printing dozens of lines of distracting information
3908 about low-level system machinery as part of the top stack frame
3909 on entry to the debugger when an undefined function was called.
3910 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3911 is not reset to COMMON-LISP-USER.
3912 * Compilation of stuff related to dyncount.lisp has been made conditional
3913 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3914 smaller. The various dyncount-related symbols have been moved into
3915 a new "SB-DYNCOUNT" package.
3916 * tty-inspect.lisp has been renamed to inspect.lisp.
3917 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3918 feature has gone away. (When we eventually port to other flavors of
3919 libc and/or Unix, we'll try to make the differences between flavors
3920 invisible at the user level.)
3921 * Various other *FEATURES* tags, and/or their associated conditionals,
3922 have been removed if obsolescent, or given better documentation, or
3923 sometimes given more-mnemonic names.
3925 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3927 * tidied up "make.sh" script
3928 * tidied up system directory structure
3929 * better "clean.sh" behavior
3930 * added doc/FOR-CMUCL-DEVELOPERS
3931 * many many small tweaks to output format, e.g. removing possibly-confusing
3932 trailing #\. character in DESCRIBE-INSTANCE
3933 * (EQUALP #\A 'A) no longer signals an error.
3934 * new hashing code, including EQUALP hashing
3935 * tidied up Lisp initialization and toplevel
3936 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3937 * command line argument processing
3938 * added POSIX-GETENV function to deal with Unix-ish environment variables
3939 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3940 terminating SBCL on EOF
3941 * non-verbose GC by default
3942 * There is no more "sbcl" shell script; the sbcl file is now the C
3943 runtime executable (just like CMU CL).
3944 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3946 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3947 symbol and package information as Lisp native symbol and package objects
3948 instead of strings naming symbols and strings naming packages. This way,
3949 whenever packages are renamed (as in warm init), debug information is
3950 transformed along with everything else.
3951 * tweaked the optimization policy declarations which control the building
3952 of SBCL itself. Now, among other things, the system no longer saves
3953 source location debugging information. (This helps two problems at once
3954 by reducing SBCL size and by keeping SBCL from trying to look for its
3955 sources -- which may not exist -- when reporting errors.)
3956 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3957 debugging and testing purposes
3958 * cleaned up printing, making the printer call PRINT-OBJECT for
3959 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3960 methods, giving nearly-ANSI behavior
3961 * converted almost all special variables to use *FOO* naming convention
3962 * deleted PARSE-TIME functionality, since it can be done portably
3963 * moved some files out of cold init into warm init
3964 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3966 * regularized formatting of source files
3967 * added an install.sh script
3968 * fixed ridiculous memory usage of cross-compiler by making
3969 compiler/alloc.lisp not try to do pooling unless it can hook
3970 itself into the GC of the cross-compilation host. Now the system
3971 builds nicely on my old laptop.
3972 * added :SB-ALLOC in target-features.lisp-expr
3973 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3974 was not implemented)
3975 * re-did condition handling and note reporting in the compiler. Notes
3976 are no longer handled by signalling conditions. Style warnings
3977 and warnings are handled more correctly and reported in such a way
3978 that it's easy to find one or the other in your output (so that you
3979 can e.g. figure out which of many problems caused COMPILE-FILE to
3981 * changed the severity of several compiler warnings from full WARNING
3982 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3983 compiler note reporting so that it doesn't use the condition system
3984 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3985 COMPILE-FILE command)
3986 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3987 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3988 no longer supported, since their implementation was deeply intertwingled
3989 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3990 * removed bogus "support" for compiler macros named (SETF FOO), and
3991 removed the compiler macro for SETF INFO (but only after making a fool
3992 of myself on the cmucl-imp mailing list by posting a bogus patch for
3993 DEFINE-COMPILER-MACRO..)
3994 * Compiled files containing forms which have side effects on the Lisp
3995 reader (such as DEFPACKAGE forms) are now handled more correctly.
3996 (Compiler queuing of top level lambdas has been suppressed by setting
3997 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3998 * deleted various currently-unused source files, e.g. gengc.lisp. They
3999 may be added back at some point e.g. when porting to other architectures,
4000 but until they are it's distracting to distribute them and to try to
4002 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4003 there were so many of them they're just distractions, and UNCROSS is
4004 known to be able to handle the current sources
4005 * moved PROFILE functionality into TRACE, so that it will be clear
4006 how the wrapping and unwrapping of functions when you profile them
4007 interacts with the wrapping and unwrapping of functions when you
4008 trace them. (Actually, the functionality isn't there yet, but at least
4009 the interface specification is there. Hopefully, the functionality will
4010 arrive with some future maintenance release.)
4011 * removed host-oops.lisp
4012 * changed signature of QUIT function to allow UNIX-CODE argument
4013 * fixed READ-SEQUENCE bug
4014 * tweaked verbose GC output so that it looks more like the progress
4015 output that ANSI specifies for functions like LOAD
4016 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4017 * added <http://sbcl.sourceforge.com> to the banner information printed by