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 * improvements to the Windows port:
8 ** floating point exceptions are now reported correctly.
9 ** stack exhaustion detection works partially.
10 ** more accurate GET-INTERNAL-REAL-TIME.
12 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
13 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
14 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
16 * bug fix: two potential GC deadlocks affecting threaded builds.
17 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
18 returns T (reported by Anton Kazennikov)
19 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
20 with non-variable places
21 * bug fix: remove a race condition in the setting of
22 funcallable-instance functions, this should make threaded CLOS
23 code more stable against memory faults.
24 * bug fix: corruption of specials when unbinding is interrupted by an
25 asynchronous unwind (reported by Hannu Koivisto)
26 * improvement: the debugger will now also display local variables that
27 are only used once, for code compiled with a DEBUG optimization quality
30 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
31 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
32 * incompatible change: External-format support for FFI calls. The
33 SB-ALIEN:C-STRING no longer implies an ASCII
34 external-format. Instead the string is subject to external-format
35 conversion. Additionally return values of type SB-ALIEN:C-STRING
36 are by default of type (SIMPLE-ARRAY CHARACTER), not
37 SIMPLE-BASE-STRING. For an alien type that behaves like the old
38 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
39 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
40 * incompatible change: SB-EXT package no longer contains the
41 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
42 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
44 * incompatible change: the single-stepper is no longer available
45 on Alpha, Mips and Sparc platforms.
46 * minor incompatible change: the direct superclasses of
47 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
48 not (STANDARD-OBJECT FUNCTION). This makes the
49 class-precedence-lists of GENERIC-FUNCTION and
50 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
52 * minor incompatible change: the default stream external format on
53 non-unicode SBCL is no longer detected from the locale, but is always
55 * new feature: Add a version of evaluator that uses an interpreter instead
56 of the compiler. EVAL still uses the compiler by default, to switch it
57 to use the interpreter, set the value of the variable
58 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
59 * minor incompatible change: the single-stepper REPL has been merged
60 with the normal debugger (see the "Stepping" heading of the debugger help
61 for more details). The debugger command STEP will no longer switch
62 to the single-stepper REPL.
63 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
65 * bug fix: loading of default sysinit file works. (thanks to Leonid
67 * bug fix: better detection of circularities in the file-compiler.
68 (reported by Marco Monteiro)
69 * bug fix: the CL pathname functions now work with files that have
70 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
71 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
72 systems with Unix98 pty semantics.
73 * bug fix: ASDF-INSTALL will now work with bsd tar.
74 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
76 * bug fix: timers expiring in dead threads no longer cause a
77 type-error (reported by Paul "Nonny Mouse").
78 * bug fix: thanks to more lightweight single-stepper instrumentation,
79 code compiled with (DEBUG 3) will compile and execute significantly faster,
80 and will have more accurate type-inferencing than before
81 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
82 whose bindings are modified
83 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
84 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
85 ** replace dummy implementations of CL:MACHINE-INSTANCE and
86 CL:SOFTWARE-VERSION with proper version
88 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
89 * feature: implemented the READER-METHOD-CLASS and
90 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
92 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
94 * optimization: faster LOGCOUNT implementation on x86 and x86-64
95 (thanks to Lutz Euler)
96 * optimization: hashing of general arrays and vectors has been
97 improved. (reported by Any Fingerhut)
98 * enhancement: SB-INTROSPECT is now able to find definitions of
99 profiled functions. (thanks to Troels Henriksen)
100 * enhancement: compiler-macro expansion applies now to FUNCALL forms
101 as well, allowing compiler-macros for SETF-functions to expand.
102 * enhancement: step-instrumentation no longer wraps known
103 single-value functions with multiple-value context, allowing
104 better type inference.
105 * fixed bug #337: use of MAKE-METHOD in method combination now works
106 even in the presence of user-defined method classes. (reported by
107 Bruno Haible and Pascal Costanza)
108 * fixed bug #339(c): if there are applicable methods not part of any
109 long-form method-combination group, call INVALID-METHOD-ERROR.
110 (reported by Bruno Haible)
111 * fixed bug #361: the :FUNCTION initarg in the protocol for
112 initialization of methods can now be used to override
113 internally-produced optimized functions. (reported by Bruno
115 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
116 system-provided lambda expression no longer cause warnings about
117 unbound #:|pv-table| symbols.
118 * bug fix: improved the handling of type declarations and the
119 detection of violations for keyword arguments with non-constant
121 * bug fix: potentially erronous calls to PATHNAME and
122 MERGE-PATHNAMES were being flushed in some cases.
123 (reported by Richard Kreuter)
124 * bug fix: compiled calls to TYPEP were mishandling obsolete
125 instances. (reported by James Bielman and Attila Lendvai)
126 * bug fix: format strings with ~> without matching ~< no longer
127 trigger an AVER, but signal an understandable error instead.
128 (reported by Antonio Martinez)
129 * bug fix: specifying an output-file without a pathname-type for
130 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
131 of input-file instead of "fasl". (reported by Robert Dodier)
132 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
133 (reported by James Y Knight).
134 * bug fix: compiler-macros-function did not consider the environment
135 argument for shadowing by local functions.
136 * bug fix: compiler-macros expansion was inhibited by local INLINE
138 * bug fix: inline expansions of known functions were subject to
139 step-instrumentation in high debug policies, leading to problems
141 * bug fix: compiler failed to differentiate between different CONS
143 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
144 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
145 * bug fix: #368: incorrect use of expressed vs. upgraded array
147 * bug fix: #306a: more precise unions of array types.
148 * thread-safety improvements:
149 ** CONDITION-WAIT could return early on Linux, if the thread was
150 interrupted and subsequently continued with SIGCONT.
151 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
153 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
154 * added support for the ucs-2 external format. (contributed by Ivan
156 * minor incompatible change: pretty printing of objects of type
157 (cons symbol) is, in the default pprint-dispatch-table, now
158 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
160 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
161 called later in a class's lifetime, possibly as late as when the
162 first instance of the class is created. Previously,
163 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
164 class became finalizeable.
165 * fixed bug: FILE-POSITION sometimes returned inconsistent results
166 for multibyte external-format streams. (thanks to "vbzoli")
167 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
168 with :ALLOCATION :CLASS inherited from superclasses of the
170 * fixed bug: anonymous classes can now be created using the :NAME
171 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
172 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
173 * fixed bug: core-files saved with :EXECUTABLE T can again be
174 executed when SBCL_HOME isn't set. (reported by James Knight)
175 * fixed bug: toplevel LOCALLY forms with declarations could
176 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
177 * fixed bug: printing from several different threads using different
178 values of *print-case* could cause invalid output, due to
179 some internal special variables of the printer not being bound
180 thread-locally (reported by Max Mikhanosha)
181 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
182 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
183 (reported by Pascal Costanza)
184 * fixed bug: FINALIZE-INHERITANCE is called from
185 REINITIALIZE-INSTANCE on classes when the class has previously
186 been finalized, as required by AMOP.
187 * minor code generation optimizations:
188 ** better register allocation in CLOS dispatching functions
189 ** overflow detection when coercing signed bytes to fixnums on x86-64
190 ** is now implemented with one IMUL instruction instead of three shifts
191 ** more efficient bit-vector access on x86 and x86-64
192 ** more efficient access to raw structure slots on x86 and x86-64
193 * fixed some bugs revealed by Paul Dietz' test suite:
194 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
197 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
198 * feature: thread support on Solaris/x86, and experimental thread support
200 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
202 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
203 default method for SLOT-UNBOUND.
204 * minor incompatible change: prevent the user from specializing the
205 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
206 somewhat counter to the intent of the protocol, I (CSR) think, and
207 additionally it just doesn't work in SBCL as currently
208 implemented, thanks to optimizations (that are always valid for
209 the other three SLOT-VALUEish functions, but not for the setter).
210 * minor incompatible change: the :SB-LDB feature is now enabled by
211 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
212 the low-level debugger.
213 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
214 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
215 down to an SB-SYS:FD-STREAM, allowing direct communication in
216 more cases. Temporary files and pipes are still used as a fallback
218 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
219 on the non-reentrant C functions localtime(3) and gmtime(3).
220 * bug fix: global optimization policy was not visible in LOCALLY and
222 * bug fix: class objects can be used as specializers in methods.
223 (reported by Pascal Costanza)
224 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
225 with a type error. (reported by blitz_ on #lisp)
226 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
227 failed with a type error. (reported by Pascal Bourguignon)
228 * bug fix: merging pathnames against defaults with :DIRECTORY
229 starting with '(:RELATIVE :BACK) should preserve the :BACK.
230 (reported by James Y Knight)
231 * bug fix: saving large (>2GB) cores on x86-64 now works
232 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
233 constant index and a (simple-array (signed-byte 32)) array
234 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
235 error (patch by Robert J. Macomber)
236 * feature: TIME now displays the amount of run-time spent in GC
237 * bug fix: The debugger now does a better job of respecting
238 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
239 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
240 workaround for bug 403.)
241 * bug fix: floating point exception handling now works on FreeBSD
242 (thanks to NIIMI Satoshi)
243 * fixed some bugs revealed by Paul Dietz' test suite:
244 ** MISC.641: LET-conversion were not supposed to work in late
246 * improvements to the Win32/x86 port:
247 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
248 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
250 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
251 * new feature: source path information is generated for macro-expansion
252 errors for use in IDE's like Slime (thanks to Helmut Eller)
253 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
254 * bug fix: compiler does not loop forever on an invalid type in
256 * improvement: compilation of most CLOS applications is significantly
258 * optimization: added a limited bytecode compiler for simple toplevel
259 forms, speeding up compilation and FASL loading
260 * bug fix: the statistical profiler now properly distinguishes anonymous
263 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
264 * minor incompatible change: in sbcl-0.9.11 (but not earlier
265 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
266 into subclasses of the finalized class. Now user calls to
267 FINALIZE-INHERITANCE finalize just the one class, and calls by the
268 system return before any subclasses are finalized.
269 * minor incompatible change: The reader no longer ignores errors
270 regarding non-existent packages in #+ and #- feature tests.
271 * new feature: command line options --no-sysinit, --no-userinit to
272 inhibit loading the corresponding init files
273 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
274 for sending data through UDP sockets (thanks to François-René Rideau)
275 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
276 error is signalled instead (thanks to François-René Rideau)
277 * minor incompatible change: Error signalling behaviour of lexical
278 operations violating package locks has changed slightly. Refer to
279 documentation on package locks for details.
280 * enhancement: EVAL can process IF-expressions without resorting to the
282 * bug fix: Some operations involving SIN, COS, or TAN weren't being
283 constant-folded properly due to missing out-of-line functions.
284 (reported by Mika Pihlajamäki)
285 * bug fix: LISTEN sometimes returned T even in cases where no data was
286 immediately available from the stream
287 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
288 were reversed. (reported by Levente Mészáros)
289 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
290 when given a vector displaced to an adjustable array. (reported
292 * bug fix: derivation of float boundaries from numbers outside the
293 appropriate float range (reported by John Wiseman)
294 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
296 * bug fix: precomputing cache entries for generic functions with
297 some subclasses of specializers as yet invalid does not attempt to
298 fill a cache line with a negative offset. (reported by Levente
300 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
301 allow condition class objects as arguments to DOCUMENTATION and
302 (SETF DOCUMENTATION); only find and set documentation for
303 structure names for the STRUCTURE doc-type. (suggested by Gary
305 * improvements to the Win32/x86 port:
306 ** (user-homedir-pathname) and default initialization file
307 locations now know about the user's "Documents and Settings"
308 directory (thanks to Yaroslav Kavenchuk)
309 ** run-program is implemented (thanks to Mike Thomas)
310 ** sockets support (thanks to Timothy Ritchey)
311 ** better backtrace support (thanks to Alastair Bridgewater)
312 ** sb-grovel supported
313 ** asdf-install and sb-posix work somewhat
314 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
315 * improvements to the Solaris/x86 port:
316 ** works on Solaris 11/Solaris Express
317 ** floating-point exception handling support
318 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
319 * fixed some bugs revealed by Paul Dietz' test suite:
320 ** REMOVE-METHOD returns its generic function argument even when
321 the method is not one of the generic functions' methods.
322 ** objects declared as MEMBER types can be admissible arguments to
324 ** printing characters should simply be printed by the FORMAT ~:C
326 ** compiler failure when compiling functions with hairy constant
327 defaults for optional parameters.
328 ** compiler produces wrong code when MAYBE-INLINE-expanding a
329 function, which is already optimized.
331 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
332 * new platform: experimental support for SBCL x86/Darwin, including
333 MacOS X 10.4.5 on Intel.
334 * new feature: Unicode character names are now known to the system
335 (through CHAR-NAME and NAME-CHAR).
336 * new feature: the filesystem location of SBCL's core file is
337 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
338 * minor incompatible change: the contrib modules SB-POSIX and
339 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
340 this change is to make it easier to distribute
341 location-independent binaries.
342 * bug fix: as implied by AMOP, standardized classes no longer have
343 slots named by external symbols of public packages. (reported by
345 * optimization: calling functions via constant symbols -- as in
346 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
347 via the function object as in (FUNCALL #'FOO).
348 * optimization: on x86 and x86-64, the calling convention for the
349 case of dealing with an unknown number of values has been altered
350 to be friendlier to the prediction heuristics implemented,
351 particularly with reference to CALL and RET pairing. (thanks to
352 Alastair Bridgewater)
353 * optimization: on x86, the code for access to thread-local symbol
354 values and binding/unbinding of thread-local symbols is smaller.
355 (thanks to Alastair Bridgewater)
356 * enhancement: CONSTANTP is now able to determine constantness of
357 more complex forms, including calls to constant-foldable standardized
358 functions and some special forms beyond QUOTE.
359 * fixed bug: occasional GC crashes on Solaris/x86
360 * optimization: x86-64 supports stack allocation of results of simple
361 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
362 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
363 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
364 (thanks to James Knight)
365 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
366 by the standard (thanks to Stephen Compall)
368 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
369 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
370 be used for bundling the runtime and the core file into one
371 executable binary. This feature is not currently supported on all SBCL
372 platforms. (thanks to James Bielman and NIIMI Satoshi)
373 * new feature: a generational or ephemeral garbage collector is now
374 the default on the PowerPC platform (both Linux and Darwin). The
375 old Cheney (stop and copy) collector is a build-time option.
376 (thanks to Cyrus Harmon, after Raymond Toy)
377 * minor incompatible change: the method by which SBCL finds its
378 contributed modules has changed; it no longer relies on symbolic
379 links from an $SBCL_HOME/systems directory, but searches directly
380 in the subdirectories of $SBCL_HOME.
381 * enhancement: the dynamic heap size on the Linux/PPC platform is
382 markedly larger, even using the older Cheney garbage collector.
383 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
384 certain complicated string types. (reported by Gary King)
385 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
386 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
387 * fixed bug: DOCUMENTATION on structures no longer signals an
388 error if no structure type of the right name exists. (reported by
390 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
391 applicable non-standard primary or :AROUND method on
392 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
393 no longer cause unbound variable errors. (reported by Kevin Reid)
394 * fixed bug: metacircle resolution in cases where methods have slots
395 added before the slots from STANDARD-METHOD. (reported by Jean
397 * fixed bug: the Power PC assembly code for calling into lisp
398 sometimes computed its return address wrongly (depending on the
399 15th bit of the address, fixed at link time). This used to
400 manifest itself in a segmentation violation while building PCL.
401 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
402 many others over the years)
403 * fixed bug: ROOM no longer reports silly numbers for stack usage if
404 the stack pointer (treated as a signed integer) is negative.
405 (thanks to Peter van Eynde)
407 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
408 * new platform: experimental support for the Windows operating
409 system has been added. (thanks to Alastair Bridgewater)
410 * fixed several bugs in and robustified the PPC FFI (including
411 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
412 * bug fix: allow non-simple string symbol names (reported by Paul
414 * bug fix: interrupt handling on NetBSD (thanks to Richard M
416 * bug fix: saving a core corrupted callbacks on x86/x86-64
417 * bug fix: closed a loophole in metacircularity detection and
418 grounding in the PCL implementation of CLOS.
419 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
420 (reported by tomppa on #lisp)
421 * bug fix: clear the direction flag on Lisp -> C transitions, as
422 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
423 (reported by Andrej Grozin and Hendrik Maryns)
424 * optimization: major improvements to GC efficiency on GENCGC platforms
425 * optimization: faster implementation of EQUAL
426 * optimization: emit more efficient opcodes for some common
427 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
428 * optimization: several other minor code-generation improvements on x86-64
429 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
431 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
432 * minor incompatible change: (SETF CLASS-NAME) and (SETF
433 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
434 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
435 by ANSI as a generic function, and for consistency (SETF
436 GENERIC-FUNCTION-NAME) is treated likewise.
437 * fixed bug #233.b: make constraint propagation notice when a variable
438 value is changed after it is referenced but before it is used
439 * fixed bug #296: no more arbitrary behaviour from filenames with
440 odd characters as --load arguments. (reported by Adam Warner)
441 * fixed bug #390: :CHARACTER-SET pathname components now work as
442 expected. (reported by Tim Daly Jr)
443 * fixed bug #391: complicated :TYPE intersections in slot
444 definitions no longer cause an error in PCL internals.
445 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
446 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
447 returning the number of octets which would be written to the
448 file-stream. (thanks to Robert J. Macomber)
449 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
450 arguments to the internal stream functions. (thanks to David
452 * bug fix: stack exhaustion is now handled on the x86/SunOS
453 platform. (thanks to Daisuke Homma)
454 * bug fix: more accurate ROOM results on GENCGC platforms
455 * optimization: improved type inference for arithmetic-for
456 index variables in LOOP
457 * optimization: faster floating-point SQRT on x86-64
458 * fixed some bugs revealed by Paul Dietz' test suite:
459 ** DOCUMENTATION returns NIL instead of "" for method combinations
460 that don't have a docstring
462 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
463 * minor incompatible change: (SETF CLASS-NAME) and (SETF
464 GENERIC-FUNCTION-NAME) are no longer generic functions, and
465 therefore are not customizeable by user code (as seems to be at
466 least permitted and maybe required by AMOP). As a consolation,
467 however, the SBCL implementation of these functions now calls
468 REINITIALIZE-INSTANCE as specified by AMOP.
469 * bug fix: slot-definition documentation is propagated as per ANSI
470 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
471 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
472 argument, as do eslotd initialization methods. (from Pascal
473 Costanza's "Closer" project)
474 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
475 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
477 * bug fix: it is now possible to have more than one subclass of
478 STANDARD-GENERIC-FUNCTION without causing stack overflow.
479 (reported by Bruno Haible, Pascal Costanza and others)
480 * bug fix: the dependent update protocol now works for generic
481 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
483 * bug fix: condition-class instances corresponding to
484 DEFINE-CONDITION forms are now created eagerly. (reported by
485 Kalle Olavi Niemitalo on comp.lang.lisp)
486 * bug fix: floating point printing is more accurate in some
487 circumstances. (thanks to Simon Alexander)
488 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
489 merged with *DEFAULT-PATHNAME-DEFAULTS*.
490 * bug fix: callbacks on OS X now preserve stack-alignment.
491 (thanks to Cyrus Harmon)
492 * enhancement: the x86-64 disassembler is much better at
493 disassembling SSE instructions. (thanks to Lutz Euler)
494 * enhancement: CHECK-TYPE on a variable now allows type-inference
495 (thanks to Vincent Arkesteijn)
496 * optimization: improved performance of EUC-JP external format.
497 (thanks to NIIMI Satoshi)
498 * optimization: performance improvements to IO on file streams of
499 :ELEMENT-TYPE CHARACTER
500 * optimization: much faster memory allocation on x86-64
501 * optimization: faster garbage collections (latency and throughput)
503 * optimization: faster bignum subtraction and division on x86 and x86-64
504 (thanks to Lutz Euler)
505 * optimization: more accurate type-inference for loops that use a
506 floating point index variable or a negative step.
508 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
509 * bug fix: add a workaround to SBCL looping infinitely at startup on
510 Linux kernels with apparently buggy implementations of personality().
511 (thanks to Svein Ove Aas)
512 * bug fix: Unicode symbols are correctly printed in LDB backtraces
513 (thanks to David Lichteblau)
514 * bug fix: local bindings shadow global symbol macros for
515 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
516 * optimization: non-open coded uses of numeric comparison operators
517 (e.g. >) no longer cons when called with more than one parameter
518 on platforms supporting dynamic-extent allocation.
519 * enhancement: saving cores with foreign code loaded is now supported
520 on MIPS/Linux in addition to the previously supported platforms.
521 * bug fix: FIND-RESTART now tests for activity, not applicability when given
522 a restart object as identifier. (reported by Helmut Eller for CMUCL)
523 * bug fix: division by zero in sb-sprof when no samples were collected
524 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
525 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
526 (thanks to Alexander Kjeldaas)
527 * new feature: ignoring the return values of destructive functions
528 like NREVERSE signals a compile-time style-warning.
529 (thanks to Kevin Reid)
530 * new feature: the alignment of alien structure fields can be
531 explicitly specified. (thanks to Cyrus Harmon)
532 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
533 and dump core on SIGQUIT
535 ** incompatible change: threads do not inherit values of specials
536 from their parents (see manual)
537 ** bug fix: threads stacks belonging to dead threads are freed by the
538 next exiting thread, no need to gc to collect thread stacks anymore
539 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
540 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
541 ** bug fix: dynamic variable and thread start related gc lossage
542 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
544 * fixed some bugs revealed by Paul Dietz' test suite:
545 ** SUBTYPEP is slightly more accurate on heinously complicated
546 CONS types where some of the members have uncertain (in the
547 NIL, NIL sense) type relationships to each other.
549 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
550 no longer purified unless :PURIFY T is explicitly specified.
551 ** Non-purified cores are significantly smaller than before
553 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
554 * new feature: timers based on Zach Beane's excellent timer package
555 * added support for the following external formats: koi8-u,
556 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
557 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
558 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
559 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
560 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
561 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
562 * incompatible change: a threaded SBCL will no longer revert to
563 non-threaded mode on non-NPTL systems, but refuse to start entirely.
564 * bug fix: interrupts are disabled until startup is complete; no
565 more sigsegvs when receiving a signal to soon
566 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
568 * bug fix: add a workaround for the memory randomization features in
569 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
570 workaround will only be in effect on systems with the proc filesystem
572 * bug fix: printing objects of type HASH-TABLE signals a
573 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
575 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
576 have been running for over 50 days. (reported by Gilbert Baumann)
577 * bug fix: the logic for getting names of functions gets less
578 confused when confronded with alternate-metaclass
579 funcallable-instances. (reported by Cyrus Harmon)
580 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
581 consistent, even on internal alternate-metaclass objects.
582 * bug fix: flush closure information collected by physical
583 environment analysis prepass before the main pass. (bug reported
585 * bug fix: compiler pack phase does not modify a hashtable, which is
586 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
587 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
588 may be expanded. (reported by Alexander <malishev> on c.l.l)
589 * bug fix: no more style-warnings when compiling code using LAST
590 under high-SPEED low-SPACE settings. (reported by David Wallin
592 * fixed some incompatibilities between SBCL's MOP and the MOP
594 ** the METAOBJECT class is now implemented;
595 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
596 STANDARD-OBJECT, as required;
597 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
598 now compatible (as required by VALIDATE-SUPERCLASS); there
599 remains a consistency requirement over the presence or absence
600 of the FUNCTION class in the superclass list of finalized
601 classes; see the manual for more details;
602 ** the :AROUND method for COMPUTE-SLOTS on
603 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
604 requested slot ordering.
606 ** bug fix: parent thread now can be gc'ed even with a live
608 ** bug fix: binding a special with PROGV to no value is not
609 the same as that symbol not having been bound (thanks to
611 ** bug fix: binding specials is thread safe (thanks to
613 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
614 and SB-SPROF do not die with 'no handler for signal XX in
615 interrupt_handle_now(..)' anymore
616 ** bug fix: WITH-TIMEOUT works with multiple running threads
617 * fixed some bugs revealed by Paul Dietz' test suite:
618 ** ENSURE-GENERIC-FUNCTION should take a method class object for
619 the :method-class keyword argument.
621 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
622 * new port: the Solaris operating system on x86 processors is now
623 mostly supported, though some rough edges in the environment
624 remain. (thanks to Daisuke Homma)
625 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
626 space for its heap. (thanks to Thiemo Seufer)
627 * optimization: SBCL on MIPS platforms now supports dynamic-extent
628 closures. (thanks to Thiemo Seufer)
629 * minor incompatible change: eof selects abort in the debugger.
630 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
631 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
633 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
634 deprecated, and will go away in a future revision of SBCL.
635 * minor incompatible change: GC-ON and GC-OFF are no longer
636 implemented with a counter, it does not matter how many times gc
637 is switched on or off
638 * bug fix: discriminating functions for generic function classes
639 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
640 make invalid assumptions about method precedence order. (reported
642 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
643 platforms. (thanks to Thiemo Seufer)
644 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
645 * bug fix: correct pathname printing with printer escaping is on.
646 (thanks to Kevin Reid)
647 * bug fix: complex VOP definitions in "user-space" no longer trigger
648 package locks. (reported by Zach Beane)
649 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
650 for funcallable-instance functions, and is no different from
652 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
653 characters in the current readtable. (reported by Nicholas Neuss)
654 * bug fix: SUBTYPEP on various CONS types returns more a more
655 accurate acknowledgment of its certainty.
656 * optimizations: REMOVE-DUPLICATES now runs in linear time on
657 lists in some cases. This partially fixes bug 384.
658 * flush all standard streams before prompting in the REPL and the
660 * bug fix: signal handling and triggering gc do not conflict
661 directly anymore, in particular a high frequency sb-sprof does
662 not prevent gc from running
663 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
664 approximation for timezone and DST information between the
665 universal time epoch and the smallest negative 32-bit time_t.
666 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
667 year 1899 so long as the encoded time is non-negative.
668 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
669 the correct answer for non-constant rotate amounts when performing
670 an inline 32-bit rotation.
672 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
673 there is only one thread in the session
674 ** bug fix: memory leak for streams created in one thread and
675 written to in another
676 ** bug fix: lockup when compiled with gcc4
677 ** bug fix: race that allows the gc to be triggered when gc is
679 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
680 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
682 ** bug fix: finished threads are gc'ed properly
683 * fixed some bugs revealed by Paul Dietz' test suite:
684 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
685 has arguments with a different set of applicable methods from
686 the orignal arguments.
687 ** The type error thrown by MAP now has a correct expected-type
689 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
690 name a compiled function.
691 ** Three MISC tests where a large bignum was improperly coerced to
692 a float that couldn't represent that bignum during type
693 derivation were fixed.
694 ** SUBTYPEP can now handle types involving the negation of a
695 list-form FUNCTION type.
696 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
697 as (COMPLEX (AND RATIO (NOT FIXNUM))).
698 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
700 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
701 * New feature: Experimental support for bivalent streams: streams
702 opened with :element-type :default now allow character and binary
703 (unsigned-byte 8) I/O
704 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
705 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
707 * bug fix: no more highly sporadic "couldn't check whether ~S is
708 readable" when reading a stream and an interrupt hits in the middle
709 of a select system call
710 * compiler better recognizes complex arrays (reported by Eduardo
712 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
713 for variables on SBCL built with linkage-tables. (reported by Luis
715 * various error reporting improvements.
716 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
717 (Thanks to Thiemo Seufer)
718 * bug fix: nested reader invokations work correctly
719 * bug fix: it is possible to have simultaneous references to foreign
720 code and foreign data with the same name.
722 ** added x86-64 support
723 ** incompatible change: the threading api now works with thread
724 objects instead of thread ids
725 ** bug fix: threads are protected from signals and interruption when
726 starting up or going down
727 ** bug fix: a race where an exiting thread could lose its stack to gc
728 ** bug fix: don't halt on infinite error in threads if possible
729 ** fixed numerous gc deadlocks introduced in the pthread merge
730 ** bug fix: fixed thread safety issues in read and print
731 ** bug fix: debugger doesn't hang on session lock if interrupted at
732 an inappropriate moment
733 ** bug fix: run-program is now thread safe(r)
734 ** bug fix: inner with-recursive-lock no longer releases the mutex
735 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
736 * fixed some bugs revealed by Paul Dietz' test suite:
737 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
738 EXPECTED-TYPE slots filled.
739 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
741 ** MAKE-CONDITION accepts classes as type-designators.
742 ** COMPILE may never return NIL.
743 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
744 range before calling Unix time functions
746 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
747 * numerous signal handling fixes to increase stability
748 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
749 * minor incompatible change: we now correctly canonize default
750 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
751 as per the MOP, rather than the historical (INITARG INITFUNCTION
752 INITFORM). (reported by Bruno Haible)
753 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
754 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
755 for more information.
756 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
757 pathname is a directory pathname.
758 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
759 * dynamic space size on PPC has been increased to 768Mb. (thanks to
761 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
762 :METACLASS argument in addition to a class name. (reported by
763 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
764 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
765 (reported by Bruno Haible; patch from Gerd Moellmann)
766 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
768 * bug fix: more cleanups to the floating point exception handling on
769 x86-64 (thanks to James Knight)
770 * bug fix: the compiler does not try to constant fold calls of
771 COERCE and BIT-* functions when they return freshly constructed
772 objects. (reported by Nikodemus Siivola and Paul Dietz)
773 * optimization: improved the MIPS versions of generic (in the
774 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
775 * optimization: direct conversion of (unsigned-byte 32) to floats on
776 the PowerPC platform.
777 * optimization: structure instances with raw slots now use less
778 memory, and probably show better memory locality. (thanks to
780 * optimization: DYNAMIC-EXTENT declarations for lists and closures
781 are treated as requests for stack allocation on the x86-64,
782 Alpha-32, PPC and SPARC platforms.
783 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
784 the wrong file descriptor; implementation of SOCKET-OPEN-P.
785 (thanks to Tony Martinez)
787 ** gcing a dead thread can no longer lead to lockups
788 ** threads block signals until they are set up properly
789 ** errno is no longer shared by threads
790 ** interrupt-thread restores the eflags register on x86
791 ** fixed some lockups due to gc/thread interaction
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** invalid dotted lists no longer raise a read error when
795 ** signal an error if a symbol that names a declaration is used
796 as the name of a type, or vice versa
797 ** allow using the (declare (typespec var*)) abbreviation for
798 (declare (type typespec var*)) with all type specifiers
799 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
800 ** always return NIL from PPRINT-POP when OBJECT is NIL
801 ** don't signal errors when pretty-printing malformed LABELS,
802 FLET or MACROLET forms
803 ** declarations in a DOLIST body are also in scope for the
805 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
807 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
810 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
811 * fixed cross-compiler leakages that prevented building a 32-bit
812 target with a 64-bit host compiler.
813 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
814 opened with :IF-EXISTS :OVERWRITE.
815 * fixed bug 281: error for an invalid qualifier in a short-form method
816 combination method is not signalled until the faulty method is called.
817 * bug fix: iteration variable type inferrer failed to deal with open
818 intervals. (reported by Alan Shields)
819 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
820 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
821 or more runtime options were provided to the sbcl binary.
822 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
823 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
824 * optimization: there's now a fast-path for fixnum arguments in the
825 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
826 * optimization: the code generated on x86-64 is more compact thanks
827 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
828 * bug fix: floating-point exception handling now partly works on
829 x86-64. (Thanks to James Knight)
830 * improvement to the MIPS backend from Thiemo Seufer: C-style
831 64-bit long long arguments and return values to alien functions
833 * fixed some bugs revealed by Paul Dietz' test suite:
834 ** the type-error signalled from WARN has a filled-in DATUM slot.
835 ** the type-error required when a stream is not associated with
836 a file has the stream as its datum.
837 ** type-errors on single-floats on x86-64 no longer have
838 :INVALID-OBJECT as the datum
839 ** the type-errors signalled for invalid function names now have
840 a correct expected type
841 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
842 for structures defined with a :TYPE. Documentation strings for
843 typed structures are no longer immediately discarded
844 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
845 broadcast streams and synonym streams. FILE-LENGTH now also works
846 on broadcast streams.
848 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
849 * incompatible change: the --noprogrammer option, deprecated since
850 version 0.7.5, has been removed. Please use the equivalent
851 --disable-debugger option instead.
852 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
854 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
855 inoperational for a while now) has been completely removed.
856 * null lexical environments are now printed as #<NULL-LEXENV>,
857 significantly reducing the amount of clutter in typical backtraces.
858 * documentation on weak pointers, finalization, and after GC hooks
859 has been added to the manual.
860 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
861 as other specialized array types, is much faster. SUBSEQ and
862 COPY-SEQ on such arrays have also been sped up.
863 * optimization: EQL is now more efficient when at least other argument
864 is known to be of type (OR FIXNUM (NOT NUMBER)).
865 * fixed bug: the runtime is now less vulnerable to changes in the
866 size of the SBCL object on OS X, and virtual memory is reserved for
867 all spaces that need to be at a fixed address.
868 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
869 * fixed bug: finalizers and after GC hooks that cause consing are now
871 * fixed bug: compiler error messages and summaries are now printed to
872 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
873 * fixed inference of the upper bound of an iteration variable.
874 (reported by Rajat Datta).
875 * fixed bug 211e: calling local functions with duplicated constant
876 keyword argument no longer causes a bogus style warning about an
878 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
879 ftype declaration to be disregarded. (reported by Dave Roberts)
880 * fixed bug 373: caused by erronous compilation of references to alien
881 variables in the runtime on ppc/darwin.
882 * fixed bug 376: CONJUGATE type deriver.
883 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
884 (reported by Baughn on #lisp)
885 * a cross-compiler bug on non-x86 platforms has been identified and
886 fixed. (thanks to Bruno Haible)
887 * improvements to the MIPS runtime code for increased stability.
888 (thanks to Thiemo Seufer)
889 * increased the maximimum compact environment size to allow
890 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
891 * partial workaround for bug 135: don't name the function we're
892 calling for hairy cases of EVAL, so as not to accumulate one environment
893 entry per call to EVAL. (reported by Kevin Reid)
894 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
895 * optimization: type testing for non-vector arrays should be faster.
896 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
898 * bug fix: setting 31st element of a bit vector to zero did not work
900 * bug fix: redefining a class definition which failed due to a
901 previous accessor / function clash now works (but see BUGS entry
902 #380 for more problems in this area). (thanks to Zach Beane)
903 * the long form of DEFINE-METHOD-COMBINATION disables method group
904 checking when given a single method group with pattern *. (thanks
906 * on x86 compiler supports stack allocation of results of simple
907 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
908 * support for building and running on Mac OS X 10.4 "Tiger" has been added
909 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
910 * fixed some bugs related to Unicode integration:
911 ** the restarts for recovering from input and output encoding
912 errors only appear when there is in fact such an error to
914 * fixed some bugs revealed by Paul Dietz' test suite:
915 ** MISC.549 and similar: late transformation of unsafe type
916 assertions into derived types caused unexpected code
918 ** SCALE-FLOAT type deriver is less wrong.
919 ** type derivers for EXP, LOG and similar functions compute result
920 types for complex arguments better.
921 ** (MISC.563) CONJUGATE type deriver works for very restricted
923 ** out-of-line type testers for character strings are available.
924 ** EQUAL compiler transform understands specialness of objects
926 ** accessing double-floats stored on the stack now works on x86-64.
927 ** debugger internals could sometimes create invalid lispobjs,
928 resulting in GC crashes.
929 ** MISC.548: type check weakening can convert required type into
931 ** initialization forms for bindings are not in scope of free special
934 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
935 * incompatible change: thread support for non-NPTL systems has
936 been removed - locking is buggy and unreliable. A threaded
937 SBCL build will now warn at startup and refuse to create
938 new threads, unless futex support is detected in the kernel
939 * incompatible change: the top level REPL now has only an ABORT
940 restart associated with it, not TOPLEVEL and ABORT as it used to.
941 TOP and TOPLEVEL are now available as debugger commands for
942 returning to the top level.
943 * incompatible change: forms evaluated in the REPL now use the
944 global optimization policy.
945 * incompatible change: user- and system-initialization files are
946 no longer processed with LOAD, but by READ and EVAL; hence the
947 global optimization policy, startup package, readtable, etc,
949 * The .fasl file format number has been incremented because of
950 various incompatible changes.
951 * internal entry point details and argument counts no longer appear
952 in backtraces unless explicitly requested by setting
953 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
954 * built-in and standard functions no longer have names like "top
955 level local call to FOO".
956 * fixed bug 32: functions defined in non-null lexical environments
957 now have more legible printed representation
958 * fixed bug 33: functions defined in non-null lexical environemnts
959 are now more amenable to inspection by INSPECT.
960 * workaround for bug 354: XEPs no longer appear in backtraces unless
961 explicitly requested.
962 * fixed bug: receiving the signal which results from attempting to
963 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
964 SIGBUS on other *BSDs) on architectures where the C stack is also
965 the Lisp stack (x86 and x86-64) from foreign code no longer leads
966 to debugger confusion or wild instability. (reported by Cheuksan
968 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
969 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
970 (reported by Lutz Euler)
971 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
972 compiled code. (reported by Paul Dietz)
973 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
974 specialized parameter to a method to something that is not TYPEP
975 the specializer is now possible.
976 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
977 face of package deletion.
978 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
979 pathname. (reported by David Tolpin on c.l.l)
980 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
981 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
982 STRING-STREAM and FILE-STREAM, respectively.
983 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
984 than previously; in addition, its test suite is now run on build.
985 (thanks to Luke Gorrie)
986 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
987 * optimization: added a immediate representation for single-floats
989 * fixed some bugs revealed by Paul Dietz' test suite:
990 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
991 ** attempting to create a package with a colliding nickname causes
992 correctable errors to be signalled.
993 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
994 ** several x86-64 backend bugs related to sign-extension of immediate
997 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
998 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
999 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1001 * fixed loading of multiply forward-referenced layouts.
1002 (thanks to Cheuksan Wang)
1003 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1004 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1005 (see COMPILE-FILE documentation for details of :PRINT option)
1006 * fixed bugs 19 and 317: fixed-format floating point printing is
1007 more accurate. This also fixes a bug reported by Adam Warner
1008 related to the ~@F format directive.
1009 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1011 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1012 dispatch table if the from-char is a dispatch macro character.
1013 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1014 by the wordsize no longer ignore the last word. (reported by Lutz
1016 * fixed bug in type checking of dynamic-extent variables. (reported
1018 * optimization: sequence traversal functions use their freedom to
1019 coerce function designators to functions.
1020 * optimization: code with many calls to CLOS methods specialized on
1021 CLOS classes has had redundant type checks removed.
1022 * optimization: added declarations to speed up operations that access
1023 the internal character database (for example STRING-UPCASE)
1024 * optimization: comparison operations between floats and sufficiently small
1025 fixnums no longer create extra rationals
1026 * fixed some bugs related to Unicode integration:
1027 ** portions of multibyte characters at the end of buffers for
1028 character-based file input are correctly transferred to the
1029 start of the buffer at the next read.
1030 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1031 passing it through to OPEN.
1032 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1033 argument given to internal calls to OPEN.
1034 ** the built-in comment readers (introduced by character sequences
1035 ";" and "#|") are more forgiving to encoding errors; they will
1036 STYLE-WARN and then attempt to resync the stream at a character
1037 boundary. (thanks to Teemu Kalvas)
1038 * fixed some bugs in the x86-64 port:
1039 ** Negative short int return values from c-calls are sign-extended
1041 ** The stack is aligned to 16-bytes for c-calls, as required by
1042 the ABI. (thanks to Cheuksan Wang)
1043 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1044 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1045 for enums. (thanks to Vincent Arkesteijn)
1046 ** Multiple small optimizations and bugfixes for floating point
1048 * fixed some bugs revealed by Paul Dietz' test suite:
1049 ** Space, Tab, Linefeed, Return and Page have the invalid
1050 secondary constituent character trait.
1051 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1053 ** WITH-INPUT-FROM-STRING should only update the index place on
1055 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1056 works more reliably.
1057 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1058 with one-dimensional arrays with fill pointers.
1059 ** TYPE-OF failed on a complex with an integer realpart and a
1061 ** compiler failure during type inference for the code of form
1062 (IF (EQL X (THE ...)) ...) (MISC.535).
1064 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1065 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1066 platform. The port supports 61-bit fixnums, large memory spaces
1067 and reloading shared object files.
1068 * enhancement: saving cores with foreign code loaded is now
1069 supported on ppc/Darwin in addition to the previously supported
1071 * enhancement: the statistical profiler now walks deeper into the
1072 call stack for more meaningful call-graphs and accrued time
1073 reports (x86/x86-64 only). It also now reports time spent in
1075 * enhancement: it is now possible to trace most individual methods
1076 of a generic function in addition to tracing the generic function
1078 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1080 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1081 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1082 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1083 produces an error. (thanks to Vincent Arkesteijn)
1084 * bug fix: NAMESTRING on pathnames with :WILD components in their
1085 directories works correctly. (thanks to Artem V. Andreev)
1086 * fixed bug 125: compiler preserves identity of closures. (reported
1088 * bug fixed: functions with &REST arguments sometimes failed with
1089 "Undefined function" when compiled with (DEBUG 3). (reported by
1091 * bug fix: overflow during compiling of setting element of a bit
1092 vector with constant index and value. (reported by Timmy Douglas)
1093 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1094 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1095 * on x86 compiler partially supports stack allocation of dynamic-extent
1097 * GO and RETURN-FROM do not check the extent of their exit points
1098 when compiled with SAFETY 0.
1099 * fixed some bugs related to Unicode integration:
1100 ** encoding and decoding errors are now much more robustly
1101 handled; it should now be possible to recover even from invalid
1102 input or output to the terminal. (thanks to Teemu Kalvas)
1103 ** provided a first cut at implementing STRING-TO-OCTETS and
1104 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1105 ** altered the SB-MD5 contributed module slightly, changing the
1106 interface just enough for it to be supportable for builds where
1107 lisp characters are not eight bits.
1108 * fixed some bugs revealed by Paul Dietz' test suite:
1109 ** the FORMATTER-generated functions for ~V[ conditionals require
1110 the correct number of arguments.
1111 ** READ-FROM-STRING returns the mandated second value when applied
1112 to displaced strings.
1113 ** the #\Rubout and #\Backspace characters are treated as invalid
1114 constituent characters by the tokenizer.
1116 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1117 * new feature: reloading changed shared object files with
1118 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1119 * new feature: references to foreign variables and functions
1120 can now be compiled and loaded before the corresponding shared
1121 object file is loaded, as long as the foreign definitions are
1122 available at runtime.
1123 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1124 supported. (thanks to Dan Debertin)
1125 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1126 just keywords. (thanks to Vincent Arkesteijn)
1127 * fixed bug #331: structure-class instances corresponding to
1128 DEFSTRUCT forms are now created eagerly.
1129 * fixed bug #345: backtraces from calls to undefined functions work
1130 on x86 as well. Related bug #61 is now also partially fixed on x86
1131 (backtraces from throws to unknown catch tags.)
1132 * bug fix: lambda-list parsing is now stricter vrt. order and number
1133 of lambda-list keywords.
1134 * bug fix: as specified by AMOP, an error is signalled if a
1135 class-option appears multiple times in a DEFCLASS form. (reported
1137 * bug fix: RANDOM can be compiled when the compiler derives the type
1138 of its numeric argument as a disjoint set of small integers.
1139 (reported by Paul Dietz)
1140 * bug fix: starting a core saved with shared objects loaded when
1141 those objects are not available no longer causes threaded SBCL to
1142 hang. (reported by Sean Ross)
1143 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1145 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1146 argument, which must always be NIL. (reported by Kalle Niemitalo)
1147 * bug fix: printing 1.0d+23 no longer results in an error.
1148 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1149 * bug fix: structure slot setters preserve evaluation order. (thanks
1151 * fixed some bugs related to Unicode integration:
1152 ** RUN-PROGRAM can allow its child to take input from a Lisp
1153 stream. (reported by Stefan Scholl)
1154 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1155 locales. (reported by Ken Causey)
1156 * fixed some bugs revealed by Paul Dietz' test suite:
1157 ** INCF, DECF and REMF evaluate their place form as specified in
1159 ** FORMATTER expands ~{ ~} iteration directives with V or #
1160 parameters correctly.
1161 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1162 consequent uses no arguments correctly.
1163 ** the system has a partial understanding of the (COMPLEX RATIO)
1165 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1168 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1169 * new feature: a build-time option (controlled by the :SB-UNICODE
1170 keyword feature, enabled by default) for building the system with
1171 support for the entire 21-bit character space defined by the
1173 * new feature: the system now has rudimentary external-format
1174 support; the primary user-visible change at this time is that
1175 characters with the high bit set (such as lower-case-e-acute) will
1176 print correctly to a terminal in a UTF-8 environment.
1177 * minor incompatible change: BASE-CHAR no longer names a class;
1178 however, CHARACTER continues to do so, as required by ANSI.
1179 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1180 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1181 has been moved to the SB-EXT package (temporarily re-exported from
1183 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1184 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1186 * fixed bug #351: better error-handlind and reporting for malformed
1188 * fixed bug #350: bignum-printing is now more memory-efficient,
1189 allowing printing of very large bignums, eg. (expt 2 10000000).
1190 (reported by Bruno Haible)
1191 * fixed bug #302: better primitive-type selection for intersection
1193 * fixed bug #308: non-graphic characters now all have names, as
1194 required. (reported by Bruno Haible)
1195 * bug fix: redefining a class with different superclasses now correctly
1196 removes it from the direct-subclasses of its previous superclasses.
1197 (reported by David Morse)
1198 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1199 new value now works. (reported by Bruno Haible)
1200 * bug fix: correct canonicalization of multiple non-standard slot
1201 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1202 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1203 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1204 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1205 now exists, an signals an error.
1206 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1207 during macroexpansion for non-iteration variables as well. (reported
1208 by Bruno Haible for CMUCL)
1209 * bug fix: Cyclic structures and unprintable objects in compiler
1210 messages no longer cause errors. (reported by Bruno Haible)
1211 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1212 and READ-FROM-STRING all now return a primary value of NIL if
1213 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1214 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1215 (reported by Bruno Haible for CMUCL)
1216 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1217 do not designate a condition. (reported by Bruno Haible for
1219 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1220 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1221 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1222 permitting resolution in favour of any of the conflicting symbols.
1223 (reported by Bruno Haible for CMUCL)
1224 * bug fix: EQUAL compiler optimizations is less aggressive on
1225 strings which can potentially compare true despite having distinct
1226 specialized array element types.
1227 * bug fix: unit enumerations can be defined without dividing by
1228 zero. (thanks to Vincent Arkesteijn)
1229 * FORMAT compile-time argument count checking has been enhanced.
1230 (report from Bruno Haible for CMUCL)
1231 * a partial workaround for the bug 262: the compiler does not try to
1232 inline-expand a local function doing RETURN-FROM from a deleted
1233 BLOCK. (thanks to Peter Denno for the bug report and to David
1234 Wragg for the simple test case)
1235 * fixed some bugs revealed by Paul Dietz' test suite:
1236 ** RENAME-PACKAGE allows all package designators as new package
1238 ** constraint propagation and lambda variable substitution are
1239 more cautious in dealing with partially deleted code.
1240 ** compiler sometimes lost reoptimization passes.
1241 ** CERROR, given a condition as condition designator, passes its
1242 remaining arguments to the continue format control without
1244 ** Case-altering FORMAT directives work correctly on non-ASCII
1246 ** The REAL type specifier handles bounds outside the floating
1247 point ranges without signalling FLOATING-POINT-OVERFLOW.
1248 ** Functions with IR1-transformations can create intercomponent
1249 references to global functions.
1250 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1252 ** FORMAT ~R treats a nil value for its first parameter correctly.
1254 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1255 * enhancement: saving cores with foreign code loaded is now
1256 supported on x86/NetBSD and sparc/Linux in addition to the previously
1257 supported platforms.
1258 * bug fix: on some platforms repeated installations caused multiple
1259 copies of HTML documentation to be installed -- should not happen
1260 any more. (reported by Stefan Scholl)
1261 * bug fix: parsing self-recursive alien record types multiple times
1262 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1263 original patch by Helmut Eller for CMUCL)
1264 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1265 (thanks to Richard Kreuter)
1266 * bug fix: defining classes whose accessors are methods on existing
1267 generic functions in other (locked) packages no longer signals
1268 bogus package lock violations. (reported by François-René Rideau)
1269 * bug fix: special variables as DEFMETHOD parameters no longer have
1270 associated bogus type declarations. (reported by David Wragg and
1272 * bug fix: read-write consistency on streams of element-type
1273 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1274 * bug fix: redefiniton of the only method of a generic function with
1275 no DEFGENERIC no longer emits a full WARNING. In addition,
1276 redefinition of generic functions with no DEFGENERIC to an
1277 incompatible lambda list now signals an error. (thanks to Zach
1279 * bug fix: DEFGENERIC now works even when there's a function of the
1280 same name in an enclosing lexical environment. (thanks to Zach
1282 * fixed compiler failure, caused by instrumenting code during
1283 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1284 * optimization: added loop analysis and improved register allocation
1285 * fixed some bugs revealed by Paul Dietz' test suite:
1286 ** POSITION on displaced vectors with non-zero displacement
1287 returns the right answer.
1288 ** (SIMPLE-STRING) is a valid type specifier for sequence
1290 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1292 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1293 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1295 ** The justification version of the FORMAT ~< directive treats
1296 non-zero minpad parameter correctly.
1298 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1299 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1300 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1301 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1302 the supported interface.
1303 * new feature: Single-stepping of code compiled with DEBUG 2 or
1304 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1305 * new feature: saving cores with foreign code loaded is now
1306 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1307 Timothy Moore's work for CMUCL)
1308 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1309 parameters to * instead of NIL if no initform is supplied.
1310 (reported by Johan Bockgaard)
1311 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1312 correctly when FUNCALL appears as the car of the form. Note:
1313 despite this FUNCALL forms are not currently subject to
1314 compiler-macro expansion. (port of Raymond Toy's fix for the
1315 same from CMUCL, reported by Johan Bockgaard)
1316 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1317 (thanks for Teemu Kalvas)
1318 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1319 case instead of terminating the iteration (thanks for Julian
1320 Squires, Sean Champ and Raymond Toy)
1321 * bug fix: incorrect expansion of defgeneric that caused a style
1322 warning. (thanks for Zach Beane)
1323 * bug fix: slot accessor effective method computation works properly
1324 for classes with multiple non-standard applicable methods on
1325 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1326 * on x86 compiler supports stack allocation of results of LIST and
1327 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1328 CMUCL implementation by Gerd Moellmann)
1329 * fixed some bugs revealed by Paul Dietz' test suite:
1330 ** FORMAT strings with both the ~<~:;~> form of the justification
1331 directive and pretty-printing directives cause an error.
1333 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1334 * incompatible change: the internal functions
1335 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1336 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1337 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1338 instead of the old functions.
1339 * new feature: on platforms where "dladdr" is available foreign
1340 function names now appear in backtraces. (based on Helmut Eller's
1342 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1343 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1345 * documentation: ASDF manual is now distributed with SBCL.
1346 * bug fix: SBCL can now load its contributed modules with REQUIRE
1347 even if the system-provided entries have been removed from
1348 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1350 * bug fix: backtraces involving undefined functions or assembly
1351 routines are more informative. (thanks to Brian Downing)
1352 * bug fix: mutually referent alien structures now work correctly.
1353 (reported by Rick Taube)
1354 * bug fix: structures defined by WITH-ALIEN can be referred to
1355 within other definitions in the same WITH-ALIEN.
1356 * bug fix: division operators (MOD, TRUNCATE and the like) with
1357 constant zero divisors and integer dividends no longer generate
1359 * bug fix: provide default methods for INPUT-STREAM-P and
1360 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1361 * bug fix: improve the detection and resolution of MOP metacycles
1362 (where effective method computation depends on the generic function
1363 having its effective method computed). (reported by Bruno Haible)
1364 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1365 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1366 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1367 * bug fix: pathnames with relative directory components can be
1368 represented relative to default pathnames.
1369 * optimization: in taking the GCD of bignums, reduce the two bignums
1370 to approximately the same size (using Euclid's algorithm) before
1371 applying the more sophisticated binary GCD. (thanks to Juho
1373 * optimization: COUNT on bitvectors now operates word-at-a-time.
1374 * optimization: ASH with a positive, but not necessarily constant,
1375 (leftwards) shift, when in a modular context, is compiled to a
1377 * fixed some bugs revealed by Paul Dietz' test suite:
1378 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1379 if the corresponding argument is NIL.
1380 ** FORMAT directives accepting a minpad parameter treat negative
1382 ** literal commas as character parameters to FORMAT directives are
1384 ** literal spaces directly after ~<Newline> directives within a
1385 format-logical-block (~:< ~@:>) do not induce :FILL-style
1386 conditional newlines.
1387 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1388 as specified: it no longer includes conditional newlines).
1389 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1391 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1392 :PER-LINE-PREFIX argument does not evaluate to a string.
1394 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1395 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1396 the manual for details; for now, package locks can be disabled by
1397 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1398 if no major problems are found then it is likely that they will be
1399 compiled in unconditionally.
1400 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1401 unsupported operators on all platforms. To load a shared library
1402 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1403 object file, link it into a shared library outside of SBCL and
1404 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1406 * bug workaround: ROOM T and the GC were not working together
1407 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1408 reporting facility aren't preserved. That reporting has been
1409 disabled, so now until and unless someone figures out how to make it
1410 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1411 * minor incompatible change: as threatened around sbcl-0.8.0, the
1412 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1413 The symbols are also exported from SB-PCL for backwards
1414 compatibility, but more so than before SB-PCL should be treated as
1415 an implementation-internal package.
1416 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1418 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1419 It is known as of this release not to work on the Alpha, however.
1420 * fixed bug #167: errors signalled due to illegal syntax in method
1421 bodies are now more legible.
1422 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1423 designators and can hence be used with TYPEP.
1424 * fixed bug #333: CHECK-TYPE now ensures that the type error
1425 signalled, if any, has the right object to be accessed by
1426 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1427 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1428 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1430 * fixed bug #334: programmatic addition of slots using specialized
1431 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1432 and :ALLOCATION :CLASS slots.
1433 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1434 float exponents. (rereported by Peter Seibel)
1435 * fixed a bug: #\Space (and other whitespace characters) are no
1436 longer considered to be macro characters in standard syntax by
1437 GET-MACRO-CHARACTER.
1438 * fixed bug: initialization of condition class metaobjects no longer
1439 causes an instance of the condition to be created. (reported by
1441 * fixed bug: it is now possible to have slots such that
1442 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1443 description is neither :INSTANCE nor :CLASS.
1444 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1445 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1446 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1447 non-standard SLOT-DEFINITION-ALLOCATION values.
1448 * fixed bugs: various race conditions handling exiting threads.
1449 CL-PPCRE's thread torture test now passes.
1450 * fixed arguably-a-bug: GC time proportional to number of threads in
1451 system even when most of them are idle
1452 * optimization: improved performance of BIT and SBIT on bit-vectors.
1453 * .fasl file incompatibility: The fasl file version number has
1454 been incremented because of changes associated with package locks.
1456 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1457 * minor incompatible change: the system no longer provides
1458 optimization hints (conditions which are TYPEP
1459 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1460 mode; these hints are still emitted when higher SPEED optimization
1462 * new contrib module: a sampling profiler (profiling by statistical
1463 sampling, rather than by instrumenting functions) is available as
1464 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1465 * the behaviour of the standard function ED is now customizeable by
1466 third parties through a hook variable: see ED's documentation
1467 string for information on the protocol.
1468 * the compiler no longer emits efficiency notes for (FUNCALL X)
1469 when the type of X is uncertain under default optimization
1471 * fixed bug 276: mutating a binding of a specialized parameter to a
1472 method to something that is not TYPEP the specializer is now
1474 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1475 exhibit the required behaviour on the x86 platform. (thanks to
1476 Peter van Eynde, Eric Marsden and Bruno Haible)
1477 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1478 even for difficult arguments. (reported by Peter Graves)
1479 * fixed bug 141a: the backquote printer now descends quoted
1481 * fixed another bug in backquote printing: no more destructive
1482 modification of the form's list structure. (reported by Brian
1484 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1485 move between its address being taken and the call to
1486 interrupt_thread, fixing a crashing race condition.
1487 * the SB-POSIX contrib implementation has been adjusted so that it
1488 no longer exhibits ridiculously poor performance when constructing
1489 instances corresponding to C structs.
1491 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1492 * minor incompatible change: the sb-grovel contrib now treats C
1493 structures as alien (in the sense of SB-ALIEN) objects rather than
1494 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1495 has implications for memory management of client code
1496 (sb-grovel-returned objects must now be manually managed) and for
1497 type safety (alien objects now have full types).
1498 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1499 used to control emission of compiler diagnostics, rather than the
1500 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1501 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1502 quality should be considered deprecated.
1503 * install.sh now installs the user manual as well
1504 * (not quite a new documentable feature, but worth considering in
1505 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1506 beginnings of a semantically meaningful condition hierarchy is
1507 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1509 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1510 designator as the defaults argument.
1511 * fixed bug: Displaced arrays whose displaced-to array has become
1512 too small now cause ARRAY-DIMENSION to signal an error, providing
1513 for safer bounds-checking. (reported by Bruno Haible)
1514 * fixed bug: DEFCLASS slot definitions with identical :READER and
1515 :WRITER names now signal a reasonable error. (reported by Thomas
1517 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1519 * fixed bug: Invalid slot specification errors now print correctly.
1520 (thanks to Zach Beane)
1521 * fixed bug 320: Shared to local slot value transfers in class
1522 redefinitions now happen corrently from superclasses as
1523 well. (reported by Bruno Haible)
1524 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1526 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1527 improper lists correctly. (reported by Bruno Haible)
1528 * fixed bug 313: source-transform for <fun-name> was erroneously
1529 applied to a call of a value of a variable with name <fun-name>.
1530 (reported by Antonio Menezes Leitao)
1531 * fixed bug 307: The obsolete instance protocol ensures that
1532 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1533 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1534 remove dead unknown-values globs from the middle of the stack.
1535 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1536 easier to resume long computations after using *BREAK-ON-SIGNALS*
1537 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1538 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1539 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1540 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1541 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1542 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1544 * fixed a (fairly theoretical) bug in string printing: if
1545 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1546 does not have array-element-type equal to the most general string
1548 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1549 function name. (thanks to Nikodemus Siivola)
1550 * fixed bug: compile-time format string checker failed on
1551 non-closed ~{. (reported by Thomas F Burdick)
1552 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1553 don't warn on legal loop code involving more than one aggregate
1554 boolean. (thanks to Nikodemus Siivola)
1555 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1556 in keyword position to MAKE-PACKAGE was not being reported as
1558 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1559 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1561 * fixed bugs 280 and 312: the checking for multiple definitions in a
1562 file is less likely to become confused by uses of inline
1564 * fixed bug: the #S reader macro performs the keyword coercion
1565 specified for slot names. (reported by Kalle Niemitalo)
1566 * fixed bug: lambda lists may contain symbols whose names start with
1567 & but are not lambda-list-keywords; their occurrence triggers a
1569 * fixed bug 321: define-method-combination argument lambda lists do
1570 not cause invalid code to be generated when &optional or &aux
1571 variables have default values. (reported by Bruno Haible)
1572 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1573 have CLOS classes; this appears also to have fixed reported
1574 instability in stack exhaustion detection.
1575 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1576 MUFFLE-WARNING when no associated restart is present is now
1578 * optimization: rearranged the expansion of various defining macros
1579 so that each expands into only one top-level form in a
1580 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1582 * optimization: used a previously little-used slot in symbols to
1583 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1585 * fixed some bugs revealed by Paul Dietz' test suite:
1586 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1588 ** Fixed an optimization bug related to inheritance of initforms
1589 from local to shared slots.
1590 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1591 ** CAST optimizer forgot to flush argument derived type.
1592 ** print/read consistency on floats is now orders of magnitude
1593 more likely. (thanks also to Bruno Haible for a similar report
1595 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1596 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1597 for objects of type REAL. Make it so.
1598 ** SXHASH is specified (infelicitously) to respect similarity,
1599 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1600 it so. (thanks to Markus Ziegler)
1601 ** on the Alpha, the compiler succeeds in compiling functions
1602 returning a known number of arguments greater than 63.
1603 ** fixed handling of invalid NIL arguments in keyword position
1605 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1607 ** provided a readably-printable representation for RANDOM-STATE
1609 ** ensured that pathnames, where they have a namestring, always
1610 print using #P"..." syntax.
1612 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1613 * Support for the forthcoming 2.0 version of the NetBSD kernel
1614 running on x86 hardware has been added. (thanks to Perry
1615 E. Metzger most immediately, and others for their past work)
1616 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1617 current release version; SBCL's OpenBSD support had been broken
1618 since about the time of OpenBSD's switch to ELF binary format.)
1619 * [placeholder for DX summary]
1620 ** user code with &REST lists declared dynamic-extent, under high
1621 speed or space and low safety and debug optimization policy.
1622 * The manual has been converted to Texinfo format and the debugger
1623 chapter from the cmucl manual has been added.
1624 * A facility has been added to extract documentation strings from
1625 sbcl and store them as Texinfo-formatted snippets for inclusion in
1626 the manual (via Texinfo's @include directive)
1627 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1628 the test case to Dave Roberts)
1629 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1630 pointers. (reported by Sean Ross)
1631 * bug fix: PROFILE output is printed nicely even for large numerical
1632 values. (thanks to Zach Beane)
1633 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1634 greater than 32 handle EOF correctly.
1635 * bug fix: on X86 an immediate argument of the IMUL instruction is
1636 correctly printed in disassembly. (thanks to Lutz Euler)
1637 * bug fix: class slots in redefined classes preserve their old
1638 values. (thanks to Bruno Haible and Nikodemus Siivola)
1639 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1640 succeeds. (reported by Marco Baringer)
1641 * fixed some bugs revealed by Paul Dietz' test suite:
1642 ** READ-SEQUENCE now works on ECHO-STREAMs.
1643 ** RATIONALIZE works more according to its specification. (thanks
1646 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1647 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1648 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1649 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1650 should matter to you only if you rebind the printer control
1651 variables and then find you want different bindings in the
1652 debugger than in the ordinary execution of your program.)
1653 * The runtime build system has been tweaked to support building
1654 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1655 assembler and linker. (thanks to Nikodemus Siivola)
1656 * Unbound, undefined, undeclared variables now trigger full
1657 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1658 more often programmer error than deliberate exploitation of undefined
1660 * optimization: the hash algorithm for strings has changed to one
1661 that is less vulnerable to spurious collisions. (thanks to Juho
1663 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1664 needless bounds checking. (thanks to Juho Snellman)
1665 * optimization: implemented multiplication as a modular
1666 (UNSIGNED-BYTE 32) operation on the PPC backend.
1667 * fixed some bugs revealed by Paul Dietz' test suite:
1668 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1670 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1671 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1672 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1674 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1676 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1678 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1679 their output stream on EOF from read.
1680 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1681 have been read to end-of-file.
1682 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1684 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1685 description of determination of which consecutive characters
1687 ** Printing the "Space" character with escaping on now yields "#\\ ",
1688 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1689 ** Reading floating-point numbers with *READ-BASE* set to a number
1690 less than 10 works correctly.
1691 ** Reading floating-point numbers with *READ-BASE* set to a number
1692 more than 10 works correctly.
1693 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1694 the readtable currently in effect.
1696 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1697 * minor incompatible change: parsing of namestrings on a physical
1698 (Unix) host has changed; numbers after the final #\. in a
1699 namestring are no longer interpreted as a version field. This is
1700 intented to be largely invisible to the user, except that the
1701 meaning of the namestring "*.*.*" has changed: it now refers to a
1702 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1703 should usually be replaced by
1704 (make-pathname :name :wild :type :wild :version :wild)
1705 with the added benefit that this is more likely to be portable.
1706 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1707 option to OPEN now signals an error if the file being opened
1708 exists; this may have an impact on existing code.
1709 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1710 signals received, on the PowerPC platforms (both Linux and
1711 Darwin). (thanks to Pierre Mai for pointing out the location of
1713 * several fixes on OS X: The system now builds and runs cleanly on
1714 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1715 (thanks to Brian Mastenbrook)
1716 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1717 second-resolution: integer multiples of 1/3600 between -24 and 24.
1718 (thanks to Vincent Arkesteijn)
1719 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1720 type when called with 1 argument; PEEK-CHAR checked type of
1721 PEEK-TYPE only after having read first character from a
1722 stream. (reported by Peter Graves)
1723 * bug fix: the garbage collector now has much better locality
1724 behaviour, and in particular no longer treats all memory as being
1725 exhausted when many small objects point to each other in a deeply
1727 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1728 recognized as being TYPEP their class.
1729 * bug fix: the PUSHNEW documentation string has been corrected.
1730 (thanks to Vincent Arkesteijn)
1731 * bug fix: defaulting of the value for the last of an atypically
1732 large number of multiple values being bound was not being
1733 performed correctly on the Alpha or PPC platforms
1734 * optimization: implemented multiplication as a modular
1735 (UNSIGNED-BYTE 32) operation on the x86 backend.
1736 * optimization: SEARCH on simple-base-strings can now be open-coded.
1737 (see also contrib/compiler-extras.lisp for inspiration for
1738 teaching the compiler about the Boyer-Moore algorithm).
1739 * value, returned by MAX (and MIN) called with several EQUALP, but
1740 not EQL, arguments now does not depend on compiler settings.
1741 * fixed some bugs revealed by Paul Dietz' test suite:
1742 ** in stack analysis liveness information is propagated from
1743 non-local entry points.
1744 ** pathwise CAST removing failed when the CAST node did not start
1746 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1747 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1749 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1750 host is already defined.
1751 ** RENAME-FILE works on streams instead of signalling an internal
1753 ** PEEK-CHAR uses the current readtable when determining whether
1754 or not a character is whitespace.
1755 ** MERGE-PATHNAMES handles the case when the pathname does not
1756 specify a name while the default-pathname specifies a version.
1757 ** Pathnames now stand a better chance of respecting print/read
1759 ** Attempting to use standardized file system operators with a
1760 pathname with invalid :DIRECTORY components signals a
1762 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1764 ** DIRECTORY on logical pathnames is more correct.
1765 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1766 signal a TYPE-ERROR if their argument is not a stream.
1767 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1768 designator argument does not designate a stream.
1769 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1770 examining the synonym.
1771 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1773 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1774 element-type larger than ([UN]SIGNED-BYTE 32).
1776 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1777 * When built with the :SB-FUTEX feature, threaded builds now take
1778 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1779 for faster/more reliable mutex and condition variable support.
1780 * Incompatible change (but one you probably shouldn't have been using
1781 anyway): the interface and code for arbitrating between multiple
1782 threads in the same user session has been redesigned.
1783 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1784 error when called without an explicit environment argument.
1785 (thanks to Nikodemus Siivola)
1786 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1787 confused about their position. (thanks to Adam Warner and Gerd
1789 * bug fix: Pretty printing backquoted forms with unquotations in the
1790 argument list position of various code constructs such as LAMBDA
1791 now works correctly. (reported by Paul Dietz)
1792 * bug fix: Pretty printing unquotations no longer loses all
1793 stream position information.
1794 * optimization: performance of string output streams is now less
1795 poor for multiple small sequence writes.
1796 * optimization: performance of CSUBTYPEP in the presence of complex
1797 expressions involving CONS and NOT many times has been improved.
1798 (reported by Paul Dietz)
1799 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1801 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1802 streams for the REPL, for the debugger, and so on. (thanks to
1804 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1805 * fixed some bugs revealed by Paul Dietz' test suite:
1806 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1807 is ignored if it is too small, rather than propagating through
1808 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1809 ** extremely complex negations of CONS types were not being
1810 sufficiently canonicalized, leading to inconsistencies in
1812 ** VALUES tranformer lost derived type.
1814 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1815 * fixed a bootstrapping bug: the build process no longer assumes
1816 that the various BOOLE-related constants have the same value in
1817 host and target lisps. (noted by Paul Dietz' test suite on an
1818 SBCL binary built from CLISP)
1819 * The system can now be dynamically linked on the MIPS platform,
1820 which enables dynamic loading of foreign code from Lisp. (thanks
1821 to Ralf Baechle for discussions on the MIPS ABI)
1822 * The system now records debugging information for its own source
1823 files in a filesystem-position-independent manner, relative to
1824 the "SYS" logical host.
1825 * fixed a compiler bug: MV-LET convertion did not check references
1826 to the "max args" entry point. (reported by Brian Downing)
1827 * tweaked disassembly notes to be less confident about proclaiming
1828 some instruction as an LRA. (thanks to Brian Downing)
1829 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1830 now each have their own history, command character, and other
1831 characteristics. (thanks to David Lichteblau)
1832 * fixed some bugs revealed by Paul Dietz' test suite:
1833 ** compiler failure in compiling LOGAND expressions including a
1835 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1836 shift greater than 32.
1837 ** FLUSH-DEST did not mark blocks for type check regeneration.
1838 ** HANDLER-CASE failed to accept declarations in handler clauses
1839 in some circumstances.
1841 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1842 * New code in contrib/sb-introspect (still probably not entirely
1843 stable yet) provides some support for smart Lisp development
1844 environments like SLIME.
1845 * The conditions signalled for errors occurring when loading .fasl
1846 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1847 in a way which should help ASDF recover gracefully.
1848 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1849 stuff has been cleaned up. If you code contrib/ stuff, this might
1850 affect you, and you can look at contrib/README, contrib/STANDARDS,
1851 and/or the 0.8.4.27 diff to check.
1852 * In full calls the compiler now does not generate checks for declared
1853 argument types for all arguments.
1854 * various threading fixes
1855 ** and some experimental patches which didn't make it into
1856 the main tree for this release, but which are shipped in
1857 contrib/experimental-thread.patch as a possible fix for some
1858 failures (deadlock, spinning...) in GC-intensive multithreaded
1860 * fixed PPC build problem (source code incompatibility of different
1861 library versions): added offsetof() hackery which attempts to divine
1862 where glibc maintainers put uc_mcontext today
1863 * fixed bug 282: compiler does not trust type assertions while passing
1864 arguments to a full call.
1865 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1866 &OPTIONAL VALUES type parameter.
1867 * fix bug 214: algorithm for noting rejected templates is now more
1868 similar to that of template seletion. (also reported by rydis on
1870 * fixed bug 141b: printing backquoted information readably and prettily
1871 inserts a space where necessary.
1872 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1873 (CHARACTER 10) are now reported as errors, rather than propagated
1874 as unknown types. (reported by piso on #lisp)
1875 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1876 with values NIL and :ERROR. (thanks to Milan Zamazal)
1877 * fixed bug 191c: CLOS now does proper keyword argument checking as
1878 described in CLHS 7.6.5 and 7.6.5.1.
1879 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1880 longer raise an error; further, using a list as a for-as-arithmetic
1881 counter now raises a meaningful error.
1882 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1883 now understood by sequence creation functions such as MAKE-SEQUENCE
1885 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1886 a STRING-INPUT-STREAM.
1887 * compiler enhancement: SIGNUM is now better able to derive the type
1889 * type declarations inside WITH-SLOTS are checked. (reported by
1891 * fixed some bugs revealed by Paul Dietz' test suite:
1892 ** incorrect optimization of TRUNCATE for a positive first
1893 argument and negative second.
1894 ** compiler failure in let-convertion during flushing dead code.
1895 ** compiler failure while deriving type of TRUNCATE on an
1896 interval, containing 0.
1897 ** ASH of a negative bignum by a negative bignum count now returns
1899 ** intersection of CONS types now canonicalizes properly, fixing
1900 inconsistencies in SUBTYPEP.
1902 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1903 * incompatible change: The --disable-debugger command line
1904 option now clobbers the debugger at a more fundamental
1905 level, by redefining #'INVOKE-DEBUGGER instead of by
1906 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1907 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1908 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1909 enter the debugger REPL and then suffer recursive errors
1910 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1911 while under the new system, BREAK in --disable-debugger mode
1912 terminates the system just as an unhandled error would.
1913 * fixed compiler performance when processing loops with a step >1;
1914 * bug fix: DOCUMENTATION now retrieves generic function
1915 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1916 support has been systematized, and now supports the methods
1917 specified by ANSI, along with a default method and a method for
1918 slot documentation. (reported by Nathan Froyd)
1919 * bug fix: effective methods associated with a generic function are
1920 no longer cached over a change of that generic function's method
1921 combination. (reported by Andreas Fuchs)
1922 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1923 names a non-existent directory. (thanks to Andreas Fuchs)
1924 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1925 the PPC platform, lead to stack corruption; this has been fixed.
1926 (reported by Rainer Joswig)
1927 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1928 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1929 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1930 arithmetic in general, is now much more reliable on the PPC
1932 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1933 platform now returns the right answer.
1934 * optimization: restored some effective method precomputation in
1935 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1936 precomputation is now tunable.
1937 * optimization: compiler-internal data structure use has been
1938 reviewed, and changes have been made that should improve the
1939 performance of the compiler by about 20%.
1940 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1941 simple-base-strings and simple-bit-vectors is improved.
1942 * optimization: the optimization of 32-bit logical and arithmetic
1943 functions introduced in version 0.8.3 on the x86 has been
1944 implemented on the mips, ppc and sparc platforms; an
1945 implementation of the same facility, but for 64-bit arithmetic,
1946 has been added for the alpha.
1947 * microoptimization: the compiler is better able to make use of the
1948 x86 LEA instruction for multiplication by constants.
1949 * bug fix: in some situations compiler did not report usage of
1950 generic arithmetic in (SPEED 3) policy.
1951 * bug 145b fix: compiler used wrong type specifier while converting
1952 MEMBER-types to numeric.
1953 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1955 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1956 index argument now works properly on the Alpha platform.
1957 * bug fix: floating point exception treatment on the Alpha platform
1959 * bug fix: FILE-POSITION works much better on string input and
1960 output streams. (thanks to Nikodemus Siivola)
1961 * bug fix: many threading/garbage collection symptoms sorted.
1962 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1963 might be pseudo-atomic.
1964 * internal change: Stopping for GC is now done with signals not
1965 ptrace. GC is now done in whichever thread wanted it, instead of
1967 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1969 * bug fix: result form in DO is not contained in the implicit
1971 * incompatible change: ICR structure is changed; the value part of
1972 CONTINUATION is now called LVAR; corresponding functions are
1973 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1974 * added type deriver for ISQRT (thanks to Robert E. Brown).
1975 * bug fix: better support for loading from the command line when an
1976 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1978 * fixed some bugs revealed by Paul Dietz' test suite:
1979 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1980 ** ROUND and FROUND now give the right answer when given very
1981 small float arguments.
1982 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1984 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1985 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1986 ** LCM should return a non-negative integer.
1987 ** PARSE-INTEGER returned the index of a terminator instead of the
1988 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1989 ** PARSE-INTEGER returned an incorrect index being applied to a
1991 ** LCM with two arguments of 0 returns 0 rather than signalling
1993 ** unsigned addition of a 32-bit constant with the high bit set no
1994 longer causes an internal compiler error on the x86.
1995 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1996 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1997 with negative last argument.
1998 ** byte specifiers with zero size and position no longer cause
1999 an error during type derivation.
2000 ** bignum multiplication on the Alpha platform now returns the
2002 * porting: The system now builds on SuSE AMD64, although it still
2003 generates a 32-bit binary.
2004 * .fasl file incompatibility: The fasl file version number has
2005 been incremented (because of the changes to internal compiler
2006 data structures referred to above).
2008 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2009 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2010 more accurately, on the Darwin kernel running on PowerPC hardware.
2011 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2012 * Compiler code deletion notes now signal a condition of type
2013 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2014 an associated MUFFLE-WARNING restart.
2015 * The compiler now performs limited argument count validation of
2016 constant format strings in FORMAT, and where appropriate in ERROR,
2017 CERROR and WARN. (thanks to Gerd Moellmann)
2018 * New ASDF-INSTALL contrib can be used for automatic download and
2019 installation of third-party Lisp code from CCLAN or other sites
2021 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2022 which forces another thread to execute a function supplied by the
2024 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2025 now accept and act upon their :ELEMENT-TYPE keyword argument.
2026 (reported by Martin Atzmueller, Edi Weitz)
2027 * bug fix: FILE-POSITION now accepts position designators up to
2028 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2029 is the greater. (thanks to Patrik Nordebo)
2030 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2031 Kalle Olavi Niemitalo)
2032 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2033 printable. (reported by Eric Marsden)
2034 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2035 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2036 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2037 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2038 (thanks to Stig Sandoe)
2039 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2040 UNREAD-CHAR now correctly includes the unread character in the
2041 target sequence. (thanks to Gerd Moellmann)
2042 * bug fix in threaded builds: the system can now be suspended and
2043 resumed by shell job control with minimal disruption.
2044 * bug fixes in times and timezones >2038AD
2045 * better handling of "where is GNU make?" problem in build scripts
2046 (thanks to Nikodemus Siivola)
2047 * new optimization: inside a named function any reference to a
2048 function with the same name is considered to be a self-reference;
2049 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2050 optimization quality.
2051 * new optimization on x86: logical functions and + now have
2052 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2053 used when the result is truncated to 32 bits.
2054 * VALUES declaration is partially enabled.
2055 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2056 * fixed some bugs revealed by Paul Dietz' test suite:
2057 ** The system now obeys the constraint imposed by
2058 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2059 types form a lattice under type intersection.
2060 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2061 ** ASSOC now ignores NIL elements in an alist.
2062 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2063 and (1+ MOST-POSITIVE-FIXNUM) answers.
2064 ** The addition of a method with invalid qualifiers to a generic
2065 function does not cause an error to be signalled immediately;
2066 a warning is signalled, and the error is generated only on
2067 calling the generic function.
2068 * changed .fasl file version number, in part to add type codes for
2069 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2070 obscure ANSI requirements
2072 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2073 * fixed bug 148: failure to inline-expand a local function left
2074 garbage, confusing the compiler.
2075 * fixed bugs 3cd: structure slot readers perform type check if the
2076 slot can have an invalid value (i.e. it is either not initialized
2077 or can be written with a less specific slot writer).
2078 * bug fix: the compiler now traps array references to elements off
2079 the end of an array; previously, the bounds checking in some
2080 circumstances could go off-by-one.
2081 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2083 * type declarations for array element types now obey the description
2084 on the CLHS page "Declaration TYPE", as per discussions on
2085 sbcl-help around 2003-05-08. This means that a declaration
2086 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2087 declaration, all references to BAR will be asserted or assumed
2088 (with THE, so dependent on compiler policy) to involve objects of
2089 type FOO. Note that no such declaration is implied in
2090 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2091 * declared types of functions from the "Conditions"
2092 chapter. (reported by Paul Dietz)
2093 * bug fix: CERROR accepts a function as its first argument.
2094 * bug fix: NTH an NTHCDR accept a bignum as index
2095 arguments. (reported by Adam Warner)
2096 * optimization: character compare routines now optimize comparing
2097 against a constant character. (reported by Gilbert Baumann)
2098 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2099 argument now works properly on the MIPS platform.
2100 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2101 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2102 declared to have a functional type, does not cause an error
2104 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2105 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2106 * fixed bug: if last continuation of a deleted block has a
2107 destination, this destination should be deleted too. (reported by
2109 * fixed a bug in the bootstrap process: the host compiler's values
2110 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2111 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2112 test case from Patrik Nordebo)
2113 * improved the ability of the disassembler on the PPC platform to
2114 provide helpful disassembly notes.
2115 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2116 the class in more cases than previously.
2117 * bug fix: FILE-POSITION now understands :START and :END for
2118 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2119 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2120 8) data. (thanks to David Lichteblau)
2121 * bug fix: it is possible to add a method to a generic function
2122 without lambda list.
2123 * bug fix: reader failed to signal END-OF-FILE inside an
2124 object representation. (reported by Nikodemus Siivola)
2125 * fixed some bugs revealed by Paul Dietz' test suite:
2126 ** LAST and [N]BUTLAST should accept a bignum.
2127 ** condition slot accessors are methods.
2128 ** (VECTOR NIL) is a subtype of STRING.
2130 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2131 * minor incompatible change: some nonsensical specialized lambda
2132 lists (used in DEFMETHOD) which were previously ignored now signal
2134 * minor incompatible change: the system is now aware of the types of
2135 variables in the COMMON-LISP package, and will signal errors for
2136 most violations of these type constraints (where previously they
2137 were silently accepted).
2138 * minor incompatible change: COMPILE-FILE now uses the freedom
2139 afforded (ANSI 3.2.2.3) to use derived function types for
2140 functions defined in the same file. This also permits the system
2141 to warn on static type mismatches and function
2142 redefinition. (Currently it does not work with high DEBUG level.)
2143 * minor incompatible change: VALUES declaration is disabled.
2144 * When issuing notes, the compiler now signals a condition of type
2145 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2146 restart for use in user handlers. It is expected that the
2147 COMPILER-NOTE condition will eventually become a condition
2148 supertype to a hierarchy of note types, which will then be
2149 handleable in a similar fashion. However, at the moment, no such
2150 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2151 but it's an implementation detail, not a classification for the
2153 * Changes in type checking closed the following bugs:
2154 ** type checking of unused values (192b, 194d, 203);
2155 ** template selection based on unsafe type assertions (192c, 236);
2156 ** type checking in branches (194bc).
2157 * A short form of VALUES type specifier has ANSI meaning (it has
2158 increased the number of situations when SBCL cannot perform type
2160 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2161 keywords or constants is permissible.
2162 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2163 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2164 classes. (thanks to Antonio Martinez)
2165 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2166 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2167 argument to a pretty-print stream. Instead, it leaves any such
2168 operations to DESCRIBE-OBJECT methods.
2169 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2171 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2172 lambda lists. (thanks to David Lichteblau)
2173 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2174 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2175 (thanks to Gerd Moellmann)
2176 * fixed bug 246: increased compilation speed of long
2177 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2179 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2181 * DEFCONSTANT now throws a condition of type
2182 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2183 constant to a non-EQL value; CONTINUE and ABORT restarts
2184 respectively change and preserve the value.
2185 * fixed bug 63: The code walker, part of the implementation of CLOS,
2186 is now better at handling symbol macros.
2187 * bug fix: There is no longer an internal implementation type named
2188 CL:LENGTH. (reported by Raymond Toy)
2189 * bug fix: In macro-like defining macros/special operators the
2190 implicit block does not enclose the lambda list.
2191 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2192 enclosed in parentheses) are not suitable as type specifiers, and
2193 their use properly signals an error now.
2194 * bug fix: An argument count mismatch for a type specifier in code
2195 being compiled no longer causes an unhandled error at compile
2196 time, but signals a compile-time warning.
2197 * fixed simple vector readable printing
2198 * bug fix: DESCRIBE takes more care over whether the class
2199 precedence list slot of a class is bound before accessing it.
2200 (reported by Markus Krummenacker)
2201 * bug fix: FORMATTER can successfully compile pretty-printer format
2202 strings which use variants of the ~* directive inside.
2203 * bug fix: SEARCH now applies its TEST predicate to the elements of
2204 the arguments in the correct order. (thanks to Wolfhard Buss)
2205 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2206 in type inference. (thanks to Robert E. Brown)
2207 * bug fix: Reading in symbols with an explicit package name of ""
2208 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2209 * fixed some bugs revealed by Paul Dietz' test suite:
2210 ** NIL is now allowed as a structure slot name.
2211 ** Arbitrary numbers, not just REALs, are allowed in certain
2212 circumstances in LOOP for-as-arithmetic clauses.
2213 ** Multiple class redefinitions before slot access no longer
2214 causes a type error.
2215 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2216 association between the name and a class.
2217 ** Generic functions with non-standard method-combination and over
2218 five methods all of which return constants no longer return NIL
2219 after the first few invocations. (thanks to Gerd Moellmann)
2220 ** CALL-NEXT-METHOD with no arguments now passes the original
2221 values of the arguments, even in the presence of assignment.
2222 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2224 ** DISASSEMBLE works with closures and funcallable instances.
2225 ** ADD-METHOD now returns the generic function, not the new method.
2226 ** FIND-METHOD signals an error if the lengths of the specializers
2227 is incompatible with the generic function, even if the ERRORP
2229 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2230 which its argument is a member.
2231 ** DEFCLASS only redefines the class named by its class-name
2232 argument if that name is the proper name of the class;
2233 otherwise, it creates a new class.
2234 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2235 of the UNBOUND-SLOT condition to the name of the slot.
2236 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2237 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2238 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2239 SLOT-MAKUNBOUND in the specified fashion.
2241 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2242 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2243 cross-compilation host. As a consequence, we can now bootstrap our
2244 way up to SBCL starting with a bare gcc toolchain and human-readable
2245 source code (first the source to CLISP, then the source to SBCL).
2246 * A contributed module containing a partial implementation of the
2247 simple-streams interface has been included. (thanks to Rudi
2249 * A contributed module implementing the RFC1321 Message Digest
2250 Algorithm, known as MD5, has been included.
2251 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2252 no longer has any effect, as the code controlled by this feature
2253 has been deleted. (As far as we know, no-one has ever built using
2254 this feature, and its semantics were confused in any case).
2255 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2256 work on conditions (as required by the ANSI specification),
2257 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2258 expected behaviour on conditions. Users should note, however,
2259 that such behaviour is not required by the ANSI specification,
2260 and so use of this behaviour may render their code unportable.
2261 * fixed some bugs revealed by Paul Dietz' test suite:
2262 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2264 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2265 ** SXHASH on condition objects no longer returns NIL.
2266 ** :ALLOCATION :CLASS slots are better treated; their values are
2267 updated on class redefinition, and initforms inherited from
2268 superclasses are applied.
2269 ** REMOVE-METHOD returns its generic function argument even when
2270 no method was removed.
2271 ** SHARED-INITIALIZE now initializes the values of the requested
2272 slots, including those with :ALLOCATION :CLASS.
2273 ** ALLOCATE-INSTANCE now works on structure classes defined via
2274 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2276 ** SLOT-EXISTS-P now works on conditions, as well as structures
2278 ** MAKE-LOAD-FORM now has the required methods on
2279 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2280 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2281 keyword, and now implements the SLOT-NAMES argument.
2282 ** methods with &OPTIONAL arguments no longer allow too many
2283 arguments to be passed in the call without error.
2284 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2285 option is consistent with the required arguments of the generic
2286 function lambda list.
2287 * bug fix: REQUIRE accepts a string designator. (Thanks to
2289 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2290 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2291 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2292 * bug fix: The compiler now checks for duplicated variables in macro
2294 * bug fix: SETQ on globals returns the correct value.
2295 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2296 not a COMPILER-ERROR (followed by some other strange error on
2297 choosing the CONTINUE restart).
2298 * bug fix: make.sh and friends are now more consistent in the way that
2299 they look for GNU "make".
2301 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2302 * experimental native threads support (on x86 Linux >=2.4 only).
2303 This is not compiled in by default: you need to add :SB-THREAD to
2304 the target features. See the "Beyond ANSI" chapter of the manual
2306 * fix for longstanding nonANSIism: The old distinction between
2307 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2308 The return value from CL:FIND-CLASS is now a CLOS class, and
2309 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2310 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2311 * An interface to the MetaObject Protocol, as described in Kiczales,
2312 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2313 MIT Press, 1991, is available from the SB-MOP package.
2314 * incompatible change: the SB-PCL package should now be considered
2315 a private implementation detail, and no longer a semi-private MOP
2317 * minor incompatible change: due to rearrangement for threads, the
2318 control stack and binding stack are now allocated at arbitrary
2319 addresses instead of being hardcoded per-port. Users affected by
2320 this probably have to be doing advanced things with shared
2321 libraries, and will know who they are.
2322 * minor incompatible change: Previously, all --eval forms used were
2323 processed with READ before any of them were processed with EVAL.
2324 Now each --eval form is processed with both READ and EVAL before
2325 the next --eval form is processed. (Thus package operations like
2326 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2327 work as the user might reasonably expect.)
2328 * minor incompatible change: *STANDARD-INPUT* is now only an
2329 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2331 * minor incompatible change: Y-OR-N-P is now character-oriented,
2332 not line oriented. Also, YES-OR-NO-P now works without errors.
2333 (thanks to Antonio Martinez)
2334 * sb-aclrepl module improvements: an integrated inspector, added
2335 repl features, and a bug fix to :trace command.
2336 * Known functions, which cannot be open coded by the backend, are
2337 considered to be able to check types of their arguments. (fixing
2338 a bug report by Nathan J. Froyd)
2339 * fixed a bug in computing method discriminating functions: It is
2340 now possible to define methods specialized on classes which have
2341 forward-referenced superclasses. (thanks to Gerd Moellmann)
2342 * fixed evaluation order in optional entries (reported by Gilbert
2344 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2345 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2346 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2347 required-by-AMOP NAME argument, as well as CLASS and
2348 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2349 * fixed bug 20: DEFMETHOD can define methods using names that are
2350 not the proper names of classes to designate class specializers.
2351 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2352 Unix file descriptors, instead of blowing up. (thanks to Antonio
2354 * Garbage collection refactoring: user-visible change is that a
2355 call to the GC function during WITHOUT-GCING will not do garbage
2356 collection until the end of the WITHOUT-GCING. If you were doing
2357 this you were probably losing anyway.
2358 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2359 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2360 TYPEP the latter but not the former.
2361 * The compiler issues a full WARNING for calls to undefined functions
2362 with names from the CL package.
2363 * MAP-INTO for a vector destination is open coded. (reported by
2364 Brian Downing on c.l.l)
2365 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2366 documentation string.
2367 * fixed some bugs revealed by Paul Dietz' test suite:
2368 ** COPY-ALIST now signals an error if its argument is a dotted
2370 ** Condition slots are now accessed more correctly in the presence
2371 of multiple initargs for a given slot.
2372 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2373 exclude restarts of the same name associated with a different
2375 ** DEFCLASS of forward-referenced classes with another
2376 forward-referenced class in the superclasses list no longer
2378 ** Condition slots are now initialized once each, not multiple
2379 times. (thanks to Gerd Moellmann)
2380 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2382 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2383 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2384 arguments contain duplicated elements.
2385 ** RESTART-CASE understands local macros.
2386 ** RESTART-CASE associates exactly its own restarts with a condition.
2387 ** ENDP in safe mode checks its argument to be of type LIST.
2388 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2389 in question is unbound.
2390 ** Optimized MAKE-INSTANCE functions no longer cause internal
2391 assertion failures in the presence of duplicate initargs.
2392 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2393 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2394 propagated even in the ctor optimized implementation of
2396 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2398 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2399 * a better implementation of SXHASH on (simple) bit vectors,
2400 measured both in execution speed and in distribution of results
2401 over the positive fixnums, has been installed. Likewise, a better
2402 implementation of EQUAL for simple bit vectors is now available.
2403 * fixed CEILING optimization for a divisor of form 2^k.
2404 * fixed bug 240 (emitting extra style warnings "using the lexical
2405 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2406 by Antonio Martinez)
2407 * fixed SXHASH, giving different results for NIL depending on type
2408 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2409 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2410 variable. (found by Rolf Wester)
2411 * SBCL does not ignore type declarations for special
2412 variables. (reported by rif on c.l.l 2003-03-05)
2413 * some bug fixes in contrib/sb-aclrepl/
2414 * fixed some bugs revealed by Paul Dietz' test suite:
2415 ** a bug in the CONS type specifier, whereby the CAR and CDR
2416 types got intertwined, has been fixed;
2417 ** the type system is now able to reason about the interaction
2418 between INTEGER and RATIO types more completely;
2419 ** APPEND, [N]REVERSE and NRECONC check that those their
2420 arguments, which must be proper lists, are really so;
2421 ** An array specialized to be unable to hold elements has been
2422 implemented, as required -- yes, really -- by ANSI;
2423 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2424 on malformed property lists;
2426 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2427 * incompatible packaging change: in line with Unix convention,
2428 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2429 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2430 * REQUIRE and PROVIDE are now optionally capable of doing something
2431 useful. See the documentation string for REQUIRE.
2432 * infrastructure for a managed SBCL contrib system: contributed
2433 modules in this release include:
2434 ** the ASDF system definition facility;
2435 ** an interface to the BSD Sockets API;
2436 ** an ACL-like convenience interface to the repl;
2437 (thanks to Kevin Rosenberg)
2438 ** an implementation of ROTATE-BYTE, with efficient implementation
2440 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2441 gives a closer approximation to the right answer than previously.
2442 (thanks to Raymond Toy)
2443 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2444 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2445 optional environment arguments, as required by ANSI.
2446 * fixed bugs in other functions taking environment objects, allowing
2447 calls with an explicit NIL environment argument to be compiled
2449 * fixed bug 228: primary return values from
2450 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2451 COMPILE or FUNCTION.
2452 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2453 :TYPE will no longer signal errors on innocuous objects.
2454 * fixed bug 231b: SETQ is better at respecting type declarations in
2455 the lexical environment.
2456 * fixed a bug in DEFCLASS: classes named by symbols with no or
2457 unprintable packages can now be defined.
2458 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2459 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2460 * fixed compiler failure related to checking types of functions.
2461 (reported by Robert E. Brown)
2462 * the compiler is now much more consistent in its error-checking
2463 treatment of bounding index arguments to sequence functions: in
2464 (SAFETY 3) code, errors will be signalled in almost all cases if
2465 invalid sequence bounding indices are passed to functions defined
2466 by ANSI to operate on sequences.
2467 * fixed a bug in the build procedure: documentation of SBCL-specific
2468 packages is now preserved and available in the final Lisp image.
2469 * lifted FDEFINITION lookup out of loops in the implementation of
2470 many list operations. (thanks to Robert E. Brown)
2471 * fixed a bug in the reader: the #n# reader macro now works for
2472 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2473 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2474 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2475 E. Brown and Tony Martinez respectively)
2476 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2477 SLOT-BOUNDP: in particular, it is now able to optimize them much
2478 better, and is now not vulnerable to having packages renamed.
2479 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2480 always called when a slot is not present in an instance. (thanks
2482 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2483 longer signal an error on unreading a character following EOF on
2484 the previous constituent stream. (thanks to Tony Martinez)
2485 * fixed some bugs revealed by Paul Dietz' test suite:
2486 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2487 not just nonnegative fixnums;
2488 ** the logical bit-array operators such as BIT-AND now accept an
2489 explicit NIL for their "opt-arg" argument (to indicate a
2490 freshly-consed result bit-array);
2491 ** ELT now signals an error on an invalid sequence index in safe
2493 ** the type system is now cleverer about negations of numeric
2494 types, and consequently understands the BIGNUM and RATIO types
2496 ** the type system is now cleverer about the interaction between
2497 INTEGER and RATIO types: while bugs still remain, many more
2498 cases are accurately computed;
2499 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2500 if it is in the last clause;
2501 ** CONSTANTLY now correctly returns a side-effect-free function in
2503 ** DECLARE is no longer treated as a special-operator; in
2504 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2505 * incremented fasl file version number due to the change in the
2506 DEFSTRUCT-SLOT-DESCRIPTION structure.
2508 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2509 * minor incompatible change: code processed by the "interpreter" or
2510 EVAL now has a compilation optimization policy of (DEBUG 2)
2511 (changed from (DEBUG 1)) to improve debuggability of interactive
2512 development, and to allow the use of the debug RETURN command in
2514 * an experimental implementation of the RETURN command for the
2515 debugger has been included. (thanks to Frederik Kuivinen)
2516 * fixed bug 62: constraints were not propagated into a loop.
2517 * fixed bug in embedded calls of SORT (reported and investigated by
2519 * fixed some bugs revealed by Paul Dietz' test suite:
2520 ** printing and reading of arrays with some dimensions having
2521 length 0 (thanks to Gerd Moellmann);
2522 ** BOA constructor with &AUX argument without a default value does
2523 not cause a type error;
2524 ** CONSTANTP now returns true for all self-evaluating objects.
2526 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2527 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2528 accessors that are related by inheritance, as specified in the
2529 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2530 Valtteri Vuorikoski)
2531 * The compiler is now able to inline functions that were defined in
2532 a complex lexical environment (e.g. inside a MACROLET).
2533 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2535 * Support for the upcoming FreeBSD-5.0 release has been included.
2536 (thanks to Dag-Erling Smorgrav)
2537 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2538 effect when it is not in a toplevel context.
2539 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2540 stand a better chance of being correct. (thanks to Gerd
2542 * fixed bug in COERCE, which now signals an error on coercing a
2543 rational to a bounded real type which excludes the expected
2545 * The compiler is now able to derive types more accurately from the
2546 COERCE and COMPILE functions.
2547 * fixed bug 223: functional binding is considered to be constant
2548 only for symbols in the CL package.
2549 * fixed bug 231: SETQ did not check the type of a variable being set
2550 (reported by Robert E. Brown)
2551 * A new optimization for MAKE-INSTANCE has been included, fixing
2552 various bugs (including relating to :ALLOCATION :CLASS slots and
2553 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2555 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2556 ** As required by ANSI, LOOP now disallows anonymous collection
2557 clauses such as COLLECT I in conjunction with aggregate boolean
2558 clauses such as THEREIS (= I 1);
2559 ** LOOP now signals an error when any variable is reused in the
2560 same loop (including the potentially useful construct analogous
2561 to WITH A = 1 WITH A = (1+ A);
2562 ** IT is only a special loop symbol within the first clause of a
2563 conditional loop clause;
2564 ** LOOP with a typed iteration variable over a hashtable now
2565 signals a type error iff it should.
2566 * fixed some other bugs revealed by Paul Dietz' test suite:
2567 ** FILE-STREAM now names the class previously known as FD-STREAM;
2568 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2569 argument) no longer signals an error;
2570 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2571 of the slot symbol, rather than using the current package
2572 ((:CONC-NAME "") continues to intern the slot's name in the
2574 * incremented fasl file version number, because of the incompatible
2575 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2576 of the new implementation of DEFINE-COMPILER-MACRO.
2578 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2579 * Support for building SBCL for MIPS platforms running in
2580 little-endian mode has now been checked in, and basic
2581 functionality on said platforms verified.
2582 * minor incompatible change: PCL now records the pathname of a file
2583 in which methods and the like are defined, rather than its
2585 * minor incompatible change: TRUENAME now considers the truename of
2586 a file naming a directory to be the pathname with :DIRECTORY
2587 component indicating that directory.
2588 * minor incompatible change: a NAMED clause in the extended form of
2589 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2590 reason for the previous behaviour is unclear.
2591 * more systematization and improvement of CLOS and MOP conformance
2592 in PCL (thanks to Gerd Moellman and Pierre Mai):
2593 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2595 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2596 primary methods with no specializers;
2597 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2599 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2600 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2601 FORWARD-REFERENCED-CLASSes; error reporting on
2602 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2604 ** SXHASH on CLOS instances now uses a slot internal to the
2605 instance to return different numbers on distinct instances,
2606 while preserving the same return value through invocations of
2608 ** DEFMETHOD signals errors when methods with longer incongruent
2609 lambda lists are added to generic functions;
2610 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2611 CLASS, as specified in AMOP;
2612 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2613 on the order returned by the primary method for classes of
2614 class STANDARD-CLASS;
2615 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2616 * fixed some bugs shown by Paul Dietz' test suite:
2617 ** DOLIST puts its body in TAGBODY;
2618 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2620 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2621 value producing form;
2622 ** if more variables are given to PROGV than values, extra
2623 variables are bound and made to have no value;
2624 ** NSUBSTITUTE on list arguments gets the right answer with
2626 ** ELT signals an error of type TYPE-ERROR when the index argument
2627 is not a valid sequence index;
2628 ** LOOP signals (at macroexpansion time) an error of type
2629 PROGRAM-ERROR when duplicate variable names are found;
2630 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2631 ** FUNCALL of special-operators now cause an error of type
2633 ** PSETQ now works as required in the presence of side-effecting
2634 symbol-macro places;
2635 ** NCONC accepts any object as its last argument;
2636 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2638 ** loop-for-as-package does not require a package to be explicitely
2640 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2642 * fixed bug 166: compiler preserves "there is a way to go"
2643 invariant when deleting code.
2644 * fixed bug 172: macro lambda lists with required arguments after
2645 &REST arguments now cause an error to be signalled. (thanks to
2647 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2648 bug. (thanks to Matthew Danish)
2649 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2651 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2652 arguments are lexically rebound. (thanks to Gerd Moellmann and
2654 * fixed bug 194: error messages are now more informative when there
2655 is no primary method applicable in a call to a generic
2656 function. (thanks to Gerd Moellmann)
2657 * fixed bug in command line argument checking (thanks to Julian
2659 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2660 (thanks to Matthew Danish)
2661 * incremented fasl file version number, because of the
2662 SXHASH-related changes in the layout of CLOS data structures
2664 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2665 * minor incompatible change: The runtime (the Unix executable named
2666 "sbcl") is now much pickier about the .core files it will load.
2667 Essentially it now requires .core files to descend from the same
2668 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2669 as the runtime does. (The intent is to prevent the crashes which
2670 can occur, and which can even be reported as mysterious failures,
2671 when people patch the sources or change the build parameters
2672 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2673 sbcl and .core files.)
2674 * fixed bug: VALUES-LIST is no longer optimized away.
2675 * fixed bug 142: The FFI conversion of C string values to Lisp
2676 string values no longer conses excessively. (thanks to Nathan
2677 Froyd porting Raymond Toy's fix to CMU CL)
2678 * began to systematize and improve MOP conformance in PCL (thanks to
2679 Nathan Froyd, Gerd Moellman and Pierre Mai):
2680 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2682 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2683 ** FINALIZE-INHERITANCE is now called on class finalization;
2684 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2685 argument precedence order.
2686 * fixed bug 202: The compiler no longer fails on functions whose
2687 derived types contradict their declared type.
2688 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2689 so it can be non-toplevel.
2690 * The fasl file version number has changed (because of the new
2691 implementation of DEFMACRO).
2692 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2693 safe code, that any length requirement by their type-specifier
2694 argument is valid. The exceptions to this are described in bug
2696 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2697 functions now check the types of their inputs as required by ANSI.
2698 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2699 introduced symbol is DECLAREd to be SPECIAL.
2700 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2701 * fixed bug: :COUNT argument to sequence functions may be negative
2702 * fixed bug: body of DO-SYMBOLS may contain declarations
2703 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2704 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2705 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2706 symbol macro only once
2707 * fixed printing of call frame when argument list is unavailable
2708 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2709 * compiler no longer signals WARNING on unknown keyword
2712 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2713 * A beta-quality port to the mips architecture running Linux,
2714 based on the old CMUCL backend, has been made. It has been tested
2715 on a big-endian kernel, and works sufficiently well to be able to
2716 rebuild itself; it has not been tested in little-endian mode.
2717 * fixed an inconsistency between gencgc.c and purify.c which made
2718 dumping/loading .core files unreliable
2719 * fixed bug 120a: The compiler now deals correctly with IFs where
2720 the consequent is the same as the alternative, instead of
2721 misderiving the return type. (thanks to Alexey Dejneka)
2722 * fixed bug 113: Logical pathnames are now dumpable (the logical
2723 host is resolved at load-time, throwing an error if it is not
2725 * fixed bug 174: FORMAT's error message is slightly clearer when a
2726 non-printing character is used in a format directive.
2727 * fixed several bugs in compiler checking of type declarations, i.e.
2728 violations of the Python "declarations are assertions" principle
2729 (thanks to Alexey Dejneka)
2730 * fixed several bugs in PCL's error checking (thanks to Gerd
2732 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2734 * fixed bug in compilation of functions as first class values
2735 (thanks to Antonio Martinez-Shotton)
2736 * The compiler's handling TYPE-ERRORs which it can prove will
2737 inevitably happen at runtime has been cleaned up and corrected
2738 in several ways. (thanks to Alexey Dejneka)
2739 * improved argument type checking for various basic arithmetic
2740 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2741 lavished on them in the past that they can be compiled in many
2742 ways in different special cases
2743 * fixed bug 181: compiler checks validity of user supplied type
2745 * cleaned up code flushing in optimization: Function calls which
2746 should signal errors for safety purposes (e.g. which ANSI says
2747 should signal errors when their arguments are of incorrect type)
2748 are no longer optimized away.
2749 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2750 * incremented fasl file version number, because changes in the
2751 implementation of sequence functions like COERCE caused
2752 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2753 in old inline expansions) to become undefined. (Actually these
2754 changes were later undone, so we might very well be binary
2755 compatible with 0.7.7 after all, but leaving the version number
2756 incremented seemed like the simplest and most conservative
2759 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2760 * An alpha-quality port to the parisc architecture running Linux,
2761 based on the old CMUCL backend, has been made. This, even more so
2762 than the other backends, should be considered still a work in
2763 progress; known problems include that the Linux kernel in 64-bit
2764 mode does not propagate the correct sigcontext structure to
2765 userspace, and consequently SBCL on a parisc64 kernel will not
2767 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2768 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2769 suppressed.) Also now that the compiler is looking at declarations
2770 in the environment, it checks optimization declarations as well,
2771 and suppresses inlining when (> DEBUG SPEED).
2772 * More fixes have been made to treatment of floating point exception
2773 treatment and other Unix signals. In particular, floating point
2774 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2775 * The detection and handling of control stack exhaustion (infinite
2776 or very deeply nested recursion) has changed. Stack exhaustion
2777 detection is now done by write-protecting pages at the OS level
2778 and applies at all optimization settings; when found, a
2779 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2780 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2781 be caught using IGNORE-ERRORS.
2782 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2783 correctly when transforming recursive LABELS functions to LETs.
2784 (thanks to Alexey Dejneka)
2785 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2786 that are names of constants or global variables.
2787 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2788 alien routines with docstrings.
2789 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2790 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2792 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2793 (thanks to Eric Marsden)
2794 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2795 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2796 * Bug fix: The SPARC backend can now compile functions involving
2797 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2798 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2799 to a functions expecting a PATHNAME-DESIGNATOR.
2800 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2801 lambda lists. (thanks to Alexey Dejneka)
2802 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2803 functions so that EXPORTing the name of the class doesn't cause
2804 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2805 to Antonio Martinez for reporting this)
2806 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2807 constant-folded without failing an assertion. (thanks to Einar
2808 Floystad Dorum for reporting this)
2809 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2810 types are now dealt with more correctly by the compiler.
2811 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2812 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2813 behaviour analogous to RENAME-FILE. This puts its behaviour more
2814 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2816 * The fasl file version number has changed again. (because of the
2817 bug fix involving the names of PCL MAKE-INSTANCE functions)
2819 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2820 * bug fix: Floating point exceptions are treated much more
2821 consistently on the x86/Linux and PPC/Linux platforms.
2822 * Array initialization with :INITIAL-ELEMENT is now much faster for
2823 cases when the compiler cannot open code the array creation, but
2824 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2825 array accesses have also seen a speed increase.
2826 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2827 specified. (This was at the root of some bad interactions between
2828 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2830 * bug fix: Internal error arguments for undefined functions are now
2831 computed correctly on the PPC/Linux platform.
2832 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2833 Raymond Toy's patch for CMU CL)
2834 * Support for the Solaris 9 operating environment has been included
2835 (thanks to Daniel Merritt)
2836 * A very ugly but hopefully complete draft of the missing FFI chapter
2837 of the manual has been created by reformatting the corresponding
2838 CMU CL manual chapter into (currently very ugly and incoherent)
2839 DocBook and bringing it up to date for SBCL behavior. Thus, the
2840 manual is now essentially complete, at least by my extreme
2841 once-and-only-once standards, whereby it's acceptable to refer to
2842 the doc strings of SB-EXT functions as the primary documentation.
2843 * The fasl file version number has changed again, due to cleanup of
2844 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2845 is no longer a static symbol.)
2847 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2848 * SBCL now builds with OpenMCL (version 0.12) as the
2849 cross-compilation host; also, more progress has been made toward
2850 bootstrapping under CLISP.
2851 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2853 * bug 158 fixed: The compiler can now deal with integer loop
2854 increments different from 1; fixing this turned out also to fix
2856 * bug 169 fixed: no more bogus warnings about using lexical bindings
2857 despite the presence of perfectly good SPECIAL declarations (thanks
2858 to David Lichteblau)
2859 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2860 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2861 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2863 * bug 180 fixed: Method combination specifications no longer ignore
2864 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2865 * bug fix: Structure type predicate functions now check their argument
2866 count as they should.
2867 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2868 correctly. (thanks to Pierre Mai)
2869 * minor incompatible change: The --noprogrammer option is deprecated
2870 in favor of the new --disable-debugger option, which is very similar.
2871 (The major difference is that it takes effect at a slightly different
2872 time at startup, causing handling of errors in --sysinit and
2873 --userinit files will be affected differently.) The
2874 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2875 been added to allow this functionality to be controlled from ordinary
2876 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2877 maintainers, who might want to run non-interactive scripts to
2878 build SBCL cores which will later be used interactively.)
2879 * minor incompatible change: The LOAD function no longer, when given
2880 a wild pathname to load, loads all files matching that pathname.
2881 Instead, an error of type FILE-ERROR is signalled.
2883 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2884 * bug 147 fixed: The compiler preserves its block link/count
2885 invariants more correctly now so that it doesn't crash. (thanks
2887 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2889 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2890 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2891 no-value-for-this-character as NIL (as specified by ANSI).
2892 * HOST-NAMESTRING on physical pathnames now returns a string that is
2893 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2894 (thanks to Christophe Rhodes)
2895 * The Alpha port handles icache flushing more correctly. (thanks to
2897 * More progress has been made toward bootstrapping under CLISP. (thanks
2898 to Christophe Rhodes)
2899 * The fasl file format has changed again, because dynamic loading
2900 on OpenBSD (which has non-ELF object files) motivated some cleanups
2901 in the way that foreign symbols are transformed and passed around.
2902 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2903 is no longer treated as whitespace by the reader, but instead as
2904 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2905 |A
\7fB|, instead of A as it used to.
2907 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2908 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2909 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2910 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2911 this as of 0.7.1.45, but was left out of the previous news section
2912 (thanks to Dan Barlow)
2913 * SBCL now runs on the Solaris operating system on SPARC architectures
2914 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2915 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2916 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2917 * SPARC backend cleanups, allowing builds of cores optimized for V8
2918 and V9 SPARCS, and also emission of code targeted to a particular
2919 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2921 * SBCL is closer to bootstrapping under CLISP, thanks to various
2922 fixes by Christophe Rhodes.
2923 * The fasl file format has changed again, to allow the compiler's
2924 INFO database to support symbol macros.
2925 * The user manual (in doc/) is formatted into HTML more nicely.
2926 (thanks to coreythomas)
2927 * The system is smarter about SUBTYPEP relationships, especially
2928 those involving NOT types (including types such as ATOM which are
2929 represented internally using NOT types). Thus SUBTYPEP is less
2930 likely to return (VALUES NIL NIL) in general, and in particular
2931 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2933 * The fasl file format has changed again, because the internal
2934 representation of types now includes a new slot to support the new
2935 SUBTYPEP-of-NOT-types logic.
2936 * (not a change in the main branch of SBCL, but a related prototype
2937 which can hopefully be merged into the main branch of SBCL in the
2938 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2939 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2940 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2941 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2942 dispatch macro character combinations. (thanks to Alexey Dejneka)
2943 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2944 that superficially look like logical namestrings correctly.
2945 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2947 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2948 default supplied-p) for &optional and &key arguments. (thanks to
2951 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2952 * incompatible change: The compiler is now less aggressive about
2953 tail call optimization, doing it only when (> SPACE DEBUG) or
2954 (> SPEED DEBUG). (This is an incompatible change because there are
2955 programs which relied on the old CMU-CL-style behavior to optimize
2956 away their unbounded recursion which will now die of stack overflow.)
2957 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2958 for non-GENCGC systems has been increased to 20M (since that
2959 seems much closer to the likely performance optimum for modern
2960 systems than the old 4M value was)
2961 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2962 and *DEBUG-PRINT-LEVEL*
2963 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2964 of CMU CL's support for SPARC, and various endianness and other
2965 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2966 * new syntactic sugar for the Unix command line: --load foo.bar is now
2967 an alternate notation for --eval '(load "foo.bar")'.
2969 ** The system now detects stack overflow and handles it gracefully,
2970 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2971 optimization settings. (This is a good thing in general, and
2972 its introduction in this version should be particularly timely
2973 for anyone whose code fails because of suppression of tail
2975 ** The system now hunts for the C variable "environ" in a more
2976 devious way, to avoid segfaults when the C library version differs
2977 between compile time and run time. (thanks to Christophe Rhodes)
2978 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2979 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2980 ** The compiler no longer issues bogus style warnings for undefined
2981 classes in the same source file as the DEFCLASSes which defined
2982 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2984 ** fixes in CONDITION class precedence list for undefined function
2985 errors (thanks to Alexei Dejneka)
2986 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2987 correctly. (thanks to Dan Barlow)
2988 ** portability fixes aiming at bootstrapping under CLISP (thanks
2989 to Dave McDonald and Christophe Rhodes)
2990 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2991 ** fixes in type translation and and type inference (thanks to
2993 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2994 ** various fixes in the new ports (thanks to Dan Barlow)
2995 * several changes related to debugging:
2996 ** suppression of tail recursion, as noted above
2997 ** stack overflow detection, as noted above
2998 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2999 is now the default. (For some time encapsulation has been more
3000 reliable than the breakpoint-based :ENCAPSULATE NIL
3001 implementation, at least on X86 systems; and I just noticed that
3002 encapsulation also seems closer to the spirit of the ANSI
3005 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3007 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3008 up properly again. (There was a packaging bug in 0.7.0 which
3009 left their definitions in SB-SYS::LOAD-FOREIGN and
3010 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3011 vital for most things which interface to C-level interfaces,
3012 like extensions working with sockets or databases or
3013 Perl-compatible regexes or whatever, and the need to fix
3014 this bug is the main reason that 0.7.1 was released so
3016 ** DEFGENERIC is now choosier about the methods it redefines, so that
3017 reLOADing a previously-LOADed file containing DEFGENERICs does
3018 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3019 cycle now works as it should. (thanks to Alexey Dejneka)
3020 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3021 fixed by Christophe Rhodes. (He actually submitted this patch
3022 months ago, and I delayed until after 0.7.0.)
3023 ** Bug 111 (internal compiler confusion about runtime checks on
3024 FUNCTION types) was fixed by Alexey Dejneka.
3025 * Some internal cleanups (getting rid of variables which aren't
3026 needed now that the byte interpreter is gone) caused the fasl
3027 file format number to change again.
3029 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3030 * major incompatible change: The default fasl file extension, i.e. the
3031 default extension for files produced by COMPILE-FILE, has changed
3032 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3034 ** There are many changes in the implementation of the compiler.
3035 SBCL is now essentially a compiler-only implementation of ANSI
3036 Common Lisp. EVAL still "interprets" a few special cases, but
3037 almost all the interesting cases are handled by creating
3038 a LAMBDA expression, calling COMPILE on it, then calling
3039 FUNCALL on the result.
3040 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3041 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3042 Since the code is newer, there might still be some new bugs
3043 (though not as many as before Martin Atzmueller's fixes:-). But
3044 the new code is substantially simpler and clearer, and hopefully
3045 any remaining bugs will be simpler, less fundamental, and more
3046 fixable then the bugs in the old code.
3047 ** The revised compiler is still a little unsteady on its feet.
3049 *** The debugging information it produces (particularly the names
3050 of FUNCTION objects) is sometimes much less useful than what
3051 the old compiler produced.
3052 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3053 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3054 within a MACROLET) has been temporarily weakened.
3055 ** There are new compiler optimizations for various functions:
3056 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3057 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3058 *** the math functions TRUNCATE, FLOOR, and CEILING
3059 *** the function-of-all-trades COERCE
3060 Mostly these should be transparent, but there's one
3061 potentially-annoying problem (bug 117): when the compiler
3062 inline-expands a function and does type analysis on the result,
3063 it can create control paths which have type mismatches, and
3064 when it can't prove that those control paths aren't taken,
3065 it will issue WARNINGs about the type mismatches. This is
3066 a particular problem in practice for the new sequence functions.
3067 It's not clear how this should be fixed, and for now, a
3068 workaround is given in the entry for 117 in the BUGS file.
3069 ** (Because of the interaction between the two previous items --
3070 occasional inlining problems and new inline expansions -- some
3071 of the new sequence function optimizations won't really kick in
3072 completely until debugging information, and then inlining, are
3073 straightened out in some future version.)
3074 * minor incompatible changes:
3075 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3076 DIRECTORY no longer implicitly promotes NIL slots of its
3077 pathname argument to :WILD. In particular, when you ask for the
3078 contents of a directory (which you used to be able to do without
3079 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3080 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3081 ** changes in behavior that ANSI explicitly defines to be
3082 implementation dependent:
3083 *** The new compiler-only implementation still conforms with ANSI,
3084 but acts a little different than before. Besides the obvious
3085 changes in performance tradeoffs (that the cost per form passed
3086 to EVAL has gone up, and the cost per form executed by EVAL
3087 has gone down), the behavior of the system changes a little
3088 because there are no longer any interpreted function objects.
3089 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3090 e.g. doing COMPILE on the output of interactive DEFUN is
3092 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3094 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3095 has changed from (:CL) to NIL.
3096 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3097 e.g. #\Newline and #\Tab, don't have names specified in the
3098 ANSI Common Lisp standard, is now based on their ASCII symbolic
3099 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3100 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3101 are no longer used for output.
3102 ** changes in internal implementation constants:
3103 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3104 4 million. (If your application spends a lot of time GCing and
3105 you have a lot of RAM, you might want to experiment with
3106 increasing it even more.)
3107 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3108 However, almost all old code should still continue to work without
3109 immediate update, as SB-C-CALL is now a (deprecated) nickname
3111 ** Old operator names in the style DEF-FOO are now deprecated in
3112 favor of new corresponding names DEFINE-FOO, for consistency with
3113 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3114 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3115 internal symbols, but a few supported extensions like
3116 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3117 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3118 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3119 etc. as you get deeper into recursive calls to the debugger
3120 command loop, instead of the old "5]", "5]]", "5]]]"
3121 sequence. (I was motivated to do this when squabbles between
3122 ILISP and SBCL left me very deeply nested in the debugger. In the
3123 short term, this change will probably provoke more ILISP/SBCL
3124 squabbles, but hopefully it will be an improvement in the long run.)
3125 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3126 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3127 function, since declaiming return types involving aliens is
3128 (1) annoyingly messy to do by hand and (2) vital to efficient
3129 compilation of code which calls such functions.
3130 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3131 longer reexported by the SB-EXT package. They're solely useful
3132 for alien code, so it seems more logical that you should get
3133 them from the SB-ALIEN package, not in SB-EXT.
3134 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3135 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3136 features. Instead, the code that they used to control is always
3137 built into the system.
3138 * many other bug fixes
3139 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3140 advantage of the new EVAL-WHEN stuff and to clean them up in
3141 general, and they are now more ANSI-compliant in a number of
3142 ways. Martin Atzmueller is responsible for a lot of this.
3143 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3145 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3146 *** correct ERROR type for various file operations
3147 *** some fixes for Lisp streams
3148 *** DEFMETHOD syntax checking
3149 *** changing old weird representation of debug information as
3150 strings (which, among their other deficiencies, don't transform
3151 correctly when you rename packages, and don't change their
3152 print representation when you change things like *PACKAGE*
3153 and *PRINT-LENGTH*) to symbols and lists of symbols
3154 He also made several improvements and fixed several bugs in DESCRIBE.
3155 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3157 *** misbehavior of WRITE-STRING/WRITE-LINE
3158 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3159 and several other LOOP problems as well
3160 *** DIRECTORY when similar filenames are present
3161 *** DEFGENERIC with :METHOD options
3162 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3163 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3164 *** argument ordering in FIND with :TEST option
3165 *** mishandled package designator argument in APROPOS-LIST
3166 *** various problems in the backquote readmacro
3167 *** a bug in APROPOS
3168 *** probably some others that I'm not describing very well here,
3169 since the CVS log documents them by reference to sbcl-devel
3170 messages, and the SourceForge archives aren't working well.:-(
3171 ** Dan Barlow improved the Alpha port (and is making progress on the
3172 PPC port, for those of you who think different).
3173 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3174 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3175 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3176 continues to work on the SPARC port (for those of you in a position
3177 to look down upon our little PC-compatible boxes from a great height).
3178 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3179 and uses that copy, rather than the current dynamic value, when
3180 it's trying to decide whether to truncate output. Thus e.g.
3181 (let ((*print-lines* 50))
3182 (pprint-logical-block (stream nil)
3184 (let ((*print-lines* 8))
3185 (print (aref possiblybigthings i) stream)))))
3186 should now truncate the logical block only at 50 lines, instead of
3187 often truncating it at 8 lines, as it did before.
3188 * The doc/cmucl/ directory, containing old CMU CL documentation
3189 from the time of the fork, is no longer part of the base system.
3190 SourceForge has shut down its anonymous FTP service, and with it
3191 my original plan for distributing the old CMU CL documentation
3192 there. For now, if you need these files you can download an old
3193 SBCL source release and extract them from it.
3194 * The fasl file version number changed again, for dozens of reasons,
3195 some of which are apparent above.
3197 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3198 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3199 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3200 patch, so that DISASSEMBLE output is much nicer.
3201 * The code in the SB-PROFILE package now seems reasonably stable.
3202 I still haven't decided what the final interface should look like
3203 (I'd like PROFILE to interact cleanly with TRACE, since both
3204 facilities use function encapsulation) but if you have a need
3205 for profiling now, you can probably use it successfully with
3206 the current CMU-CL-style interface.
3207 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3208 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3209 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3210 link now returns the dangling link itself, and for similar
3211 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3212 link itself. (In these cases the old code signalled an error and
3213 looped endlessly, respectively.) Thus, DIRECTORY now works even
3214 in the presence of dangling and cyclic symbolic links.
3215 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3216 is now a supported extension again, since the consensus on
3217 sbcl-devel was that it can be useful for ordinary development
3218 work, not just for debugging SBCL itself.
3219 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3220 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3221 that currently-defined functions might be redefined later with
3222 different return types.
3223 * Hash tables can be printed readably, as inspired by CMU CL code
3224 of Eric Marsden and SBCL code of Martin Atzmueller.
3225 * better error handling in CLOS method combination, thanks to
3226 Martin Atzmueller porting Pierre Mai's CMU CL patches
3227 * more overflow fixes for >16Mbyte I/O buffers
3228 * A bug in READ has been fixed, so that now a single Ctrl-D
3229 character suffices to cause end-of-file on character streams.
3230 In particular, now you only need one Ctrl-D at the command
3231 line (not two) to exit SBCL.
3232 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3234 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3235 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3236 does the right thing.
3237 * The implementation of some type tests, especially for CONDITION
3238 types, is now tidier and maybe faster, due to CMU CL code
3239 originally by Douglas Crosher, ported by Martin Atzmueller.
3240 * Some math functions have been fixed, and there are new
3241 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3242 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3243 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3244 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3245 likely to be built into sbcl-0.7.0.)
3246 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3247 CL, ported to SBCL by Martin Atzmueller)
3248 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3249 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3250 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3251 * There's a new slam.sh hack to shorten the edit/compile/debug
3252 cycle for low-level changes to SBCL itself, and a new
3253 :SB-AFTER-XC-CORE target feature to control the generation of
3254 the after-xc.core file needed by slam.sh.
3255 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3256 COMPILE-FILE is no longer supported, so that now every function
3257 gets an entry point, so that block compilation looks a little
3258 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3259 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3260 returns the number of bytes consed since the system started,
3261 rather than the number consed since the first time the function
3262 was called. (The new definition parallels ANSI functions like
3263 CL:GET-INTERNAL-RUN-TIME.)
3264 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3265 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3266 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3267 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3268 * Fasl file version numbers are now independent of the target CPU,
3269 since historically most system changes which required version
3270 number changes have affected all CPUs equally. Similarly,
3271 the byte fasl file version is now equal to the ordinary
3274 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3275 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3276 is no longer recognized. I apologize for this, because it was
3277 listed in SB-EXT as a supported extension, but I found that
3278 its existing behavior was poorly specified, as well as incorrectly
3279 specified, and it looked like too much of a mess to straighten it
3280 out. I have enough on my hands trying to get ANSI stuff to work..
3281 * many patches ported from CMU CL by Martin Atzmueller, with
3282 half a dozen bug fixes in pretty-printing and the debugger, and
3283 half a dozen others elsewhere
3284 * fixed bug 13: Floating point infinities are now supported again.
3285 They might still be a little bit flaky, but thanks to bug reports
3286 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3287 as flaky as they were.
3288 * The --noprogrammer command line option is now supported. (Its
3289 behavior is slightly different in detail from what the old man
3290 page claimed it would do, but it's still appropriate under the
3291 same circumstances that the man page talks about.)
3292 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3293 are now supported, and enabled by default. Thus, the compiler can
3294 handle many floating point and complex operations much less
3295 inefficiently. (Thus e.g. you can implement a complex FFT
3297 * The compiler now detects type mismatches between DECLAIM FTYPE
3298 and DEFUN better, and implements CHECK-TYPE more correctly, and
3299 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3300 thanks to patches from Martin Atzmueller.
3301 * various fixes to make the cross-compiler more portable to
3302 ANSI-conforming-but-different cross-compilation hosts (notably
3303 Lispworks for Windows, following bug reports from Arthur Lemmens)
3304 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3305 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3306 fixed thanks to Pierre Mai's CMU CL patches.
3307 * fixes to keep the system from overflowing internal counters when
3308 it tries to use i/o buffers larger than 16M bytes
3309 * fixed bug 45a: Various internal functions required to support
3310 complex special functions have been merged from CMU CL sources.
3311 (When I was first setting up SBCL, I misunderstood a compile-time
3312 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3313 * improved support for type intersection and union, fixing bug 12
3314 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3315 more obscure bugs as well
3316 * some steps toward byte-compiling non-performance-critical
3317 parts of the system, courtesy of patches from Martin Atzmueller
3318 * Christophe Rhodes has made some debian packages of sbcl at
3319 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3320 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3321 stable, but are nonetheless usable. When he's ready, I'd be happy
3322 to add them to the SourceForge "File Releases" section. (And if
3323 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3324 * new fasl file format version number (because of changes in
3325 internal representation of (OR ..) types to accommodate the new
3326 support for (AND ..) types, among other things)
3328 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3329 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3331 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3333 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3334 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3335 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3336 * DESCRIBE now works on structure objects again.
3337 * Most function call argument type mismatches are now handled as
3338 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3339 know whether the function will be redefined before the call is
3340 executed. (The compiler could flag local calls with full WARNINGs,
3341 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3342 it doesn't keep track of enough information to know whether calls
3343 are local in this sense.)
3344 * Compiler output is now more verbose, with messages truncated
3345 later than before. (There should be some supported way for users
3346 to override the default verbosity, but I haven't decided how to
3347 provide it yet, so this behavior is still controlled by the internal
3348 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3349 src/compiler/ir1util.lisp.)
3350 * Fasl file format version numbers have increased again, because
3351 support for the Gray streams extension changes the layout of the
3352 system's STREAM objects.
3353 * The Gray subclassable streams extension now works, thanks to a
3354 patch from Martin Atzmueller.
3355 * The full LOAD-FOREIGN extension (not just the primitive
3356 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3357 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3358 but like most other programs, it defaults to copying the Unix
3359 environment from the original process instead of starting the
3360 new process in an empty environment.
3361 * Extensions which manipulate the Unix environment now support
3362 an :ENVIRONMENT keyword option which doesn't smash case or
3363 do other bad things. The CMU-CL-style :ENV option is retained
3364 for porting convenience.
3365 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3366 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3368 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3370 * A patch from Martin Atzmueller seems to have solved the SIGINT
3371 problem, and as far as we know, signal-handling now works cleanly.
3372 (If you find any new bugs, please report them!)
3373 * The system no longer defaults Lisp source file names to types
3374 ".l", ".cl", or ".lsp", but only to ".lisp".
3375 * The compiler no longer uses special default file extensions for
3376 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3377 seems to expect a single default extension for all compiled code,
3378 and there's no compelling reason to try to stretch the standard
3379 to allow two different extensions.) Instead, byte-compiled files
3380 default to the same extension as native-compiled files.
3381 * Fasl file format version numbers have increased again, because
3382 a rearrangement of internal implementation packages made some
3383 dumped symbols in old fasl files unreadable in new cores.
3384 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3385 many fewer weird special cases.
3386 * Bug #17 (differing COMPILE-FILE behavior between logical and
3387 physical pathnames) has been fixed, and some related misbehavior too,
3388 thanks to a patch from Martin Atzmueller.
3389 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3390 by Tim Moore, ported to SBCL by Martin Atzmueller.
3391 * Martin Atzmueller fixed several filesystem-related problems,
3392 including bug #36, in part by porting CMU CL patches, which were
3393 written in part by Paul Werkowski.
3394 * More compiler warnings in src/runtime/ are gone, thanks to
3395 more patches from Martin Atzmueller.
3396 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3399 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3401 * DESCRIBE now works on CONDITION objects.
3402 * The debugger now handles errors which arise when trying to print
3403 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3405 * The build system now uses an additional file, customize-target-features.lisp,
3406 to allow local modifications to the target *FEATURES* list. (The point of
3407 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3408 debugging features enabled, without having to worry about propagating it
3409 into everyone's system when I do a "cvs update".) When no
3410 customize-target-features.lisp file exists, the target *FEATURES* list
3411 should be constructed the same way as before.
3412 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3413 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3414 warn about behavior which is undefined under the ANSI spec. Note: This
3415 is specified by ANSI, but it's not very popular with programmers.
3416 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3417 macro in the SBCL sources for an example of a workaround which you
3418 might use to make portable ANSI-standard code which does what you want.
3419 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3420 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3421 when it's used as a non-toplevel form. (This is required in order
3422 to implement the DEFCONSTANT-EQX macro.)
3423 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3424 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3425 funny things, due to bugs in the implementation of EVAL-WHEN
3426 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3427 non-toplevel DEFCONSTANTs are uncommon.)
3428 * The core file version number and fasl file version number have been
3429 incremented, because the old noncompliant DEFCONSTANT behavior involved
3430 calling functions which no longer exist, and because I also took the
3431 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3432 * fixed bug 1 (error handling before read-eval-print loop starts), and
3433 redid debugger restarts and related debugger commands somewhat while
3435 ** The QUIT debugger command is gone, since it did something
3436 rather different than the SB-EXT:QUIT command, and since it never
3437 worked properly outside the main toplevel read/eval/print loop.
3438 Invoking the new TOPLEVEL restart provides the same functionality.
3439 ** The GO debugger command is also gone, since you can just invoke
3440 the CONTINUE restart directly instead.
3441 ** The TOP debugger command is also gone, since it's redundant with the
3442 FRAME 0 command, and since it interfered with abbreviations for the
3444 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3446 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3447 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3448 some porting work by Martin Atzmueller)
3449 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3450 with a boatload of patches which he ported to SBCL. Now that those
3452 ** The system tries to make sure that its low-priority messages
3453 are prefixed by semicolons, to help people who like to use
3454 syntax highlighting in their ILISP buffer. (This patch
3455 was originally due to Raymond Toy.)
3456 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3457 patches originally written by Raymond Toy.
3458 ** The compiler understands coercion between single-value and
3459 multiple-VALUES type expressions better, getting rid of some very
3460 weird behavior, thanks to patches originally by Robert MacLachlan
3461 and Douglas Crosher.
3462 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3463 lambda lists, thanks to a patch originally by Pierre Mai.
3464 ** The system no longer bogusly warns about "abbreviated type
3466 ** The compiler gets less confused by inlining and RETURN-FROM,
3467 thanks to some patches originally by Tim Moore.
3468 ** The system no longer hangs when dumping circular lists to fasl
3469 files, thanks to a patch originally from Douglas Crosher.
3470 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3471 undefined function error.
3472 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3473 in structure slot initforms) for now, documented workaround instead:-|
3474 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3475 * fixed bug 5: added stubs for various Gray stream functions called
3476 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3477 installed, at least appropriate type errors are generated
3478 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3479 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3480 consistently in DEFMETHOD forms.
3481 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3482 it doesn't seem to affect SBCL after all
3483 * The C runtime system now builds with better optimization and many
3484 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3486 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3488 * The system is now under CVS at SourceForge (instead of the
3489 CVS repository on my home machine).
3490 * The new signal handling code has been tweaked to treat register
3491 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3492 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3493 downstream, e.g. in debug-int.lisp, has implicit dependencies
3494 on the unsignedness of integer representation of machine words,
3495 and that caused the system to bomb out with infinite regress
3496 when trying to recover from type errors involving signed values,
3497 e.g. (BUTLAST '(1 2 3) -1).)
3498 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3499 (This was one of the bugs Peter Van Eynde reported back in July.)
3500 * The system now uses code inspired by Colin Walters' O(N)
3501 implementation of MAP (from the cmucl-imp@cons.org mailing
3502 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3503 inline the MAP operation, and there is more than one
3504 sequence argument to the MAP call (so that it can't just
3505 do ETYPECASE once and for all based on the type of the
3506 single sequence argument). (The old non-inline implementation
3507 of the general M-argument sequence-of-length-N case required
3508 O(M*N*N) time when any of the sequence arguments were LISTs.)
3509 * The QUIT :UNIX-CODE keyword argument has been renamed to
3510 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3511 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3513 * Raymond Wiker's patches to port dynamic loading from Linux to
3514 FreeBSD have been added.
3515 * The BUGS file is now more nearly up to date, thanks in large part
3516 to Martin Atzmueller's review of it.
3517 * The debugger now flushes standard output streams before it begins
3518 its output ("debugger invoked" and so forth).
3519 * The core version number and fasl file version number have both
3520 been incremented, because of incompatible changes in the layout
3522 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3523 used to not be called for a saved Lisp image.)
3524 * Martin Atzmueller's version of a patch to fix a compiler crash,
3525 as posted on sbcl-devel 13 September 2000, has been installed.
3526 * Instead of installing Martin Atzmueller's patch for the
3527 compiler transform for SUBSEQ, I deleted the compiler transform,
3528 and transforms for some similar consing operations.
3529 * A bug in signal handling which kept TRACE from working on OpenBSD
3531 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3532 away by constant folding
3533 * The system now defines its address space constants in one place
3534 (in the Lisp sources), and propagates them automatically elsewhere
3535 (through GENESIS and the sbcl.h file). Therefore, patching the
3536 address map is less unnecessarily tedious and error-prone. The
3537 Lisp names of address space constants have also been systematized.
3538 * CVS tags like dollar-Header-dollar have been removed from
3539 the sources, because they have never saved me trouble and
3540 they've been source of trouble working with patches and other
3541 diff-related operations.
3542 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3543 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3545 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3547 * The system has been ported to OpenBSD.
3548 * The system now compiles with a simple "sh make.sh" on the systems
3549 that it's supported on. I.e., now you no longer need to tweak
3550 text in the target-features.lisp-expr and symlinks in src/runtime/
3551 by hand, the make.sh takes care of it for you.
3552 * The system is no longer so grossly inefficient when compiling code
3553 involving vectors implemented as general (not simple) vectors (VECTOR T),
3554 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3555 which dares to use the various sequence functions on non-simple
3556 vectors, takes less of a performance hit.
3557 * There is now a primitive type predicate VECTOR-T-P
3558 to test for the (VECTOR T) type, so that e.g.
3559 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3560 can now be compiled with some semblance of efficiency. (The old code
3561 turned the type declaration into a full call to %TYPEP at runtime!)
3562 * AREF on (VECTOR T) is still not fast, since it's still compiled
3563 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3564 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3565 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3566 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3567 without making full calls to SUBTYPEP at runtime.
3568 (Some analogous efficiency issues for non-simple vectors specialized to
3569 element types other than T, or for non-simple multidimensional arrays,
3570 have not been addressed. They could almost certainly be handled the
3571 same way if anyone is motivated to do so.)
3572 * The changes in array handling break binary compatibility, so
3573 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3574 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3575 instead of (VALUES T T).
3576 * By following the instructions that Dan Barlow posted to sbcl-devel
3577 on 2 July 2000, I was able to enable primitive dynamic object
3578 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3579 functionality is not implemented (since it calls ld to resolve
3580 library references automatically, requiring RUN-PROGRAM for its
3581 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3582 try to resolve library references) is now supported.
3583 * The system now flushes the standard output streams when it terminates,
3584 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3585 them at several other probably-convenient times, e.g. in each pass of
3586 the toplevel read-eval-print loop, and after evaluating a form given
3587 as an "--eval" command-line option. (These changes were motivated by a
3588 discussion of stream flushing issues on cmucl-imp in August 2000.)
3589 * The source transform for TYPEP of array types no longer assumes
3590 that an array whose element type is a not-yet-defined type
3591 is implemented as an array of T, but instead punts, so that the
3592 type will be interpreted at runtime.
3593 * There is now some support for cross-compiling in make.sh: each of
3594 the phases of make.sh has its own script. (This should be transparent
3595 to people doing ordinary, non-cross-compile builds.)
3596 * Since my laptop doesn't have hundreds of megabytes of memory like
3597 my desktop machine, I became more motivated to do some items on
3598 my to-do list in order to reduce the size of the system a little:
3599 ** Arrange for various needed-only-at-cold-init things to be
3600 uninterned after cold init. To support this, those things have
3601 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3602 symbols with such names are now uninterned after cold init).
3603 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3604 fasl files for cold load.
3605 ** Remove the old compiler structure pooling code (which used to
3606 be conditional on the target feature :SB-ALLOC) completely.
3607 ** Redo the representation of some data in cold init to be more compact.
3608 (I also looked into supporting byte compiled code at bootstrap time,
3609 which would probably reduce the size of the system a lot, but that
3610 looked too complicated, so I punted for now.)
3611 * The maximum signal nesting depth in the src/runtime/ support code has
3612 been reduced from 4096 to 256. (I don't know any reason for the very
3613 large old value. If the new smaller value turns out to break something,
3614 I'll probably just bump it back up.)
3615 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3617 * Many, many bugs reported by Peter Van Eynde have been added to
3618 the BUGS list; some have even been fixed.
3619 * While enabling dynamic object file loading, I tried to make the
3620 code easier to understand, renaming various functions and variables
3621 with less ambiguous names, and changing some function calling
3622 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3623 * While trying to figure out how to do the OpenBSD port, I tried to
3624 clean up some of the code in src/runtime/. In particular, I dropped
3625 support for non-POSIX signal handling, added various comments,
3626 tweaked the code to reduce the number of compilation warnings, and
3627 renamed some files to increase consistency.
3628 * To support the new automatic configuration functionality in make.sh,
3629 the source file target-features.lisp-expr has been replaced with the
3630 source file base-target-features.lisp-expr and the machine-generated
3631 file local-target-features.lisp-expr.
3632 * fixed a stupid quoting error in make.sh so that using CMU CL
3633 "lisp -batch" as cross-compilation host works again
3635 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3637 * DESCRIBE no longer tries to call itself recursively to describe
3638 bound/fbound values, so that it no longer fails on symbols which are
3639 bound to themselves (like keywords, T, and NIL).
3640 * DESCRIBE now works on generic functions.
3641 * The printer now prints less-screwed-up representations of closures
3642 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3643 * A private symbol is used instead of the :EMPTY keyword previously
3644 used to mark empty slots in hash tables. Thus
3645 (DEFVAR *HT* (MAKE-HASH-TABLE))
3646 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3647 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3648 now does what ANSI says that it should. (You can still get
3649 similar noncompliant behavior if bang on the hash table
3650 implementation with all the symbols you get back from
3651 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3652 This breaks binary compatibility, since tests for equality to
3653 :EMPTY are wired into things like the macroexpansion of
3654 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3656 * There's now a minimal placeholder implementation for CL:STEP,
3657 as required by ANSI.
3658 * An obscure bug in the interaction of the normal compiler, the byte
3659 compiler, inlining, and structure predicates has been patched
3660 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3661 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3663 * Missing ordinary arguments in a macro call are now detected even
3664 when the macro lambda list contains &KEY or &REST.
3665 * The debugger no longer complains about encountering the top of the
3666 stack when you type "FRAME 0" to explicitly instruct it to go to
3667 the top of the stack. And it now prints the frame you request even
3668 if it's the current frame (instead of saying "You are here.").
3669 * As specified by ANSI, the system now always prints keywords
3670 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3672 * The default initial SIZE of HASH-TABLEs is now smaller.
3673 * Type information from CLOS class dispatch is now propagated
3674 into DEFMETHOD bodies, so that e.g.
3675 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3677 is now basically equivalent to
3678 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3679 (DECLARE (TYPE SINGLE-FLOAT X))
3681 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3682 operation, without having to do run-time type dispatch.
3683 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3684 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3685 * Since I don't seem to be making big file reorganizations very often
3686 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3687 is overflowing my ability to conveniently back them up), I've finally
3688 checked the system into CVS. (The CVS repository is on my home system,
3689 not at SourceForge -- putting it on SourceForge might come later.)
3690 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3691 high-level GC-NOTIFY-FOO functions send their output. (There's
3692 still very little control of where low-level verbose GC functions
3693 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3694 less than it used to -- the GC-NOTIFY-FOO functions are now under
3695 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3696 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3697 in only one place in the source code, and propagates it automatically
3698 everywhere that it's needed. Thus e.g. when I bump the version from
3699 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3700 * The C source files now include boilerplate legalese and documentation
3701 at the head of each file (just as the Lisp source files already did).
3702 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3703 to his page will be replaced with a link to his new CLiki service.
3705 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3707 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3708 * The build process now looks for GNU make under the default name "gmake",
3709 instead of "make" as it used to. If GNU make is not available as "gmake"
3710 on your system, you can change this default behavior by setting the
3711 GNUMAKE environment variable.
3712 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3713 can build without error under CMU CL.
3715 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3717 * There is now a partial SBCL user manual (with some new text and some
3718 text cribbed from the CMU CL manual).
3719 * The beginnings of a profiler have been added (starting with the
3720 CMU CL profiler and simplifying and cleaning up). Eventually the
3721 main interface should be through the TRACE macro, but for now,
3722 it's still accessed through vaguely CMU-CL-style functions and macros
3723 exported from the package SB-PROFILE.
3724 * Some problems left over from porting CMU CL to the new
3725 cross-compilation bootstrap process have been cleaned up:
3726 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3727 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3728 part when getting rid of the compiler *BACKEND* stuff.)
3729 ** The value of *NULL-TYPE* was screwed up, because it was
3730 being initialized before the type system knew the final
3731 definition of the 'NULL type. This screwed up several key
3732 optimizations in the compiler, causing inefficiency in all sorts
3733 of places. (I found it because I wanted to understand why
3734 GET-INTERNAL-RUN-TIME was consing.)
3735 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3736 PROCLAIM FTYPE information. Unfortunately this broke binary
3737 compatibility again, since now the forms output by DEFGENERIC
3738 to refer to functions which didn't exist in 0.6.3.
3739 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3740 can use the new (as of 0.6.3) transform for SOME into MAP into
3742 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3743 values to (INTEGER 0 1000000), so that the time code will no longer
3744 occasionally get blown up by Linux returning 1000000 microseconds
3745 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3746 its output conform to the ANSI spec. (Alas, this makes its output
3747 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3748 * A full call to MAP NIL with a single sequence argument no longer conses.
3749 * fixes to problems pointed out by Martin Atzmueller:
3750 * The manual page no longer talks about multiprocessing as though
3751 it were currently supported.
3752 * The ILISP support patches have been removed from the distribution,
3753 because as of version 5.10.1, ILISP now supports SBCL without us
3754 having to maintain patches.
3755 * added a modified version of Raymond Toy's recent CMU CL patch for
3756 EQUALP comparison of HASH-TABLE
3758 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3760 * The system still can't cross-compile itself with
3761 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3762 entails), but at least it can compile more of itself that way
3763 than it used to be able to, and various buglets which were uncovered
3764 by trying to cross-compile itself that way have now been fixed.
3765 * This release breaks binary compatibility again. This time
3766 at least I've incremented the FASL file format version to 2, so that the
3767 problem can be detected reliably instead of just causing weird errors.
3768 * various new style warnings:
3769 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3770 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3771 ** using the lexical binding of a variable named in the *FOO* style
3772 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3773 as specified by ANSI.
3774 * *RANDOM-STATE* is no longer automatically initialized from
3775 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3776 default behavior of the system is to repeat its behavior every time
3777 it's run. If you'd like to change this behavior, you can always
3778 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3779 old convention there was no comparably easy way to get the system to
3780 repeat its behavior every time it was run.
3781 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3782 FTYPE declarations has been removed, in favor of their ANSI
3783 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3784 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3785 the types of their sequence arguments can be determined at compile time.
3786 This is done through a new open code expansion for MAP which eliminates
3787 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3788 argument types can be determined at compile time.
3789 * The optimizer now transforms COERCE into an identity operation when it
3790 can prove that the coerced object is already of the correct type. (This
3791 can be a win for machine generated code, including the output of other
3792 optimization transforms, such as the MAP transform above.)
3793 * Credit information has been moved from source file headers into CREDITS.
3794 * Source file headers have been made more standard.
3795 * The CASE macro now compiles without complaining even when it has
3798 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3800 * (Note that the way that the PCL macroexpansions were rewritten
3801 to accommodate the change in DEFGENERIC below breaks binary
3802 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3803 not run under sbcl-0.6.2. Once we get out of alpha releases,
3804 i.e. hit release 1.0.0, we'll probably try to maintain binary
3805 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3806 and sbcl-1.4.4. Until then, however, it might be fairly common
3807 for maintenance releases to break binary compatibility.)
3808 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3810 * The reporting of unhandled signals has been changed to print some
3811 explanatory text as well as the report form. (Previously only
3812 the report form was printed.)
3813 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3814 it defines, so that the compiler no longer issues undefined function
3815 warnings for compiled-but-not-yet-loaded generic functions.
3816 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3817 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3818 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3819 section of the ANSI spec are supported.
3820 * The "" nickname for the "KEYWORD" package has been removed.
3821 The reader still handles symbol tokens which begin with a package marker
3822 as keywords, but it doesn't expose its mechanism for doing so in the
3823 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3824 * The system now issues STYLE-WARNINGs for contradictory TYPE
3825 proclamations. (Warnings for contradictory FTYPE proclamations would
3826 be nice too, but those can't be done usefully unless the type system
3827 is made smarter about FUNCTION types.)
3828 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3829 systematized, so that "*target-*.lisp is supposed to exist only on the
3830 target and imply that there's a related file which exists on the
3831 host, and *host-*.lisp is supposed to exist only on the host and imply
3832 that there's a related file which exists on the target. This involves a
3833 lot of renaming. Hopefully the acute confusion caused by the renaming
3834 will be justified by the reduction in chronic confusion..
3835 ** runtime-type.lisp -> early-target-type.lisp
3836 ** target-type.lisp -> late-target-type.lisp
3837 ** early-host-format.lisp -> early-format.lisp
3838 ** late-host-format.lisp -> late-format.lisp
3839 ** host-error.lisp -> misc-error.lisp
3840 ** early-error.lisp -> early-target-error.lisp
3841 ** late-error.lisp -> late-target-error.lisp
3842 ** host-defboot.lisp -> early-defboot.lisp
3843 ** code/misc.lisp -> code/target-misc.lisp
3844 ** code/host-misc.lisp -> code/misc.lisp
3845 ** code/numbers.lisp -> code/target-numbers.lisp
3846 ** code/early-numbers.lisp -> numbers.lisp
3847 ** early-host-type.lisp -> early-type.lisp
3848 ** late-host-type.lisp -> late-type.lisp
3849 ** host-typep.lisp -> typep.lisp
3850 ** load.lisp -> target-load.lisp
3851 ** host-load.lisp -> load.lisp
3852 ** host-disassem.lisp -> disassem.lisp
3853 ** host-insts.lisp -> insts.lisp
3854 ** byte-comp.lisp -> target-byte-comp.lisp
3855 ** host-byte-comp.lisp -> byte-comp.lisp
3856 ** host-signal.lisp -> signal.lisp
3857 ** host-defstruct.lisp -> defstruct.lisp
3858 ** late-target-type.lisp -> deftypes-for-target.lisp
3859 Furthermore, several other previously target-only files foo.lisp (e.g.
3860 hash-table.lisp and random.lisp) have been split into a target-and-host
3861 foo.lisp file and a target-only target-foo.lisp file, with their key type
3862 definitions in the target-and-host part, so that the cross-compiler will
3863 know more about target types.
3864 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3865 gone away. In their place are various *BACKEND-FOO* variables
3866 corresponding to the slots of the old structure.
3867 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3868 into the target SBCL has been fixed.
3869 * The chill.lisp system for loading cold code into a running SBCL
3871 * Support for the CMU CL "scavenger hook" extension has been removed.
3872 (It was undocumented and unused in the CMU CL sources that SBCL was
3873 derived from, and stale in sbcl-0.6.1.)
3874 * Various errors in the cross-compiler type system were detected
3875 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3876 (enabling various consistency checks). Many of them were fixed,
3877 but some hard problems remain, so the compiler is back to
3878 running without *TYPE-SYSTEM-INITIALIZED* for now.
3879 * As part of the cross-compiler type system cleanup, I implemented
3880 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3881 * I have started adding UNCROSS calls throughout the type system
3882 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3883 on cross-compiler input files will be able to go away, and various
3885 * CONSTANTP now returns true for quoted forms (as explicitly required
3888 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3890 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3891 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3892 (MAKE-LIST -1). In the longer term, it ought to become true
3893 that declarations are assertions even at SAFETY 1. For now, it's not
3894 quite true even at SAFETY 3, but it's at least more nearly true..
3895 (Note that this change seems to increases the size of the system by
3896 O(5%) and to decrease the speed of the compiler by 20% or more.)
3897 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3898 for the problem of printing dozens of lines of distracting information
3899 about low-level system machinery as part of the top stack frame
3900 on entry to the debugger when an undefined function was called.
3901 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3902 is not reset to COMMON-LISP-USER.
3903 * Compilation of stuff related to dyncount.lisp has been made conditional
3904 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3905 smaller. The various dyncount-related symbols have been moved into
3906 a new "SB-DYNCOUNT" package.
3907 * tty-inspect.lisp has been renamed to inspect.lisp.
3908 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3909 feature has gone away. (When we eventually port to other flavors of
3910 libc and/or Unix, we'll try to make the differences between flavors
3911 invisible at the user level.)
3912 * Various other *FEATURES* tags, and/or their associated conditionals,
3913 have been removed if obsolescent, or given better documentation, or
3914 sometimes given more-mnemonic names.
3916 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3918 * tidied up "make.sh" script
3919 * tidied up system directory structure
3920 * better "clean.sh" behavior
3921 * added doc/FOR-CMUCL-DEVELOPERS
3922 * many many small tweaks to output format, e.g. removing possibly-confusing
3923 trailing #\. character in DESCRIBE-INSTANCE
3924 * (EQUALP #\A 'A) no longer signals an error.
3925 * new hashing code, including EQUALP hashing
3926 * tidied up Lisp initialization and toplevel
3927 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3928 * command line argument processing
3929 * added POSIX-GETENV function to deal with Unix-ish environment variables
3930 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3931 terminating SBCL on EOF
3932 * non-verbose GC by default
3933 * There is no more "sbcl" shell script; the sbcl file is now the C
3934 runtime executable (just like CMU CL).
3935 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3937 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3938 symbol and package information as Lisp native symbol and package objects
3939 instead of strings naming symbols and strings naming packages. This way,
3940 whenever packages are renamed (as in warm init), debug information is
3941 transformed along with everything else.
3942 * tweaked the optimization policy declarations which control the building
3943 of SBCL itself. Now, among other things, the system no longer saves
3944 source location debugging information. (This helps two problems at once
3945 by reducing SBCL size and by keeping SBCL from trying to look for its
3946 sources -- which may not exist -- when reporting errors.)
3947 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3948 debugging and testing purposes
3949 * cleaned up printing, making the printer call PRINT-OBJECT for
3950 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3951 methods, giving nearly-ANSI behavior
3952 * converted almost all special variables to use *FOO* naming convention
3953 * deleted PARSE-TIME functionality, since it can be done portably
3954 * moved some files out of cold init into warm init
3955 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3957 * regularized formatting of source files
3958 * added an install.sh script
3959 * fixed ridiculous memory usage of cross-compiler by making
3960 compiler/alloc.lisp not try to do pooling unless it can hook
3961 itself into the GC of the cross-compilation host. Now the system
3962 builds nicely on my old laptop.
3963 * added :SB-ALLOC in target-features.lisp-expr
3964 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3965 was not implemented)
3966 * re-did condition handling and note reporting in the compiler. Notes
3967 are no longer handled by signalling conditions. Style warnings
3968 and warnings are handled more correctly and reported in such a way
3969 that it's easy to find one or the other in your output (so that you
3970 can e.g. figure out which of many problems caused COMPILE-FILE to
3972 * changed the severity of several compiler warnings from full WARNING
3973 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3974 compiler note reporting so that it doesn't use the condition system
3975 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3976 COMPILE-FILE command)
3977 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3978 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3979 no longer supported, since their implementation was deeply intertwingled
3980 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3981 * removed bogus "support" for compiler macros named (SETF FOO), and
3982 removed the compiler macro for SETF INFO (but only after making a fool
3983 of myself on the cmucl-imp mailing list by posting a bogus patch for
3984 DEFINE-COMPILER-MACRO..)
3985 * Compiled files containing forms which have side effects on the Lisp
3986 reader (such as DEFPACKAGE forms) are now handled more correctly.
3987 (Compiler queuing of top level lambdas has been suppressed by setting
3988 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3989 * deleted various currently-unused source files, e.g. gengc.lisp. They
3990 may be added back at some point e.g. when porting to other architectures,
3991 but until they are it's distracting to distribute them and to try to
3993 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3994 there were so many of them they're just distractions, and UNCROSS is
3995 known to be able to handle the current sources
3996 * moved PROFILE functionality into TRACE, so that it will be clear
3997 how the wrapping and unwrapping of functions when you profile them
3998 interacts with the wrapping and unwrapping of functions when you
3999 trace them. (Actually, the functionality isn't there yet, but at least
4000 the interface specification is there. Hopefully, the functionality will
4001 arrive with some future maintenance release.)
4002 * removed host-oops.lisp
4003 * changed signature of QUIT function to allow UNIX-CODE argument
4004 * fixed READ-SEQUENCE bug
4005 * tweaked verbose GC output so that it looks more like the progress
4006 output that ANSI specifies for functions like LOAD
4007 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4008 * added <http://sbcl.sourceforge.com> to the banner information printed by