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.
6 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
7 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
8 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
10 * bug fix: two potential GC deadlocks affecting threaded builds.
11 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
12 returns T (reported by Anton Kazennikov)
13 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
14 with non-variable places
15 * bug fix: remove a race condition in the setting of
16 funcallable-instance functions, this should make threaded CLOS
17 code more stable against memory faults.
18 * bug fix: corruption of specials when unbinding is interrupted by an
19 asynchronous unwind (reported by Hannu Koivisto)
20 * improvement: the debugger will now also display local variables that
21 are only used once, for code compiled with a DEBUG optimization quality
24 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
25 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
26 * incompatible change: External-format support for FFI calls. The
27 SB-ALIEN:C-STRING no longer implies an ASCII
28 external-format. Instead the string is subject to external-format
29 conversion. Additionally return values of type SB-ALIEN:C-STRING
30 are by default of type (SIMPLE-ARRAY CHARACTER), not
31 SIMPLE-BASE-STRING. For an alien type that behaves like the old
32 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
33 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
34 * incompatible change: SB-EXT package no longer contains the
35 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
36 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
38 * incompatible change: the single-stepper is no longer available
39 on Alpha, Mips and Sparc platforms.
40 * minor incompatible change: the direct superclasses of
41 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
42 not (STANDARD-OBJECT FUNCTION). This makes the
43 class-precedence-lists of GENERIC-FUNCTION and
44 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
46 * minor incompatible change: the default stream external format on
47 non-unicode SBCL is no longer detected from the locale, but is always
49 * new feature: Add a version of evaluator that uses an interpreter instead
50 of the compiler. EVAL still uses the compiler by default, to switch it
51 to use the interpreter, set the value of the variable
52 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
53 * minor incompatible change: the single-stepper REPL has been merged
54 with the normal debugger (see the "Stepping" heading of the debugger help
55 for more details). The debugger command STEP will no longer switch
56 to the single-stepper REPL.
57 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
59 * bug fix: loading of default sysinit file works. (thanks to Leonid
61 * bug fix: better detection of circularities in the file-compiler.
62 (reported by Marco Monteiro)
63 * bug fix: the CL pathname functions now work with files that have
64 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
65 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
66 systems with Unix98 pty semantics.
67 * bug fix: ASDF-INSTALL will now work with bsd tar.
68 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
70 * bug fix: timers expiring in dead threads no longer cause a
71 type-error (reported by Paul "Nonny Mouse").
72 * bug fix: thanks to more lightweight single-stepper instrumentation,
73 code compiled with (DEBUG 3) will compile and execute significantly faster,
74 and will have more accurate type-inferencing than before
75 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
76 whose bindings are modified
77 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
78 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
79 ** replace dummy implementations of CL:MACHINE-INSTANCE and
80 CL:SOFTWARE-VERSION with proper version
82 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
83 * feature: implemented the READER-METHOD-CLASS and
84 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
86 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
88 * optimization: faster LOGCOUNT implementation on x86 and x86-64
89 (thanks to Lutz Euler)
90 * optimization: hashing of general arrays and vectors has been
91 improved. (reported by Any Fingerhut)
92 * enhancement: SB-INTROSPECT is now able to find definitions of
93 profiled functions. (thanks to Troels Henriksen)
94 * enhancement: compiler-macro expansion applies now to FUNCALL forms
95 as well, allowing compiler-macros for SETF-functions to expand.
96 * enhancement: step-instrumentation no longer wraps known
97 single-value functions with multiple-value context, allowing
98 better type inference.
99 * fixed bug #337: use of MAKE-METHOD in method combination now works
100 even in the presence of user-defined method classes. (reported by
101 Bruno Haible and Pascal Costanza)
102 * fixed bug #339(c): if there are applicable methods not part of any
103 long-form method-combination group, call INVALID-METHOD-ERROR.
104 (reported by Bruno Haible)
105 * fixed bug #361: the :FUNCTION initarg in the protocol for
106 initialization of methods can now be used to override
107 internally-produced optimized functions. (reported by Bruno
109 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
110 system-provided lambda expression no longer cause warnings about
111 unbound #:|pv-table| symbols.
112 * bug fix: improved the handling of type declarations and the
113 detection of violations for keyword arguments with non-constant
115 * bug fix: potentially erronous calls to PATHNAME and
116 MERGE-PATHNAMES were being flushed in some cases.
117 (reported by Richard Kreuter)
118 * bug fix: compiled calls to TYPEP were mishandling obsolete
119 instances. (reported by James Bielman and Attila Lendvai)
120 * bug fix: format strings with ~> without matching ~< no longer
121 trigger an AVER, but signal an understandable error instead.
122 (reported by Antonio Martinez)
123 * bug fix: specifying an output-file without a pathname-type for
124 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
125 of input-file instead of "fasl". (reported by Robert Dodier)
126 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
127 (reported by James Y Knight).
128 * bug fix: compiler-macros-function did not consider the environment
129 argument for shadowing by local functions.
130 * bug fix: compiler-macros expansion was inhibited by local INLINE
132 * bug fix: inline expansions of known functions were subject to
133 step-instrumentation in high debug policies, leading to problems
135 * bug fix: compiler failed to differentiate between different CONS
137 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
138 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
139 * bug fix: #368: incorrect use of expressed vs. upgraded array
141 * bug fix: #306a: more precise unions of array types.
142 * thread-safety improvements:
143 ** CONDITION-WAIT could return early on Linux, if the thread was
144 interrupted and subsequently continued with SIGCONT.
145 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
147 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
148 * added support for the ucs-2 external format. (contributed by Ivan
150 * minor incompatible change: pretty printing of objects of type
151 (cons symbol) is, in the default pprint-dispatch-table, now
152 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
154 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
155 called later in a class's lifetime, possibly as late as when the
156 first instance of the class is created. Previously,
157 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
158 class became finalizeable.
159 * fixed bug: FILE-POSITION sometimes returned inconsistent results
160 for multibyte external-format streams. (thanks to "vbzoli")
161 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
162 with :ALLOCATION :CLASS inherited from superclasses of the
164 * fixed bug: anonymous classes can now be created using the :NAME
165 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
166 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
167 * fixed bug: core-files saved with :EXECUTABLE T can again be
168 executed when SBCL_HOME isn't set. (reported by James Knight)
169 * fixed bug: toplevel LOCALLY forms with declarations could
170 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
171 * fixed bug: printing from several different threads using different
172 values of *print-case* could cause invalid output, due to
173 some internal special variables of the printer not being bound
174 thread-locally (reported by Max Mikhanosha)
175 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
176 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
177 (reported by Pascal Costanza)
178 * fixed bug: FINALIZE-INHERITANCE is called from
179 REINITIALIZE-INSTANCE on classes when the class has previously
180 been finalized, as required by AMOP.
181 * minor code generation optimizations:
182 ** better register allocation in CLOS dispatching functions
183 ** overflow detection when coercing signed bytes to fixnums on x86-64
184 ** is now implemented with one IMUL instruction instead of three shifts
185 ** more efficient bit-vector access on x86 and x86-64
186 ** more efficient access to raw structure slots on x86 and x86-64
187 * fixed some bugs revealed by Paul Dietz' test suite:
188 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
191 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
192 * feature: thread support on Solaris/x86, and experimental thread support
194 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
196 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
197 default method for SLOT-UNBOUND.
198 * minor incompatible change: prevent the user from specializing the
199 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
200 somewhat counter to the intent of the protocol, I (CSR) think, and
201 additionally it just doesn't work in SBCL as currently
202 implemented, thanks to optimizations (that are always valid for
203 the other three SLOT-VALUEish functions, but not for the setter).
204 * minor incompatible change: the :SB-LDB feature is now enabled by
205 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
206 the low-level debugger.
207 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
208 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
209 down to an SB-SYS:FD-STREAM, allowing direct communication in
210 more cases. Temporary files and pipes are still used as a fallback
212 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
213 on the non-reentrant C functions localtime(3) and gmtime(3).
214 * bug fix: global optimization policy was not visible in LOCALLY and
216 * bug fix: class objects can be used as specializers in methods.
217 (reported by Pascal Costanza)
218 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
219 with a type error. (reported by blitz_ on #lisp)
220 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
221 failed with a type error. (reported by Pascal Bourguignon)
222 * bug fix: merging pathnames against defaults with :DIRECTORY
223 starting with '(:RELATIVE :BACK) should preserve the :BACK.
224 (reported by James Y Knight)
225 * bug fix: saving large (>2GB) cores on x86-64 now works
226 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
227 constant index and a (simple-array (signed-byte 32)) array
228 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
229 error (patch by Robert J. Macomber)
230 * feature: TIME now displays the amount of run-time spent in GC
231 * bug fix: The debugger now does a better job of respecting
232 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
233 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
234 workaround for bug 403.)
235 * bug fix: floating point exception handling now works on FreeBSD
236 (thanks to NIIMI Satoshi)
237 * fixed some bugs revealed by Paul Dietz' test suite:
238 ** MISC.641: LET-conversion were not supposed to work in late
240 * improvements to the Win32/x86 port:
241 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
242 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
244 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
245 * new feature: source path information is generated for macro-expansion
246 errors for use in IDE's like Slime (thanks to Helmut Eller)
247 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
248 * bug fix: compiler does not loop forever on an invalid type in
250 * improvement: compilation of most CLOS applications is significantly
252 * optimization: added a limited bytecode compiler for simple toplevel
253 forms, speeding up compilation and FASL loading
254 * bug fix: the statistical profiler now properly distinguishes anonymous
257 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
258 * minor incompatible change: in sbcl-0.9.11 (but not earlier
259 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
260 into subclasses of the finalized class. Now user calls to
261 FINALIZE-INHERITANCE finalize just the one class, and calls by the
262 system return before any subclasses are finalized.
263 * minor incompatible change: The reader no longer ignores errors
264 regarding non-existent packages in #+ and #- feature tests.
265 * new feature: command line options --no-sysinit, --no-userinit to
266 inhibit loading the corresponding init files
267 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
268 for sending data through UDP sockets (thanks to François-René Rideau)
269 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
270 error is signalled instead (thanks to François-René Rideau)
271 * minor incompatible change: Error signalling behaviour of lexical
272 operations violating package locks has changed slightly. Refer to
273 documentation on package locks for details.
274 * enhancement: EVAL can process IF-expressions without resorting to the
276 * bug fix: Some operations involving SIN, COS, or TAN weren't being
277 constant-folded properly due to missing out-of-line functions.
278 (reported by Mika Pihlajamäki)
279 * bug fix: LISTEN sometimes returned T even in cases where no data was
280 immediately available from the stream
281 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
282 were reversed. (reported by Levente Mészáros)
283 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
284 when given a vector displaced to an adjustable array. (reported
286 * bug fix: derivation of float boundaries from numbers outside the
287 appropriate float range (reported by John Wiseman)
288 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
290 * bug fix: precomputing cache entries for generic functions with
291 some subclasses of specializers as yet invalid does not attempt to
292 fill a cache line with a negative offset. (reported by Levente
294 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
295 allow condition class objects as arguments to DOCUMENTATION and
296 (SETF DOCUMENTATION); only find and set documentation for
297 structure names for the STRUCTURE doc-type. (suggested by Gary
299 * improvements to the Win32/x86 port:
300 ** (user-homedir-pathname) and default initialization file
301 locations now know about the user's "Documents and Settings"
302 directory (thanks to Yaroslav Kavenchuk)
303 ** run-program is implemented (thanks to Mike Thomas)
304 ** sockets support (thanks to Timothy Ritchey)
305 ** better backtrace support (thanks to Alastair Bridgewater)
306 ** sb-grovel supported
307 ** asdf-install and sb-posix work somewhat
308 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
309 * improvements to the Solaris/x86 port:
310 ** works on Solaris 11/Solaris Express
311 ** floating-point exception handling support
312 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
313 * fixed some bugs revealed by Paul Dietz' test suite:
314 ** REMOVE-METHOD returns its generic function argument even when
315 the method is not one of the generic functions' methods.
316 ** objects declared as MEMBER types can be admissible arguments to
318 ** printing characters should simply be printed by the FORMAT ~:C
320 ** compiler failure when compiling functions with hairy constant
321 defaults for optional parameters.
322 ** compiler produces wrong code when MAYBE-INLINE-expanding a
323 function, which is already optimized.
325 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
326 * new platform: experimental support for SBCL x86/Darwin, including
327 MacOS X 10.4.5 on Intel.
328 * new feature: Unicode character names are now known to the system
329 (through CHAR-NAME and NAME-CHAR).
330 * new feature: the filesystem location of SBCL's core file is
331 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
332 * minor incompatible change: the contrib modules SB-POSIX and
333 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
334 this change is to make it easier to distribute
335 location-independent binaries.
336 * bug fix: as implied by AMOP, standardized classes no longer have
337 slots named by external symbols of public packages. (reported by
339 * optimization: calling functions via constant symbols -- as in
340 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
341 via the function object as in (FUNCALL #'FOO).
342 * optimization: on x86 and x86-64, the calling convention for the
343 case of dealing with an unknown number of values has been altered
344 to be friendlier to the prediction heuristics implemented,
345 particularly with reference to CALL and RET pairing. (thanks to
346 Alastair Bridgewater)
347 * optimization: on x86, the code for access to thread-local symbol
348 values and binding/unbinding of thread-local symbols is smaller.
349 (thanks to Alastair Bridgewater)
350 * enhancement: CONSTANTP is now able to determine constantness of
351 more complex forms, including calls to constant-foldable standardized
352 functions and some special forms beyond QUOTE.
353 * fixed bug: occasional GC crashes on Solaris/x86
354 * optimization: x86-64 supports stack allocation of results of simple
355 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
356 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
357 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
358 (thanks to James Knight)
359 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
360 by the standard (thanks to Stephen Compall)
362 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
363 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
364 be used for bundling the runtime and the core file into one
365 executable binary. This feature is not currently supported on all SBCL
366 platforms. (thanks to James Bielman and NIIMI Satoshi)
367 * new feature: a generational or ephemeral garbage collector is now
368 the default on the PowerPC platform (both Linux and Darwin). The
369 old Cheney (stop and copy) collector is a build-time option.
370 (thanks to Cyrus Harmon, after Raymond Toy)
371 * minor incompatible change: the method by which SBCL finds its
372 contributed modules has changed; it no longer relies on symbolic
373 links from an $SBCL_HOME/systems directory, but searches directly
374 in the subdirectories of $SBCL_HOME.
375 * enhancement: the dynamic heap size on the Linux/PPC platform is
376 markedly larger, even using the older Cheney garbage collector.
377 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
378 certain complicated string types. (reported by Gary King)
379 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
380 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
381 * fixed bug: DOCUMENTATION on structures no longer signals an
382 error if no structure type of the right name exists. (reported by
384 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
385 applicable non-standard primary or :AROUND method on
386 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
387 no longer cause unbound variable errors. (reported by Kevin Reid)
388 * fixed bug: metacircle resolution in cases where methods have slots
389 added before the slots from STANDARD-METHOD. (reported by Jean
391 * fixed bug: the Power PC assembly code for calling into lisp
392 sometimes computed its return address wrongly (depending on the
393 15th bit of the address, fixed at link time). This used to
394 manifest itself in a segmentation violation while building PCL.
395 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
396 many others over the years)
397 * fixed bug: ROOM no longer reports silly numbers for stack usage if
398 the stack pointer (treated as a signed integer) is negative.
399 (thanks to Peter van Eynde)
401 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
402 * new platform: experimental support for the Windows operating
403 system has been added. (thanks to Alastair Bridgewater)
404 * fixed several bugs in and robustified the PPC FFI (including
405 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
406 * bug fix: allow non-simple string symbol names (reported by Paul
408 * bug fix: interrupt handling on NetBSD (thanks to Richard M
410 * bug fix: saving a core corrupted callbacks on x86/x86-64
411 * bug fix: closed a loophole in metacircularity detection and
412 grounding in the PCL implementation of CLOS.
413 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
414 (reported by tomppa on #lisp)
415 * bug fix: clear the direction flag on Lisp -> C transitions, as
416 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
417 (reported by Andrej Grozin and Hendrik Maryns)
418 * optimization: major improvements to GC efficiency on GENCGC platforms
419 * optimization: faster implementation of EQUAL
420 * optimization: emit more efficient opcodes for some common
421 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
422 * optimization: several other minor code-generation improvements on x86-64
423 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
425 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
426 * minor incompatible change: (SETF CLASS-NAME) and (SETF
427 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
428 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
429 by ANSI as a generic function, and for consistency (SETF
430 GENERIC-FUNCTION-NAME) is treated likewise.
431 * fixed bug #233.b: make constraint propagation notice when a variable
432 value is changed after it is referenced but before it is used
433 * fixed bug #296: no more arbitrary behaviour from filenames with
434 odd characters as --load arguments. (reported by Adam Warner)
435 * fixed bug #390: :CHARACTER-SET pathname components now work as
436 expected. (reported by Tim Daly Jr)
437 * fixed bug #391: complicated :TYPE intersections in slot
438 definitions no longer cause an error in PCL internals.
439 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
440 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
441 returning the number of octets which would be written to the
442 file-stream. (thanks to Robert J. Macomber)
443 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
444 arguments to the internal stream functions. (thanks to David
446 * bug fix: stack exhaustion is now handled on the x86/SunOS
447 platform. (thanks to Daisuke Homma)
448 * bug fix: more accurate ROOM results on GENCGC platforms
449 * optimization: improved type inference for arithmetic-for
450 index variables in LOOP
451 * optimization: faster floating-point SQRT on x86-64
452 * fixed some bugs revealed by Paul Dietz' test suite:
453 ** DOCUMENTATION returns NIL instead of "" for method combinations
454 that don't have a docstring
456 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
457 * minor incompatible change: (SETF CLASS-NAME) and (SETF
458 GENERIC-FUNCTION-NAME) are no longer generic functions, and
459 therefore are not customizeable by user code (as seems to be at
460 least permitted and maybe required by AMOP). As a consolation,
461 however, the SBCL implementation of these functions now calls
462 REINITIALIZE-INSTANCE as specified by AMOP.
463 * bug fix: slot-definition documentation is propagated as per ANSI
464 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
465 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
466 argument, as do eslotd initialization methods. (from Pascal
467 Costanza's "Closer" project)
468 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
469 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
471 * bug fix: it is now possible to have more than one subclass of
472 STANDARD-GENERIC-FUNCTION without causing stack overflow.
473 (reported by Bruno Haible, Pascal Costanza and others)
474 * bug fix: the dependent update protocol now works for generic
475 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
477 * bug fix: condition-class instances corresponding to
478 DEFINE-CONDITION forms are now created eagerly. (reported by
479 Kalle Olavi Niemitalo on comp.lang.lisp)
480 * bug fix: floating point printing is more accurate in some
481 circumstances. (thanks to Simon Alexander)
482 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
483 merged with *DEFAULT-PATHNAME-DEFAULTS*.
484 * bug fix: callbacks on OS X now preserve stack-alignment.
485 (thanks to Cyrus Harmon)
486 * enhancement: the x86-64 disassembler is much better at
487 disassembling SSE instructions. (thanks to Lutz Euler)
488 * enhancement: CHECK-TYPE on a variable now allows type-inference
489 (thanks to Vincent Arkesteijn)
490 * optimization: improved performance of EUC-JP external format.
491 (thanks to NIIMI Satoshi)
492 * optimization: performance improvements to IO on file streams of
493 :ELEMENT-TYPE CHARACTER
494 * optimization: much faster memory allocation on x86-64
495 * optimization: faster garbage collections (latency and throughput)
497 * optimization: faster bignum subtraction and division on x86 and x86-64
498 (thanks to Lutz Euler)
499 * optimization: more accurate type-inference for loops that use a
500 floating point index variable or a negative step.
502 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
503 * bug fix: add a workaround to SBCL looping infinitely at startup on
504 Linux kernels with apparently buggy implementations of personality().
505 (thanks to Svein Ove Aas)
506 * bug fix: Unicode symbols are correctly printed in LDB backtraces
507 (thanks to David Lichteblau)
508 * bug fix: local bindings shadow global symbol macros for
509 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
510 * optimization: non-open coded uses of numeric comparison operators
511 (e.g. >) no longer cons when called with more than one parameter
512 on platforms supporting dynamic-extent allocation.
513 * enhancement: saving cores with foreign code loaded is now supported
514 on MIPS/Linux in addition to the previously supported platforms.
515 * bug fix: FIND-RESTART now tests for activity, not applicability when given
516 a restart object as identifier. (reported by Helmut Eller for CMUCL)
517 * bug fix: division by zero in sb-sprof when no samples were collected
518 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
519 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
520 (thanks to Alexander Kjeldaas)
521 * new feature: ignoring the return values of destructive functions
522 like NREVERSE signals a compile-time style-warning.
523 (thanks to Kevin Reid)
524 * new feature: the alignment of alien structure fields can be
525 explicitly specified. (thanks to Cyrus Harmon)
526 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
527 and dump core on SIGQUIT
529 ** incompatible change: threads do not inherit values of specials
530 from their parents (see manual)
531 ** bug fix: threads stacks belonging to dead threads are freed by the
532 next exiting thread, no need to gc to collect thread stacks anymore
533 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
534 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
535 ** bug fix: dynamic variable and thread start related gc lossage
536 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
538 * fixed some bugs revealed by Paul Dietz' test suite:
539 ** SUBTYPEP is slightly more accurate on heinously complicated
540 CONS types where some of the members have uncertain (in the
541 NIL, NIL sense) type relationships to each other.
543 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
544 no longer purified unless :PURIFY T is explicitly specified.
545 ** Non-purified cores are significantly smaller than before
547 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
548 * new feature: timers based on Zach Beane's excellent timer package
549 * added support for the following external formats: koi8-u,
550 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
551 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
552 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
553 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
554 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
555 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
556 * incompatible change: a threaded SBCL will no longer revert to
557 non-threaded mode on non-NPTL systems, but refuse to start entirely.
558 * bug fix: interrupts are disabled until startup is complete; no
559 more sigsegvs when receiving a signal to soon
560 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
562 * bug fix: add a workaround for the memory randomization features in
563 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
564 workaround will only be in effect on systems with the proc filesystem
566 * bug fix: printing objects of type HASH-TABLE signals a
567 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
569 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
570 have been running for over 50 days. (reported by Gilbert Baumann)
571 * bug fix: the logic for getting names of functions gets less
572 confused when confronded with alternate-metaclass
573 funcallable-instances. (reported by Cyrus Harmon)
574 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
575 consistent, even on internal alternate-metaclass objects.
576 * bug fix: flush closure information collected by physical
577 environment analysis prepass before the main pass. (bug reported
579 * bug fix: compiler pack phase does not modify a hashtable, which is
580 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
581 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
582 may be expanded. (reported by Alexander <malishev> on c.l.l)
583 * bug fix: no more style-warnings when compiling code using LAST
584 under high-SPEED low-SPACE settings. (reported by David Wallin
586 * fixed some incompatibilities between SBCL's MOP and the MOP
588 ** the METAOBJECT class is now implemented;
589 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
590 STANDARD-OBJECT, as required;
591 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
592 now compatible (as required by VALIDATE-SUPERCLASS); there
593 remains a consistency requirement over the presence or absence
594 of the FUNCTION class in the superclass list of finalized
595 classes; see the manual for more details;
596 ** the :AROUND method for COMPUTE-SLOTS on
597 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
598 requested slot ordering.
600 ** bug fix: parent thread now can be gc'ed even with a live
602 ** bug fix: binding a special with PROGV to no value is not
603 the same as that symbol not having been bound (thanks to
605 ** bug fix: binding specials is thread safe (thanks to
607 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
608 and SB-SPROF do not die with 'no handler for signal XX in
609 interrupt_handle_now(..)' anymore
610 ** bug fix: WITH-TIMEOUT works with multiple running threads
611 * fixed some bugs revealed by Paul Dietz' test suite:
612 ** ENSURE-GENERIC-FUNCTION should take a method class object for
613 the :method-class keyword argument.
615 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
616 * new port: the Solaris operating system on x86 processors is now
617 mostly supported, though some rough edges in the environment
618 remain. (thanks to Daisuke Homma)
619 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
620 space for its heap. (thanks to Thiemo Seufer)
621 * optimization: SBCL on MIPS platforms now supports dynamic-extent
622 closures. (thanks to Thiemo Seufer)
623 * minor incompatible change: eof selects abort in the debugger.
624 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
625 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
627 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
628 deprecated, and will go away in a future revision of SBCL.
629 * minor incompatible change: GC-ON and GC-OFF are no longer
630 implemented with a counter, it does not matter how many times gc
631 is switched on or off
632 * bug fix: discriminating functions for generic function classes
633 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
634 make invalid assumptions about method precedence order. (reported
636 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
637 platforms. (thanks to Thiemo Seufer)
638 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
639 * bug fix: correct pathname printing with printer escaping is on.
640 (thanks to Kevin Reid)
641 * bug fix: complex VOP definitions in "user-space" no longer trigger
642 package locks. (reported by Zach Beane)
643 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
644 for funcallable-instance functions, and is no different from
646 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
647 characters in the current readtable. (reported by Nicholas Neuss)
648 * bug fix: SUBTYPEP on various CONS types returns more a more
649 accurate acknowledgment of its certainty.
650 * optimizations: REMOVE-DUPLICATES now runs in linear time on
651 lists in some cases. This partially fixes bug 384.
652 * flush all standard streams before prompting in the REPL and the
654 * bug fix: signal handling and triggering gc do not conflict
655 directly anymore, in particular a high frequency sb-sprof does
656 not prevent gc from running
657 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
658 approximation for timezone and DST information between the
659 universal time epoch and the smallest negative 32-bit time_t.
660 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
661 year 1899 so long as the encoded time is non-negative.
662 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
663 the correct answer for non-constant rotate amounts when performing
664 an inline 32-bit rotation.
666 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
667 there is only one thread in the session
668 ** bug fix: memory leak for streams created in one thread and
669 written to in another
670 ** bug fix: lockup when compiled with gcc4
671 ** bug fix: race that allows the gc to be triggered when gc is
673 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
674 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
676 ** bug fix: finished threads are gc'ed properly
677 * fixed some bugs revealed by Paul Dietz' test suite:
678 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
679 has arguments with a different set of applicable methods from
680 the orignal arguments.
681 ** The type error thrown by MAP now has a correct expected-type
683 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
684 name a compiled function.
685 ** Three MISC tests where a large bignum was improperly coerced to
686 a float that couldn't represent that bignum during type
687 derivation were fixed.
688 ** SUBTYPEP can now handle types involving the negation of a
689 list-form FUNCTION type.
690 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
691 as (COMPLEX (AND RATIO (NOT FIXNUM))).
692 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
694 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
695 * New feature: Experimental support for bivalent streams: streams
696 opened with :element-type :default now allow character and binary
697 (unsigned-byte 8) I/O
698 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
699 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
701 * bug fix: no more highly sporadic "couldn't check whether ~S is
702 readable" when reading a stream and an interrupt hits in the middle
703 of a select system call
704 * compiler better recognizes complex arrays (reported by Eduardo
706 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
707 for variables on SBCL built with linkage-tables. (reported by Luis
709 * various error reporting improvements.
710 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
711 (Thanks to Thiemo Seufer)
712 * bug fix: nested reader invokations work correctly
713 * bug fix: it is possible to have simultaneous references to foreign
714 code and foreign data with the same name.
716 ** added x86-64 support
717 ** incompatible change: the threading api now works with thread
718 objects instead of thread ids
719 ** bug fix: threads are protected from signals and interruption when
720 starting up or going down
721 ** bug fix: a race where an exiting thread could lose its stack to gc
722 ** bug fix: don't halt on infinite error in threads if possible
723 ** fixed numerous gc deadlocks introduced in the pthread merge
724 ** bug fix: fixed thread safety issues in read and print
725 ** bug fix: debugger doesn't hang on session lock if interrupted at
726 an inappropriate moment
727 ** bug fix: run-program is now thread safe(r)
728 ** bug fix: inner with-recursive-lock no longer releases the mutex
729 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
730 * fixed some bugs revealed by Paul Dietz' test suite:
731 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
732 EXPECTED-TYPE slots filled.
733 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
735 ** MAKE-CONDITION accepts classes as type-designators.
736 ** COMPILE may never return NIL.
737 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
738 range before calling Unix time functions
740 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
741 * numerous signal handling fixes to increase stability
742 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
743 * minor incompatible change: we now correctly canonize default
744 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
745 as per the MOP, rather than the historical (INITARG INITFUNCTION
746 INITFORM). (reported by Bruno Haible)
747 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
748 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
749 for more information.
750 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
751 pathname is a directory pathname.
752 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
753 * dynamic space size on PPC has been increased to 768Mb. (thanks to
755 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
756 :METACLASS argument in addition to a class name. (reported by
757 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
758 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
759 (reported by Bruno Haible; patch from Gerd Moellmann)
760 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
762 * bug fix: more cleanups to the floating point exception handling on
763 x86-64 (thanks to James Knight)
764 * bug fix: the compiler does not try to constant fold calls of
765 COERCE and BIT-* functions when they return freshly constructed
766 objects. (reported by Nikodemus Siivola and Paul Dietz)
767 * optimization: improved the MIPS versions of generic (in the
768 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
769 * optimization: direct conversion of (unsigned-byte 32) to floats on
770 the PowerPC platform.
771 * optimization: structure instances with raw slots now use less
772 memory, and probably show better memory locality. (thanks to
774 * optimization: DYNAMIC-EXTENT declarations for lists and closures
775 are treated as requests for stack allocation on the x86-64,
776 Alpha-32, PPC and SPARC platforms.
777 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
778 the wrong file descriptor; implementation of SOCKET-OPEN-P.
779 (thanks to Tony Martinez)
781 ** gcing a dead thread can no longer lead to lockups
782 ** threads block signals until they are set up properly
783 ** errno is no longer shared by threads
784 ** interrupt-thread restores the eflags register on x86
785 ** fixed some lockups due to gc/thread interaction
786 * fixed some bugs revealed by Paul Dietz' test suite:
787 ** invalid dotted lists no longer raise a read error when
789 ** signal an error if a symbol that names a declaration is used
790 as the name of a type, or vice versa
791 ** allow using the (declare (typespec var*)) abbreviation for
792 (declare (type typespec var*)) with all type specifiers
793 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
794 ** always return NIL from PPRINT-POP when OBJECT is NIL
795 ** don't signal errors when pretty-printing malformed LABELS,
796 FLET or MACROLET forms
797 ** declarations in a DOLIST body are also in scope for the
799 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
801 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
804 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
805 * fixed cross-compiler leakages that prevented building a 32-bit
806 target with a 64-bit host compiler.
807 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
808 opened with :IF-EXISTS :OVERWRITE.
809 * fixed bug 281: error for an invalid qualifier in a short-form method
810 combination method is not signalled until the faulty method is called.
811 * bug fix: iteration variable type inferrer failed to deal with open
812 intervals. (reported by Alan Shields)
813 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
814 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
815 or more runtime options were provided to the sbcl binary.
816 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
817 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
818 * optimization: there's now a fast-path for fixnum arguments in the
819 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
820 * optimization: the code generated on x86-64 is more compact thanks
821 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
822 * bug fix: floating-point exception handling now partly works on
823 x86-64. (Thanks to James Knight)
824 * improvement to the MIPS backend from Thiemo Seufer: C-style
825 64-bit long long arguments and return values to alien functions
827 * fixed some bugs revealed by Paul Dietz' test suite:
828 ** the type-error signalled from WARN has a filled-in DATUM slot.
829 ** the type-error required when a stream is not associated with
830 a file has the stream as its datum.
831 ** type-errors on single-floats on x86-64 no longer have
832 :INVALID-OBJECT as the datum
833 ** the type-errors signalled for invalid function names now have
834 a correct expected type
835 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
836 for structures defined with a :TYPE. Documentation strings for
837 typed structures are no longer immediately discarded
838 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
839 broadcast streams and synonym streams. FILE-LENGTH now also works
840 on broadcast streams.
842 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
843 * incompatible change: the --noprogrammer option, deprecated since
844 version 0.7.5, has been removed. Please use the equivalent
845 --disable-debugger option instead.
846 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
848 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
849 inoperational for a while now) has been completely removed.
850 * null lexical environments are now printed as #<NULL-LEXENV>,
851 significantly reducing the amount of clutter in typical backtraces.
852 * documentation on weak pointers, finalization, and after GC hooks
853 has been added to the manual.
854 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
855 as other specialized array types, is much faster. SUBSEQ and
856 COPY-SEQ on such arrays have also been sped up.
857 * optimization: EQL is now more efficient when at least other argument
858 is known to be of type (OR FIXNUM (NOT NUMBER)).
859 * fixed bug: the runtime is now less vulnerable to changes in the
860 size of the SBCL object on OS X, and virtual memory is reserved for
861 all spaces that need to be at a fixed address.
862 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
863 * fixed bug: finalizers and after GC hooks that cause consing are now
865 * fixed bug: compiler error messages and summaries are now printed to
866 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
867 * fixed inference of the upper bound of an iteration variable.
868 (reported by Rajat Datta).
869 * fixed bug 211e: calling local functions with duplicated constant
870 keyword argument no longer causes a bogus style warning about an
872 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
873 ftype declaration to be disregarded. (reported by Dave Roberts)
874 * fixed bug 373: caused by erronous compilation of references to alien
875 variables in the runtime on ppc/darwin.
876 * fixed bug 376: CONJUGATE type deriver.
877 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
878 (reported by Baughn on #lisp)
879 * a cross-compiler bug on non-x86 platforms has been identified and
880 fixed. (thanks to Bruno Haible)
881 * improvements to the MIPS runtime code for increased stability.
882 (thanks to Thiemo Seufer)
883 * increased the maximimum compact environment size to allow
884 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
885 * partial workaround for bug 135: don't name the function we're
886 calling for hairy cases of EVAL, so as not to accumulate one environment
887 entry per call to EVAL. (reported by Kevin Reid)
888 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
889 * optimization: type testing for non-vector arrays should be faster.
890 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
892 * bug fix: setting 31st element of a bit vector to zero did not work
894 * bug fix: redefining a class definition which failed due to a
895 previous accessor / function clash now works (but see BUGS entry
896 #380 for more problems in this area). (thanks to Zach Beane)
897 * the long form of DEFINE-METHOD-COMBINATION disables method group
898 checking when given a single method group with pattern *. (thanks
900 * on x86 compiler supports stack allocation of results of simple
901 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
902 * support for building and running on Mac OS X 10.4 "Tiger" has been added
903 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
904 * fixed some bugs related to Unicode integration:
905 ** the restarts for recovering from input and output encoding
906 errors only appear when there is in fact such an error to
908 * fixed some bugs revealed by Paul Dietz' test suite:
909 ** MISC.549 and similar: late transformation of unsafe type
910 assertions into derived types caused unexpected code
912 ** SCALE-FLOAT type deriver is less wrong.
913 ** type derivers for EXP, LOG and similar functions compute result
914 types for complex arguments better.
915 ** (MISC.563) CONJUGATE type deriver works for very restricted
917 ** out-of-line type testers for character strings are available.
918 ** EQUAL compiler transform understands specialness of objects
920 ** accessing double-floats stored on the stack now works on x86-64.
921 ** debugger internals could sometimes create invalid lispobjs,
922 resulting in GC crashes.
923 ** MISC.548: type check weakening can convert required type into
925 ** initialization forms for bindings are not in scope of free special
928 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
929 * incompatible change: thread support for non-NPTL systems has
930 been removed - locking is buggy and unreliable. A threaded
931 SBCL build will now warn at startup and refuse to create
932 new threads, unless futex support is detected in the kernel
933 * incompatible change: the top level REPL now has only an ABORT
934 restart associated with it, not TOPLEVEL and ABORT as it used to.
935 TOP and TOPLEVEL are now available as debugger commands for
936 returning to the top level.
937 * incompatible change: forms evaluated in the REPL now use the
938 global optimization policy.
939 * incompatible change: user- and system-initialization files are
940 no longer processed with LOAD, but by READ and EVAL; hence the
941 global optimization policy, startup package, readtable, etc,
943 * The .fasl file format number has been incremented because of
944 various incompatible changes.
945 * internal entry point details and argument counts no longer appear
946 in backtraces unless explicitly requested by setting
947 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
948 * built-in and standard functions no longer have names like "top
949 level local call to FOO".
950 * fixed bug 32: functions defined in non-null lexical environments
951 now have more legible printed representation
952 * fixed bug 33: functions defined in non-null lexical environemnts
953 are now more amenable to inspection by INSPECT.
954 * workaround for bug 354: XEPs no longer appear in backtraces unless
955 explicitly requested.
956 * fixed bug: receiving the signal which results from attempting to
957 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
958 SIGBUS on other *BSDs) on architectures where the C stack is also
959 the Lisp stack (x86 and x86-64) from foreign code no longer leads
960 to debugger confusion or wild instability. (reported by Cheuksan
962 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
963 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
964 (reported by Lutz Euler)
965 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
966 compiled code. (reported by Paul Dietz)
967 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
968 specialized parameter to a method to something that is not TYPEP
969 the specializer is now possible.
970 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
971 face of package deletion.
972 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
973 pathname. (reported by David Tolpin on c.l.l)
974 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
975 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
976 STRING-STREAM and FILE-STREAM, respectively.
977 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
978 than previously; in addition, its test suite is now run on build.
979 (thanks to Luke Gorrie)
980 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
981 * optimization: added a immediate representation for single-floats
983 * fixed some bugs revealed by Paul Dietz' test suite:
984 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
985 ** attempting to create a package with a colliding nickname causes
986 correctable errors to be signalled.
987 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
988 ** several x86-64 backend bugs related to sign-extension of immediate
991 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
992 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
993 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
995 * fixed loading of multiply forward-referenced layouts.
996 (thanks to Cheuksan Wang)
997 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
998 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
999 (see COMPILE-FILE documentation for details of :PRINT option)
1000 * fixed bugs 19 and 317: fixed-format floating point printing is
1001 more accurate. This also fixes a bug reported by Adam Warner
1002 related to the ~@F format directive.
1003 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1005 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1006 dispatch table if the from-char is a dispatch macro character.
1007 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1008 by the wordsize no longer ignore the last word. (reported by Lutz
1010 * fixed bug in type checking of dynamic-extent variables. (reported
1012 * optimization: sequence traversal functions use their freedom to
1013 coerce function designators to functions.
1014 * optimization: code with many calls to CLOS methods specialized on
1015 CLOS classes has had redundant type checks removed.
1016 * optimization: added declarations to speed up operations that access
1017 the internal character database (for example STRING-UPCASE)
1018 * optimization: comparison operations between floats and sufficiently small
1019 fixnums no longer create extra rationals
1020 * fixed some bugs related to Unicode integration:
1021 ** portions of multibyte characters at the end of buffers for
1022 character-based file input are correctly transferred to the
1023 start of the buffer at the next read.
1024 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1025 passing it through to OPEN.
1026 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1027 argument given to internal calls to OPEN.
1028 ** the built-in comment readers (introduced by character sequences
1029 ";" and "#|") are more forgiving to encoding errors; they will
1030 STYLE-WARN and then attempt to resync the stream at a character
1031 boundary. (thanks to Teemu Kalvas)
1032 * fixed some bugs in the x86-64 port:
1033 ** Negative short int return values from c-calls are sign-extended
1035 ** The stack is aligned to 16-bytes for c-calls, as required by
1036 the ABI. (thanks to Cheuksan Wang)
1037 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1038 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1039 for enums. (thanks to Vincent Arkesteijn)
1040 ** Multiple small optimizations and bugfixes for floating point
1042 * fixed some bugs revealed by Paul Dietz' test suite:
1043 ** Space, Tab, Linefeed, Return and Page have the invalid
1044 secondary constituent character trait.
1045 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1047 ** WITH-INPUT-FROM-STRING should only update the index place on
1049 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1050 works more reliably.
1051 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1052 with one-dimensional arrays with fill pointers.
1053 ** TYPE-OF failed on a complex with an integer realpart and a
1055 ** compiler failure during type inference for the code of form
1056 (IF (EQL X (THE ...)) ...) (MISC.535).
1058 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1059 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1060 platform. The port supports 61-bit fixnums, large memory spaces
1061 and reloading shared object files.
1062 * enhancement: saving cores with foreign code loaded is now
1063 supported on ppc/Darwin in addition to the previously supported
1065 * enhancement: the statistical profiler now walks deeper into the
1066 call stack for more meaningful call-graphs and accrued time
1067 reports (x86/x86-64 only). It also now reports time spent in
1069 * enhancement: it is now possible to trace most individual methods
1070 of a generic function in addition to tracing the generic function
1072 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1074 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1075 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1076 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1077 produces an error. (thanks to Vincent Arkesteijn)
1078 * bug fix: NAMESTRING on pathnames with :WILD components in their
1079 directories works correctly. (thanks to Artem V. Andreev)
1080 * fixed bug 125: compiler preserves identity of closures. (reported
1082 * bug fixed: functions with &REST arguments sometimes failed with
1083 "Undefined function" when compiled with (DEBUG 3). (reported by
1085 * bug fix: overflow during compiling of setting element of a bit
1086 vector with constant index and value. (reported by Timmy Douglas)
1087 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1088 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1089 * on x86 compiler partially supports stack allocation of dynamic-extent
1091 * GO and RETURN-FROM do not check the extent of their exit points
1092 when compiled with SAFETY 0.
1093 * fixed some bugs related to Unicode integration:
1094 ** encoding and decoding errors are now much more robustly
1095 handled; it should now be possible to recover even from invalid
1096 input or output to the terminal. (thanks to Teemu Kalvas)
1097 ** provided a first cut at implementing STRING-TO-OCTETS and
1098 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1099 ** altered the SB-MD5 contributed module slightly, changing the
1100 interface just enough for it to be supportable for builds where
1101 lisp characters are not eight bits.
1102 * fixed some bugs revealed by Paul Dietz' test suite:
1103 ** the FORMATTER-generated functions for ~V[ conditionals require
1104 the correct number of arguments.
1105 ** READ-FROM-STRING returns the mandated second value when applied
1106 to displaced strings.
1107 ** the #\Rubout and #\Backspace characters are treated as invalid
1108 constituent characters by the tokenizer.
1110 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1111 * new feature: reloading changed shared object files with
1112 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1113 * new feature: references to foreign variables and functions
1114 can now be compiled and loaded before the corresponding shared
1115 object file is loaded, as long as the foreign definitions are
1116 available at runtime.
1117 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1118 supported. (thanks to Dan Debertin)
1119 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1120 just keywords. (thanks to Vincent Arkesteijn)
1121 * fixed bug #331: structure-class instances corresponding to
1122 DEFSTRUCT forms are now created eagerly.
1123 * fixed bug #345: backtraces from calls to undefined functions work
1124 on x86 as well. Related bug #61 is now also partially fixed on x86
1125 (backtraces from throws to unknown catch tags.)
1126 * bug fix: lambda-list parsing is now stricter vrt. order and number
1127 of lambda-list keywords.
1128 * bug fix: as specified by AMOP, an error is signalled if a
1129 class-option appears multiple times in a DEFCLASS form. (reported
1131 * bug fix: RANDOM can be compiled when the compiler derives the type
1132 of its numeric argument as a disjoint set of small integers.
1133 (reported by Paul Dietz)
1134 * bug fix: starting a core saved with shared objects loaded when
1135 those objects are not available no longer causes threaded SBCL to
1136 hang. (reported by Sean Ross)
1137 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1139 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1140 argument, which must always be NIL. (reported by Kalle Niemitalo)
1141 * bug fix: printing 1.0d+23 no longer results in an error.
1142 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1143 * bug fix: structure slot setters preserve evaluation order. (thanks
1145 * fixed some bugs related to Unicode integration:
1146 ** RUN-PROGRAM can allow its child to take input from a Lisp
1147 stream. (reported by Stefan Scholl)
1148 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1149 locales. (reported by Ken Causey)
1150 * fixed some bugs revealed by Paul Dietz' test suite:
1151 ** INCF, DECF and REMF evaluate their place form as specified in
1153 ** FORMATTER expands ~{ ~} iteration directives with V or #
1154 parameters correctly.
1155 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1156 consequent uses no arguments correctly.
1157 ** the system has a partial understanding of the (COMPLEX RATIO)
1159 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1162 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1163 * new feature: a build-time option (controlled by the :SB-UNICODE
1164 keyword feature, enabled by default) for building the system with
1165 support for the entire 21-bit character space defined by the
1167 * new feature: the system now has rudimentary external-format
1168 support; the primary user-visible change at this time is that
1169 characters with the high bit set (such as lower-case-e-acute) will
1170 print correctly to a terminal in a UTF-8 environment.
1171 * minor incompatible change: BASE-CHAR no longer names a class;
1172 however, CHARACTER continues to do so, as required by ANSI.
1173 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1174 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1175 has been moved to the SB-EXT package (temporarily re-exported from
1177 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1178 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1180 * fixed bug #351: better error-handlind and reporting for malformed
1182 * fixed bug #350: bignum-printing is now more memory-efficient,
1183 allowing printing of very large bignums, eg. (expt 2 10000000).
1184 (reported by Bruno Haible)
1185 * fixed bug #302: better primitive-type selection for intersection
1187 * fixed bug #308: non-graphic characters now all have names, as
1188 required. (reported by Bruno Haible)
1189 * bug fix: redefining a class with different superclasses now correctly
1190 removes it from the direct-subclasses of its previous superclasses.
1191 (reported by David Morse)
1192 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1193 new value now works. (reported by Bruno Haible)
1194 * bug fix: correct canonicalization of multiple non-standard slot
1195 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1196 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1197 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1198 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1199 now exists, an signals an error.
1200 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1201 during macroexpansion for non-iteration variables as well. (reported
1202 by Bruno Haible for CMUCL)
1203 * bug fix: Cyclic structures and unprintable objects in compiler
1204 messages no longer cause errors. (reported by Bruno Haible)
1205 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1206 and READ-FROM-STRING all now return a primary value of NIL if
1207 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1208 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1209 (reported by Bruno Haible for CMUCL)
1210 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1211 do not designate a condition. (reported by Bruno Haible for
1213 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1214 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1215 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1216 permitting resolution in favour of any of the conflicting symbols.
1217 (reported by Bruno Haible for CMUCL)
1218 * bug fix: EQUAL compiler optimizations is less aggressive on
1219 strings which can potentially compare true despite having distinct
1220 specialized array element types.
1221 * bug fix: unit enumerations can be defined without dividing by
1222 zero. (thanks to Vincent Arkesteijn)
1223 * FORMAT compile-time argument count checking has been enhanced.
1224 (report from Bruno Haible for CMUCL)
1225 * a partial workaround for the bug 262: the compiler does not try to
1226 inline-expand a local function doing RETURN-FROM from a deleted
1227 BLOCK. (thanks to Peter Denno for the bug report and to David
1228 Wragg for the simple test case)
1229 * fixed some bugs revealed by Paul Dietz' test suite:
1230 ** RENAME-PACKAGE allows all package designators as new package
1232 ** constraint propagation and lambda variable substitution are
1233 more cautious in dealing with partially deleted code.
1234 ** compiler sometimes lost reoptimization passes.
1235 ** CERROR, given a condition as condition designator, passes its
1236 remaining arguments to the continue format control without
1238 ** Case-altering FORMAT directives work correctly on non-ASCII
1240 ** The REAL type specifier handles bounds outside the floating
1241 point ranges without signalling FLOATING-POINT-OVERFLOW.
1242 ** Functions with IR1-transformations can create intercomponent
1243 references to global functions.
1244 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1246 ** FORMAT ~R treats a nil value for its first parameter correctly.
1248 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1249 * enhancement: saving cores with foreign code loaded is now
1250 supported on x86/NetBSD and sparc/Linux in addition to the previously
1251 supported platforms.
1252 * bug fix: on some platforms repeated installations caused multiple
1253 copies of HTML documentation to be installed -- should not happen
1254 any more. (reported by Stefan Scholl)
1255 * bug fix: parsing self-recursive alien record types multiple times
1256 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1257 original patch by Helmut Eller for CMUCL)
1258 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1259 (thanks to Richard Kreuter)
1260 * bug fix: defining classes whose accessors are methods on existing
1261 generic functions in other (locked) packages no longer signals
1262 bogus package lock violations. (reported by François-René Rideau)
1263 * bug fix: special variables as DEFMETHOD parameters no longer have
1264 associated bogus type declarations. (reported by David Wragg and
1266 * bug fix: read-write consistency on streams of element-type
1267 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1268 * bug fix: redefiniton of the only method of a generic function with
1269 no DEFGENERIC no longer emits a full WARNING. In addition,
1270 redefinition of generic functions with no DEFGENERIC to an
1271 incompatible lambda list now signals an error. (thanks to Zach
1273 * bug fix: DEFGENERIC now works even when there's a function of the
1274 same name in an enclosing lexical environment. (thanks to Zach
1276 * fixed compiler failure, caused by instrumenting code during
1277 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1278 * optimization: added loop analysis and improved register allocation
1279 * fixed some bugs revealed by Paul Dietz' test suite:
1280 ** POSITION on displaced vectors with non-zero displacement
1281 returns the right answer.
1282 ** (SIMPLE-STRING) is a valid type specifier for sequence
1284 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1286 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1287 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1289 ** The justification version of the FORMAT ~< directive treats
1290 non-zero minpad parameter correctly.
1292 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1293 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1294 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1295 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1296 the supported interface.
1297 * new feature: Single-stepping of code compiled with DEBUG 2 or
1298 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1299 * new feature: saving cores with foreign code loaded is now
1300 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1301 Timothy Moore's work for CMUCL)
1302 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1303 parameters to * instead of NIL if no initform is supplied.
1304 (reported by Johan Bockgaard)
1305 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1306 correctly when FUNCALL appears as the car of the form. Note:
1307 despite this FUNCALL forms are not currently subject to
1308 compiler-macro expansion. (port of Raymond Toy's fix for the
1309 same from CMUCL, reported by Johan Bockgaard)
1310 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1311 (thanks for Teemu Kalvas)
1312 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1313 case instead of terminating the iteration (thanks for Julian
1314 Squires, Sean Champ and Raymond Toy)
1315 * bug fix: incorrect expansion of defgeneric that caused a style
1316 warning. (thanks for Zach Beane)
1317 * bug fix: slot accessor effective method computation works properly
1318 for classes with multiple non-standard applicable methods on
1319 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1320 * on x86 compiler supports stack allocation of results of LIST and
1321 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1322 CMUCL implementation by Gerd Moellmann)
1323 * fixed some bugs revealed by Paul Dietz' test suite:
1324 ** FORMAT strings with both the ~<~:;~> form of the justification
1325 directive and pretty-printing directives cause an error.
1327 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1328 * incompatible change: the internal functions
1329 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1330 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1331 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1332 instead of the old functions.
1333 * new feature: on platforms where "dladdr" is available foreign
1334 function names now appear in backtraces. (based on Helmut Eller's
1336 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1337 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1339 * documentation: ASDF manual is now distributed with SBCL.
1340 * bug fix: SBCL can now load its contributed modules with REQUIRE
1341 even if the system-provided entries have been removed from
1342 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1344 * bug fix: backtraces involving undefined functions or assembly
1345 routines are more informative. (thanks to Brian Downing)
1346 * bug fix: mutually referent alien structures now work correctly.
1347 (reported by Rick Taube)
1348 * bug fix: structures defined by WITH-ALIEN can be referred to
1349 within other definitions in the same WITH-ALIEN.
1350 * bug fix: division operators (MOD, TRUNCATE and the like) with
1351 constant zero divisors and integer dividends no longer generate
1353 * bug fix: provide default methods for INPUT-STREAM-P and
1354 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1355 * bug fix: improve the detection and resolution of MOP metacycles
1356 (where effective method computation depends on the generic function
1357 having its effective method computed). (reported by Bruno Haible)
1358 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1359 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1360 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1361 * bug fix: pathnames with relative directory components can be
1362 represented relative to default pathnames.
1363 * optimization: in taking the GCD of bignums, reduce the two bignums
1364 to approximately the same size (using Euclid's algorithm) before
1365 applying the more sophisticated binary GCD. (thanks to Juho
1367 * optimization: COUNT on bitvectors now operates word-at-a-time.
1368 * optimization: ASH with a positive, but not necessarily constant,
1369 (leftwards) shift, when in a modular context, is compiled to a
1371 * fixed some bugs revealed by Paul Dietz' test suite:
1372 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1373 if the corresponding argument is NIL.
1374 ** FORMAT directives accepting a minpad parameter treat negative
1376 ** literal commas as character parameters to FORMAT directives are
1378 ** literal spaces directly after ~<Newline> directives within a
1379 format-logical-block (~:< ~@:>) do not induce :FILL-style
1380 conditional newlines.
1381 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1382 as specified: it no longer includes conditional newlines).
1383 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1385 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1386 :PER-LINE-PREFIX argument does not evaluate to a string.
1388 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1389 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1390 the manual for details; for now, package locks can be disabled by
1391 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1392 if no major problems are found then it is likely that they will be
1393 compiled in unconditionally.
1394 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1395 unsupported operators on all platforms. To load a shared library
1396 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1397 object file, link it into a shared library outside of SBCL and
1398 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1400 * bug workaround: ROOM T and the GC were not working together
1401 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1402 reporting facility aren't preserved. That reporting has been
1403 disabled, so now until and unless someone figures out how to make it
1404 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1405 * minor incompatible change: as threatened around sbcl-0.8.0, the
1406 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1407 The symbols are also exported from SB-PCL for backwards
1408 compatibility, but more so than before SB-PCL should be treated as
1409 an implementation-internal package.
1410 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1412 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1413 It is known as of this release not to work on the Alpha, however.
1414 * fixed bug #167: errors signalled due to illegal syntax in method
1415 bodies are now more legible.
1416 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1417 designators and can hence be used with TYPEP.
1418 * fixed bug #333: CHECK-TYPE now ensures that the type error
1419 signalled, if any, has the right object to be accessed by
1420 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1421 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1422 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1424 * fixed bug #334: programmatic addition of slots using specialized
1425 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1426 and :ALLOCATION :CLASS slots.
1427 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1428 float exponents. (rereported by Peter Seibel)
1429 * fixed a bug: #\Space (and other whitespace characters) are no
1430 longer considered to be macro characters in standard syntax by
1431 GET-MACRO-CHARACTER.
1432 * fixed bug: initialization of condition class metaobjects no longer
1433 causes an instance of the condition to be created. (reported by
1435 * fixed bug: it is now possible to have slots such that
1436 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1437 description is neither :INSTANCE nor :CLASS.
1438 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1439 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1440 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1441 non-standard SLOT-DEFINITION-ALLOCATION values.
1442 * fixed bugs: various race conditions handling exiting threads.
1443 CL-PPCRE's thread torture test now passes.
1444 * fixed arguably-a-bug: GC time proportional to number of threads in
1445 system even when most of them are idle
1446 * optimization: improved performance of BIT and SBIT on bit-vectors.
1447 * .fasl file incompatibility: The fasl file version number has
1448 been incremented because of changes associated with package locks.
1450 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1451 * minor incompatible change: the system no longer provides
1452 optimization hints (conditions which are TYPEP
1453 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1454 mode; these hints are still emitted when higher SPEED optimization
1456 * new contrib module: a sampling profiler (profiling by statistical
1457 sampling, rather than by instrumenting functions) is available as
1458 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1459 * the behaviour of the standard function ED is now customizeable by
1460 third parties through a hook variable: see ED's documentation
1461 string for information on the protocol.
1462 * the compiler no longer emits efficiency notes for (FUNCALL X)
1463 when the type of X is uncertain under default optimization
1465 * fixed bug 276: mutating a binding of a specialized parameter to a
1466 method to something that is not TYPEP the specializer is now
1468 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1469 exhibit the required behaviour on the x86 platform. (thanks to
1470 Peter van Eynde, Eric Marsden and Bruno Haible)
1471 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1472 even for difficult arguments. (reported by Peter Graves)
1473 * fixed bug 141a: the backquote printer now descends quoted
1475 * fixed another bug in backquote printing: no more destructive
1476 modification of the form's list structure. (reported by Brian
1478 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1479 move between its address being taken and the call to
1480 interrupt_thread, fixing a crashing race condition.
1481 * the SB-POSIX contrib implementation has been adjusted so that it
1482 no longer exhibits ridiculously poor performance when constructing
1483 instances corresponding to C structs.
1485 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1486 * minor incompatible change: the sb-grovel contrib now treats C
1487 structures as alien (in the sense of SB-ALIEN) objects rather than
1488 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1489 has implications for memory management of client code
1490 (sb-grovel-returned objects must now be manually managed) and for
1491 type safety (alien objects now have full types).
1492 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1493 used to control emission of compiler diagnostics, rather than the
1494 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1495 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1496 quality should be considered deprecated.
1497 * install.sh now installs the user manual as well
1498 * (not quite a new documentable feature, but worth considering in
1499 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1500 beginnings of a semantically meaningful condition hierarchy is
1501 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1503 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1504 designator as the defaults argument.
1505 * fixed bug: Displaced arrays whose displaced-to array has become
1506 too small now cause ARRAY-DIMENSION to signal an error, providing
1507 for safer bounds-checking. (reported by Bruno Haible)
1508 * fixed bug: DEFCLASS slot definitions with identical :READER and
1509 :WRITER names now signal a reasonable error. (reported by Thomas
1511 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1513 * fixed bug: Invalid slot specification errors now print correctly.
1514 (thanks to Zach Beane)
1515 * fixed bug 320: Shared to local slot value transfers in class
1516 redefinitions now happen corrently from superclasses as
1517 well. (reported by Bruno Haible)
1518 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1520 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1521 improper lists correctly. (reported by Bruno Haible)
1522 * fixed bug 313: source-transform for <fun-name> was erroneously
1523 applied to a call of a value of a variable with name <fun-name>.
1524 (reported by Antonio Menezes Leitao)
1525 * fixed bug 307: The obsolete instance protocol ensures that
1526 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1527 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1528 remove dead unknown-values globs from the middle of the stack.
1529 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1530 easier to resume long computations after using *BREAK-ON-SIGNALS*
1531 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1532 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1533 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1534 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1535 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1536 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1538 * fixed a (fairly theoretical) bug in string printing: if
1539 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1540 does not have array-element-type equal to the most general string
1542 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1543 function name. (thanks to Nikodemus Siivola)
1544 * fixed bug: compile-time format string checker failed on
1545 non-closed ~{. (reported by Thomas F Burdick)
1546 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1547 don't warn on legal loop code involving more than one aggregate
1548 boolean. (thanks to Nikodemus Siivola)
1549 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1550 in keyword position to MAKE-PACKAGE was not being reported as
1552 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1553 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1555 * fixed bugs 280 and 312: the checking for multiple definitions in a
1556 file is less likely to become confused by uses of inline
1558 * fixed bug: the #S reader macro performs the keyword coercion
1559 specified for slot names. (reported by Kalle Niemitalo)
1560 * fixed bug: lambda lists may contain symbols whose names start with
1561 & but are not lambda-list-keywords; their occurrence triggers a
1563 * fixed bug 321: define-method-combination argument lambda lists do
1564 not cause invalid code to be generated when &optional or &aux
1565 variables have default values. (reported by Bruno Haible)
1566 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1567 have CLOS classes; this appears also to have fixed reported
1568 instability in stack exhaustion detection.
1569 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1570 MUFFLE-WARNING when no associated restart is present is now
1572 * optimization: rearranged the expansion of various defining macros
1573 so that each expands into only one top-level form in a
1574 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1576 * optimization: used a previously little-used slot in symbols to
1577 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1579 * fixed some bugs revealed by Paul Dietz' test suite:
1580 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1582 ** Fixed an optimization bug related to inheritance of initforms
1583 from local to shared slots.
1584 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1585 ** CAST optimizer forgot to flush argument derived type.
1586 ** print/read consistency on floats is now orders of magnitude
1587 more likely. (thanks also to Bruno Haible for a similar report
1589 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1590 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1591 for objects of type REAL. Make it so.
1592 ** SXHASH is specified (infelicitously) to respect similarity,
1593 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1594 it so. (thanks to Markus Ziegler)
1595 ** on the Alpha, the compiler succeeds in compiling functions
1596 returning a known number of arguments greater than 63.
1597 ** fixed handling of invalid NIL arguments in keyword position
1599 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1601 ** provided a readably-printable representation for RANDOM-STATE
1603 ** ensured that pathnames, where they have a namestring, always
1604 print using #P"..." syntax.
1606 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1607 * Support for the forthcoming 2.0 version of the NetBSD kernel
1608 running on x86 hardware has been added. (thanks to Perry
1609 E. Metzger most immediately, and others for their past work)
1610 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1611 current release version; SBCL's OpenBSD support had been broken
1612 since about the time of OpenBSD's switch to ELF binary format.)
1613 * [placeholder for DX summary]
1614 ** user code with &REST lists declared dynamic-extent, under high
1615 speed or space and low safety and debug optimization policy.
1616 * The manual has been converted to Texinfo format and the debugger
1617 chapter from the cmucl manual has been added.
1618 * A facility has been added to extract documentation strings from
1619 sbcl and store them as Texinfo-formatted snippets for inclusion in
1620 the manual (via Texinfo's @include directive)
1621 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1622 the test case to Dave Roberts)
1623 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1624 pointers. (reported by Sean Ross)
1625 * bug fix: PROFILE output is printed nicely even for large numerical
1626 values. (thanks to Zach Beane)
1627 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1628 greater than 32 handle EOF correctly.
1629 * bug fix: on X86 an immediate argument of the IMUL instruction is
1630 correctly printed in disassembly. (thanks to Lutz Euler)
1631 * bug fix: class slots in redefined classes preserve their old
1632 values. (thanks to Bruno Haible and Nikodemus Siivola)
1633 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1634 succeeds. (reported by Marco Baringer)
1635 * fixed some bugs revealed by Paul Dietz' test suite:
1636 ** READ-SEQUENCE now works on ECHO-STREAMs.
1637 ** RATIONALIZE works more according to its specification. (thanks
1640 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1641 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1642 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1643 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1644 should matter to you only if you rebind the printer control
1645 variables and then find you want different bindings in the
1646 debugger than in the ordinary execution of your program.)
1647 * The runtime build system has been tweaked to support building
1648 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1649 assembler and linker. (thanks to Nikodemus Siivola)
1650 * Unbound, undefined, undeclared variables now trigger full
1651 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1652 more often programmer error than deliberate exploitation of undefined
1654 * optimization: the hash algorithm for strings has changed to one
1655 that is less vulnerable to spurious collisions. (thanks to Juho
1657 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1658 needless bounds checking. (thanks to Juho Snellman)
1659 * optimization: implemented multiplication as a modular
1660 (UNSIGNED-BYTE 32) operation on the PPC backend.
1661 * fixed some bugs revealed by Paul Dietz' test suite:
1662 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1664 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1665 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1666 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1668 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1670 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1672 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1673 their output stream on EOF from read.
1674 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1675 have been read to end-of-file.
1676 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1678 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1679 description of determination of which consecutive characters
1681 ** Printing the "Space" character with escaping on now yields "#\\ ",
1682 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1683 ** Reading floating-point numbers with *READ-BASE* set to a number
1684 less than 10 works correctly.
1685 ** Reading floating-point numbers with *READ-BASE* set to a number
1686 more than 10 works correctly.
1687 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1688 the readtable currently in effect.
1690 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1691 * minor incompatible change: parsing of namestrings on a physical
1692 (Unix) host has changed; numbers after the final #\. in a
1693 namestring are no longer interpreted as a version field. This is
1694 intented to be largely invisible to the user, except that the
1695 meaning of the namestring "*.*.*" has changed: it now refers to a
1696 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1697 should usually be replaced by
1698 (make-pathname :name :wild :type :wild :version :wild)
1699 with the added benefit that this is more likely to be portable.
1700 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1701 option to OPEN now signals an error if the file being opened
1702 exists; this may have an impact on existing code.
1703 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1704 signals received, on the PowerPC platforms (both Linux and
1705 Darwin). (thanks to Pierre Mai for pointing out the location of
1707 * several fixes on OS X: The system now builds and runs cleanly on
1708 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1709 (thanks to Brian Mastenbrook)
1710 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1711 second-resolution: integer multiples of 1/3600 between -24 and 24.
1712 (thanks to Vincent Arkesteijn)
1713 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1714 type when called with 1 argument; PEEK-CHAR checked type of
1715 PEEK-TYPE only after having read first character from a
1716 stream. (reported by Peter Graves)
1717 * bug fix: the garbage collector now has much better locality
1718 behaviour, and in particular no longer treats all memory as being
1719 exhausted when many small objects point to each other in a deeply
1721 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1722 recognized as being TYPEP their class.
1723 * bug fix: the PUSHNEW documentation string has been corrected.
1724 (thanks to Vincent Arkesteijn)
1725 * bug fix: defaulting of the value for the last of an atypically
1726 large number of multiple values being bound was not being
1727 performed correctly on the Alpha or PPC platforms
1728 * optimization: implemented multiplication as a modular
1729 (UNSIGNED-BYTE 32) operation on the x86 backend.
1730 * optimization: SEARCH on simple-base-strings can now be open-coded.
1731 (see also contrib/compiler-extras.lisp for inspiration for
1732 teaching the compiler about the Boyer-Moore algorithm).
1733 * value, returned by MAX (and MIN) called with several EQUALP, but
1734 not EQL, arguments now does not depend on compiler settings.
1735 * fixed some bugs revealed by Paul Dietz' test suite:
1736 ** in stack analysis liveness information is propagated from
1737 non-local entry points.
1738 ** pathwise CAST removing failed when the CAST node did not start
1740 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1741 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1743 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1744 host is already defined.
1745 ** RENAME-FILE works on streams instead of signalling an internal
1747 ** PEEK-CHAR uses the current readtable when determining whether
1748 or not a character is whitespace.
1749 ** MERGE-PATHNAMES handles the case when the pathname does not
1750 specify a name while the default-pathname specifies a version.
1751 ** Pathnames now stand a better chance of respecting print/read
1753 ** Attempting to use standardized file system operators with a
1754 pathname with invalid :DIRECTORY components signals a
1756 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1758 ** DIRECTORY on logical pathnames is more correct.
1759 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1760 signal a TYPE-ERROR if their argument is not a stream.
1761 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1762 designator argument does not designate a stream.
1763 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1764 examining the synonym.
1765 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1767 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1768 element-type larger than ([UN]SIGNED-BYTE 32).
1770 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1771 * When built with the :SB-FUTEX feature, threaded builds now take
1772 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1773 for faster/more reliable mutex and condition variable support.
1774 * Incompatible change (but one you probably shouldn't have been using
1775 anyway): the interface and code for arbitrating between multiple
1776 threads in the same user session has been redesigned.
1777 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1778 error when called without an explicit environment argument.
1779 (thanks to Nikodemus Siivola)
1780 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1781 confused about their position. (thanks to Adam Warner and Gerd
1783 * bug fix: Pretty printing backquoted forms with unquotations in the
1784 argument list position of various code constructs such as LAMBDA
1785 now works correctly. (reported by Paul Dietz)
1786 * bug fix: Pretty printing unquotations no longer loses all
1787 stream position information.
1788 * optimization: performance of string output streams is now less
1789 poor for multiple small sequence writes.
1790 * optimization: performance of CSUBTYPEP in the presence of complex
1791 expressions involving CONS and NOT many times has been improved.
1792 (reported by Paul Dietz)
1793 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1795 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1796 streams for the REPL, for the debugger, and so on. (thanks to
1798 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1799 * fixed some bugs revealed by Paul Dietz' test suite:
1800 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1801 is ignored if it is too small, rather than propagating through
1802 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1803 ** extremely complex negations of CONS types were not being
1804 sufficiently canonicalized, leading to inconsistencies in
1806 ** VALUES tranformer lost derived type.
1808 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1809 * fixed a bootstrapping bug: the build process no longer assumes
1810 that the various BOOLE-related constants have the same value in
1811 host and target lisps. (noted by Paul Dietz' test suite on an
1812 SBCL binary built from CLISP)
1813 * The system can now be dynamically linked on the MIPS platform,
1814 which enables dynamic loading of foreign code from Lisp. (thanks
1815 to Ralf Baechle for discussions on the MIPS ABI)
1816 * The system now records debugging information for its own source
1817 files in a filesystem-position-independent manner, relative to
1818 the "SYS" logical host.
1819 * fixed a compiler bug: MV-LET convertion did not check references
1820 to the "max args" entry point. (reported by Brian Downing)
1821 * tweaked disassembly notes to be less confident about proclaiming
1822 some instruction as an LRA. (thanks to Brian Downing)
1823 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1824 now each have their own history, command character, and other
1825 characteristics. (thanks to David Lichteblau)
1826 * fixed some bugs revealed by Paul Dietz' test suite:
1827 ** compiler failure in compiling LOGAND expressions including a
1829 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1830 shift greater than 32.
1831 ** FLUSH-DEST did not mark blocks for type check regeneration.
1832 ** HANDLER-CASE failed to accept declarations in handler clauses
1833 in some circumstances.
1835 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1836 * New code in contrib/sb-introspect (still probably not entirely
1837 stable yet) provides some support for smart Lisp development
1838 environments like SLIME.
1839 * The conditions signalled for errors occurring when loading .fasl
1840 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1841 in a way which should help ASDF recover gracefully.
1842 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1843 stuff has been cleaned up. If you code contrib/ stuff, this might
1844 affect you, and you can look at contrib/README, contrib/STANDARDS,
1845 and/or the 0.8.4.27 diff to check.
1846 * In full calls the compiler now does not generate checks for declared
1847 argument types for all arguments.
1848 * various threading fixes
1849 ** and some experimental patches which didn't make it into
1850 the main tree for this release, but which are shipped in
1851 contrib/experimental-thread.patch as a possible fix for some
1852 failures (deadlock, spinning...) in GC-intensive multithreaded
1854 * fixed PPC build problem (source code incompatibility of different
1855 library versions): added offsetof() hackery which attempts to divine
1856 where glibc maintainers put uc_mcontext today
1857 * fixed bug 282: compiler does not trust type assertions while passing
1858 arguments to a full call.
1859 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1860 &OPTIONAL VALUES type parameter.
1861 * fix bug 214: algorithm for noting rejected templates is now more
1862 similar to that of template seletion. (also reported by rydis on
1864 * fixed bug 141b: printing backquoted information readably and prettily
1865 inserts a space where necessary.
1866 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1867 (CHARACTER 10) are now reported as errors, rather than propagated
1868 as unknown types. (reported by piso on #lisp)
1869 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1870 with values NIL and :ERROR. (thanks to Milan Zamazal)
1871 * fixed bug 191c: CLOS now does proper keyword argument checking as
1872 described in CLHS 7.6.5 and 7.6.5.1.
1873 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1874 longer raise an error; further, using a list as a for-as-arithmetic
1875 counter now raises a meaningful error.
1876 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1877 now understood by sequence creation functions such as MAKE-SEQUENCE
1879 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1880 a STRING-INPUT-STREAM.
1881 * compiler enhancement: SIGNUM is now better able to derive the type
1883 * type declarations inside WITH-SLOTS are checked. (reported by
1885 * fixed some bugs revealed by Paul Dietz' test suite:
1886 ** incorrect optimization of TRUNCATE for a positive first
1887 argument and negative second.
1888 ** compiler failure in let-convertion during flushing dead code.
1889 ** compiler failure while deriving type of TRUNCATE on an
1890 interval, containing 0.
1891 ** ASH of a negative bignum by a negative bignum count now returns
1893 ** intersection of CONS types now canonicalizes properly, fixing
1894 inconsistencies in SUBTYPEP.
1896 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1897 * incompatible change: The --disable-debugger command line
1898 option now clobbers the debugger at a more fundamental
1899 level, by redefining #'INVOKE-DEBUGGER instead of by
1900 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1901 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1902 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1903 enter the debugger REPL and then suffer recursive errors
1904 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1905 while under the new system, BREAK in --disable-debugger mode
1906 terminates the system just as an unhandled error would.
1907 * fixed compiler performance when processing loops with a step >1;
1908 * bug fix: DOCUMENTATION now retrieves generic function
1909 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1910 support has been systematized, and now supports the methods
1911 specified by ANSI, along with a default method and a method for
1912 slot documentation. (reported by Nathan Froyd)
1913 * bug fix: effective methods associated with a generic function are
1914 no longer cached over a change of that generic function's method
1915 combination. (reported by Andreas Fuchs)
1916 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1917 names a non-existent directory. (thanks to Andreas Fuchs)
1918 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1919 the PPC platform, lead to stack corruption; this has been fixed.
1920 (reported by Rainer Joswig)
1921 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1922 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1923 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1924 arithmetic in general, is now much more reliable on the PPC
1926 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1927 platform now returns the right answer.
1928 * optimization: restored some effective method precomputation in
1929 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1930 precomputation is now tunable.
1931 * optimization: compiler-internal data structure use has been
1932 reviewed, and changes have been made that should improve the
1933 performance of the compiler by about 20%.
1934 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1935 simple-base-strings and simple-bit-vectors is improved.
1936 * optimization: the optimization of 32-bit logical and arithmetic
1937 functions introduced in version 0.8.3 on the x86 has been
1938 implemented on the mips, ppc and sparc platforms; an
1939 implementation of the same facility, but for 64-bit arithmetic,
1940 has been added for the alpha.
1941 * microoptimization: the compiler is better able to make use of the
1942 x86 LEA instruction for multiplication by constants.
1943 * bug fix: in some situations compiler did not report usage of
1944 generic arithmetic in (SPEED 3) policy.
1945 * bug 145b fix: compiler used wrong type specifier while converting
1946 MEMBER-types to numeric.
1947 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1949 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1950 index argument now works properly on the Alpha platform.
1951 * bug fix: floating point exception treatment on the Alpha platform
1953 * bug fix: FILE-POSITION works much better on string input and
1954 output streams. (thanks to Nikodemus Siivola)
1955 * bug fix: many threading/garbage collection symptoms sorted.
1956 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1957 might be pseudo-atomic.
1958 * internal change: Stopping for GC is now done with signals not
1959 ptrace. GC is now done in whichever thread wanted it, instead of
1961 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1963 * bug fix: result form in DO is not contained in the implicit
1965 * incompatible change: ICR structure is changed; the value part of
1966 CONTINUATION is now called LVAR; corresponding functions are
1967 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1968 * added type deriver for ISQRT (thanks to Robert E. Brown).
1969 * bug fix: better support for loading from the command line when an
1970 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1972 * fixed some bugs revealed by Paul Dietz' test suite:
1973 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1974 ** ROUND and FROUND now give the right answer when given very
1975 small float arguments.
1976 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1978 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1979 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1980 ** LCM should return a non-negative integer.
1981 ** PARSE-INTEGER returned the index of a terminator instead of the
1982 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1983 ** PARSE-INTEGER returned an incorrect index being applied to a
1985 ** LCM with two arguments of 0 returns 0 rather than signalling
1987 ** unsigned addition of a 32-bit constant with the high bit set no
1988 longer causes an internal compiler error on the x86.
1989 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1990 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1991 with negative last argument.
1992 ** byte specifiers with zero size and position no longer cause
1993 an error during type derivation.
1994 ** bignum multiplication on the Alpha platform now returns the
1996 * porting: The system now builds on SuSE AMD64, although it still
1997 generates a 32-bit binary.
1998 * .fasl file incompatibility: The fasl file version number has
1999 been incremented (because of the changes to internal compiler
2000 data structures referred to above).
2002 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2003 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2004 more accurately, on the Darwin kernel running on PowerPC hardware.
2005 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2006 * Compiler code deletion notes now signal a condition of type
2007 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2008 an associated MUFFLE-WARNING restart.
2009 * The compiler now performs limited argument count validation of
2010 constant format strings in FORMAT, and where appropriate in ERROR,
2011 CERROR and WARN. (thanks to Gerd Moellmann)
2012 * New ASDF-INSTALL contrib can be used for automatic download and
2013 installation of third-party Lisp code from CCLAN or other sites
2015 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2016 which forces another thread to execute a function supplied by the
2018 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2019 now accept and act upon their :ELEMENT-TYPE keyword argument.
2020 (reported by Martin Atzmueller, Edi Weitz)
2021 * bug fix: FILE-POSITION now accepts position designators up to
2022 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2023 is the greater. (thanks to Patrik Nordebo)
2024 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2025 Kalle Olavi Niemitalo)
2026 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2027 printable. (reported by Eric Marsden)
2028 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2029 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2030 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2031 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2032 (thanks to Stig Sandoe)
2033 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2034 UNREAD-CHAR now correctly includes the unread character in the
2035 target sequence. (thanks to Gerd Moellmann)
2036 * bug fix in threaded builds: the system can now be suspended and
2037 resumed by shell job control with minimal disruption.
2038 * bug fixes in times and timezones >2038AD
2039 * better handling of "where is GNU make?" problem in build scripts
2040 (thanks to Nikodemus Siivola)
2041 * new optimization: inside a named function any reference to a
2042 function with the same name is considered to be a self-reference;
2043 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2044 optimization quality.
2045 * new optimization on x86: logical functions and + now have
2046 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2047 used when the result is truncated to 32 bits.
2048 * VALUES declaration is partially enabled.
2049 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2050 * fixed some bugs revealed by Paul Dietz' test suite:
2051 ** The system now obeys the constraint imposed by
2052 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2053 types form a lattice under type intersection.
2054 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2055 ** ASSOC now ignores NIL elements in an alist.
2056 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2057 and (1+ MOST-POSITIVE-FIXNUM) answers.
2058 ** The addition of a method with invalid qualifiers to a generic
2059 function does not cause an error to be signalled immediately;
2060 a warning is signalled, and the error is generated only on
2061 calling the generic function.
2062 * changed .fasl file version number, in part to add type codes for
2063 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2064 obscure ANSI requirements
2066 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2067 * fixed bug 148: failure to inline-expand a local function left
2068 garbage, confusing the compiler.
2069 * fixed bugs 3cd: structure slot readers perform type check if the
2070 slot can have an invalid value (i.e. it is either not initialized
2071 or can be written with a less specific slot writer).
2072 * bug fix: the compiler now traps array references to elements off
2073 the end of an array; previously, the bounds checking in some
2074 circumstances could go off-by-one.
2075 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2077 * type declarations for array element types now obey the description
2078 on the CLHS page "Declaration TYPE", as per discussions on
2079 sbcl-help around 2003-05-08. This means that a declaration
2080 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2081 declaration, all references to BAR will be asserted or assumed
2082 (with THE, so dependent on compiler policy) to involve objects of
2083 type FOO. Note that no such declaration is implied in
2084 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2085 * declared types of functions from the "Conditions"
2086 chapter. (reported by Paul Dietz)
2087 * bug fix: CERROR accepts a function as its first argument.
2088 * bug fix: NTH an NTHCDR accept a bignum as index
2089 arguments. (reported by Adam Warner)
2090 * optimization: character compare routines now optimize comparing
2091 against a constant character. (reported by Gilbert Baumann)
2092 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2093 argument now works properly on the MIPS platform.
2094 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2095 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2096 declared to have a functional type, does not cause an error
2098 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2099 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2100 * fixed bug: if last continuation of a deleted block has a
2101 destination, this destination should be deleted too. (reported by
2103 * fixed a bug in the bootstrap process: the host compiler's values
2104 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2105 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2106 test case from Patrik Nordebo)
2107 * improved the ability of the disassembler on the PPC platform to
2108 provide helpful disassembly notes.
2109 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2110 the class in more cases than previously.
2111 * bug fix: FILE-POSITION now understands :START and :END for
2112 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2113 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2114 8) data. (thanks to David Lichteblau)
2115 * bug fix: it is possible to add a method to a generic function
2116 without lambda list.
2117 * bug fix: reader failed to signal END-OF-FILE inside an
2118 object representation. (reported by Nikodemus Siivola)
2119 * fixed some bugs revealed by Paul Dietz' test suite:
2120 ** LAST and [N]BUTLAST should accept a bignum.
2121 ** condition slot accessors are methods.
2122 ** (VECTOR NIL) is a subtype of STRING.
2124 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2125 * minor incompatible change: some nonsensical specialized lambda
2126 lists (used in DEFMETHOD) which were previously ignored now signal
2128 * minor incompatible change: the system is now aware of the types of
2129 variables in the COMMON-LISP package, and will signal errors for
2130 most violations of these type constraints (where previously they
2131 were silently accepted).
2132 * minor incompatible change: COMPILE-FILE now uses the freedom
2133 afforded (ANSI 3.2.2.3) to use derived function types for
2134 functions defined in the same file. This also permits the system
2135 to warn on static type mismatches and function
2136 redefinition. (Currently it does not work with high DEBUG level.)
2137 * minor incompatible change: VALUES declaration is disabled.
2138 * When issuing notes, the compiler now signals a condition of type
2139 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2140 restart for use in user handlers. It is expected that the
2141 COMPILER-NOTE condition will eventually become a condition
2142 supertype to a hierarchy of note types, which will then be
2143 handleable in a similar fashion. However, at the moment, no such
2144 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2145 but it's an implementation detail, not a classification for the
2147 * Changes in type checking closed the following bugs:
2148 ** type checking of unused values (192b, 194d, 203);
2149 ** template selection based on unsafe type assertions (192c, 236);
2150 ** type checking in branches (194bc).
2151 * A short form of VALUES type specifier has ANSI meaning (it has
2152 increased the number of situations when SBCL cannot perform type
2154 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2155 keywords or constants is permissible.
2156 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2157 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2158 classes. (thanks to Antonio Martinez)
2159 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2160 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2161 argument to a pretty-print stream. Instead, it leaves any such
2162 operations to DESCRIBE-OBJECT methods.
2163 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2165 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2166 lambda lists. (thanks to David Lichteblau)
2167 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2168 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2169 (thanks to Gerd Moellmann)
2170 * fixed bug 246: increased compilation speed of long
2171 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2173 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2175 * DEFCONSTANT now throws a condition of type
2176 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2177 constant to a non-EQL value; CONTINUE and ABORT restarts
2178 respectively change and preserve the value.
2179 * fixed bug 63: The code walker, part of the implementation of CLOS,
2180 is now better at handling symbol macros.
2181 * bug fix: There is no longer an internal implementation type named
2182 CL:LENGTH. (reported by Raymond Toy)
2183 * bug fix: In macro-like defining macros/special operators the
2184 implicit block does not enclose the lambda list.
2185 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2186 enclosed in parentheses) are not suitable as type specifiers, and
2187 their use properly signals an error now.
2188 * bug fix: An argument count mismatch for a type specifier in code
2189 being compiled no longer causes an unhandled error at compile
2190 time, but signals a compile-time warning.
2191 * fixed simple vector readable printing
2192 * bug fix: DESCRIBE takes more care over whether the class
2193 precedence list slot of a class is bound before accessing it.
2194 (reported by Markus Krummenacker)
2195 * bug fix: FORMATTER can successfully compile pretty-printer format
2196 strings which use variants of the ~* directive inside.
2197 * bug fix: SEARCH now applies its TEST predicate to the elements of
2198 the arguments in the correct order. (thanks to Wolfhard Buss)
2199 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2200 in type inference. (thanks to Robert E. Brown)
2201 * bug fix: Reading in symbols with an explicit package name of ""
2202 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2203 * fixed some bugs revealed by Paul Dietz' test suite:
2204 ** NIL is now allowed as a structure slot name.
2205 ** Arbitrary numbers, not just REALs, are allowed in certain
2206 circumstances in LOOP for-as-arithmetic clauses.
2207 ** Multiple class redefinitions before slot access no longer
2208 causes a type error.
2209 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2210 association between the name and a class.
2211 ** Generic functions with non-standard method-combination and over
2212 five methods all of which return constants no longer return NIL
2213 after the first few invocations. (thanks to Gerd Moellmann)
2214 ** CALL-NEXT-METHOD with no arguments now passes the original
2215 values of the arguments, even in the presence of assignment.
2216 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2218 ** DISASSEMBLE works with closures and funcallable instances.
2219 ** ADD-METHOD now returns the generic function, not the new method.
2220 ** FIND-METHOD signals an error if the lengths of the specializers
2221 is incompatible with the generic function, even if the ERRORP
2223 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2224 which its argument is a member.
2225 ** DEFCLASS only redefines the class named by its class-name
2226 argument if that name is the proper name of the class;
2227 otherwise, it creates a new class.
2228 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2229 of the UNBOUND-SLOT condition to the name of the slot.
2230 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2231 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2232 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2233 SLOT-MAKUNBOUND in the specified fashion.
2235 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2236 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2237 cross-compilation host. As a consequence, we can now bootstrap our
2238 way up to SBCL starting with a bare gcc toolchain and human-readable
2239 source code (first the source to CLISP, then the source to SBCL).
2240 * A contributed module containing a partial implementation of the
2241 simple-streams interface has been included. (thanks to Rudi
2243 * A contributed module implementing the RFC1321 Message Digest
2244 Algorithm, known as MD5, has been included.
2245 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2246 no longer has any effect, as the code controlled by this feature
2247 has been deleted. (As far as we know, no-one has ever built using
2248 this feature, and its semantics were confused in any case).
2249 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2250 work on conditions (as required by the ANSI specification),
2251 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2252 expected behaviour on conditions. Users should note, however,
2253 that such behaviour is not required by the ANSI specification,
2254 and so use of this behaviour may render their code unportable.
2255 * fixed some bugs revealed by Paul Dietz' test suite:
2256 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2258 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2259 ** SXHASH on condition objects no longer returns NIL.
2260 ** :ALLOCATION :CLASS slots are better treated; their values are
2261 updated on class redefinition, and initforms inherited from
2262 superclasses are applied.
2263 ** REMOVE-METHOD returns its generic function argument even when
2264 no method was removed.
2265 ** SHARED-INITIALIZE now initializes the values of the requested
2266 slots, including those with :ALLOCATION :CLASS.
2267 ** ALLOCATE-INSTANCE now works on structure classes defined via
2268 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2270 ** SLOT-EXISTS-P now works on conditions, as well as structures
2272 ** MAKE-LOAD-FORM now has the required methods on
2273 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2274 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2275 keyword, and now implements the SLOT-NAMES argument.
2276 ** methods with &OPTIONAL arguments no longer allow too many
2277 arguments to be passed in the call without error.
2278 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2279 option is consistent with the required arguments of the generic
2280 function lambda list.
2281 * bug fix: REQUIRE accepts a string designator. (Thanks to
2283 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2284 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2285 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2286 * bug fix: The compiler now checks for duplicated variables in macro
2288 * bug fix: SETQ on globals returns the correct value.
2289 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2290 not a COMPILER-ERROR (followed by some other strange error on
2291 choosing the CONTINUE restart).
2292 * bug fix: make.sh and friends are now more consistent in the way that
2293 they look for GNU "make".
2295 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2296 * experimental native threads support (on x86 Linux >=2.4 only).
2297 This is not compiled in by default: you need to add :SB-THREAD to
2298 the target features. See the "Beyond ANSI" chapter of the manual
2300 * fix for longstanding nonANSIism: The old distinction between
2301 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2302 The return value from CL:FIND-CLASS is now a CLOS class, and
2303 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2304 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2305 * An interface to the MetaObject Protocol, as described in Kiczales,
2306 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2307 MIT Press, 1991, is available from the SB-MOP package.
2308 * incompatible change: the SB-PCL package should now be considered
2309 a private implementation detail, and no longer a semi-private MOP
2311 * minor incompatible change: due to rearrangement for threads, the
2312 control stack and binding stack are now allocated at arbitrary
2313 addresses instead of being hardcoded per-port. Users affected by
2314 this probably have to be doing advanced things with shared
2315 libraries, and will know who they are.
2316 * minor incompatible change: Previously, all --eval forms used were
2317 processed with READ before any of them were processed with EVAL.
2318 Now each --eval form is processed with both READ and EVAL before
2319 the next --eval form is processed. (Thus package operations like
2320 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2321 work as the user might reasonably expect.)
2322 * minor incompatible change: *STANDARD-INPUT* is now only an
2323 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2325 * minor incompatible change: Y-OR-N-P is now character-oriented,
2326 not line oriented. Also, YES-OR-NO-P now works without errors.
2327 (thanks to Antonio Martinez)
2328 * sb-aclrepl module improvements: an integrated inspector, added
2329 repl features, and a bug fix to :trace command.
2330 * Known functions, which cannot be open coded by the backend, are
2331 considered to be able to check types of their arguments. (fixing
2332 a bug report by Nathan J. Froyd)
2333 * fixed a bug in computing method discriminating functions: It is
2334 now possible to define methods specialized on classes which have
2335 forward-referenced superclasses. (thanks to Gerd Moellmann)
2336 * fixed evaluation order in optional entries (reported by Gilbert
2338 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2339 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2340 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2341 required-by-AMOP NAME argument, as well as CLASS and
2342 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2343 * fixed bug 20: DEFMETHOD can define methods using names that are
2344 not the proper names of classes to designate class specializers.
2345 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2346 Unix file descriptors, instead of blowing up. (thanks to Antonio
2348 * Garbage collection refactoring: user-visible change is that a
2349 call to the GC function during WITHOUT-GCING will not do garbage
2350 collection until the end of the WITHOUT-GCING. If you were doing
2351 this you were probably losing anyway.
2352 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2353 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2354 TYPEP the latter but not the former.
2355 * The compiler issues a full WARNING for calls to undefined functions
2356 with names from the CL package.
2357 * MAP-INTO for a vector destination is open coded. (reported by
2358 Brian Downing on c.l.l)
2359 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2360 documentation string.
2361 * fixed some bugs revealed by Paul Dietz' test suite:
2362 ** COPY-ALIST now signals an error if its argument is a dotted
2364 ** Condition slots are now accessed more correctly in the presence
2365 of multiple initargs for a given slot.
2366 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2367 exclude restarts of the same name associated with a different
2369 ** DEFCLASS of forward-referenced classes with another
2370 forward-referenced class in the superclasses list no longer
2372 ** Condition slots are now initialized once each, not multiple
2373 times. (thanks to Gerd Moellmann)
2374 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2376 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2377 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2378 arguments contain duplicated elements.
2379 ** RESTART-CASE understands local macros.
2380 ** RESTART-CASE associates exactly its own restarts with a condition.
2381 ** ENDP in safe mode checks its argument to be of type LIST.
2382 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2383 in question is unbound.
2384 ** Optimized MAKE-INSTANCE functions no longer cause internal
2385 assertion failures in the presence of duplicate initargs.
2386 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2387 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2388 propagated even in the ctor optimized implementation of
2390 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2392 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2393 * a better implementation of SXHASH on (simple) bit vectors,
2394 measured both in execution speed and in distribution of results
2395 over the positive fixnums, has been installed. Likewise, a better
2396 implementation of EQUAL for simple bit vectors is now available.
2397 * fixed CEILING optimization for a divisor of form 2^k.
2398 * fixed bug 240 (emitting extra style warnings "using the lexical
2399 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2400 by Antonio Martinez)
2401 * fixed SXHASH, giving different results for NIL depending on type
2402 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2403 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2404 variable. (found by Rolf Wester)
2405 * SBCL does not ignore type declarations for special
2406 variables. (reported by rif on c.l.l 2003-03-05)
2407 * some bug fixes in contrib/sb-aclrepl/
2408 * fixed some bugs revealed by Paul Dietz' test suite:
2409 ** a bug in the CONS type specifier, whereby the CAR and CDR
2410 types got intertwined, has been fixed;
2411 ** the type system is now able to reason about the interaction
2412 between INTEGER and RATIO types more completely;
2413 ** APPEND, [N]REVERSE and NRECONC check that those their
2414 arguments, which must be proper lists, are really so;
2415 ** An array specialized to be unable to hold elements has been
2416 implemented, as required -- yes, really -- by ANSI;
2417 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2418 on malformed property lists;
2420 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2421 * incompatible packaging change: in line with Unix convention,
2422 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2423 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2424 * REQUIRE and PROVIDE are now optionally capable of doing something
2425 useful. See the documentation string for REQUIRE.
2426 * infrastructure for a managed SBCL contrib system: contributed
2427 modules in this release include:
2428 ** the ASDF system definition facility;
2429 ** an interface to the BSD Sockets API;
2430 ** an ACL-like convenience interface to the repl;
2431 (thanks to Kevin Rosenberg)
2432 ** an implementation of ROTATE-BYTE, with efficient implementation
2434 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2435 gives a closer approximation to the right answer than previously.
2436 (thanks to Raymond Toy)
2437 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2438 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2439 optional environment arguments, as required by ANSI.
2440 * fixed bugs in other functions taking environment objects, allowing
2441 calls with an explicit NIL environment argument to be compiled
2443 * fixed bug 228: primary return values from
2444 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2445 COMPILE or FUNCTION.
2446 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2447 :TYPE will no longer signal errors on innocuous objects.
2448 * fixed bug 231b: SETQ is better at respecting type declarations in
2449 the lexical environment.
2450 * fixed a bug in DEFCLASS: classes named by symbols with no or
2451 unprintable packages can now be defined.
2452 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2453 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2454 * fixed compiler failure related to checking types of functions.
2455 (reported by Robert E. Brown)
2456 * the compiler is now much more consistent in its error-checking
2457 treatment of bounding index arguments to sequence functions: in
2458 (SAFETY 3) code, errors will be signalled in almost all cases if
2459 invalid sequence bounding indices are passed to functions defined
2460 by ANSI to operate on sequences.
2461 * fixed a bug in the build procedure: documentation of SBCL-specific
2462 packages is now preserved and available in the final Lisp image.
2463 * lifted FDEFINITION lookup out of loops in the implementation of
2464 many list operations. (thanks to Robert E. Brown)
2465 * fixed a bug in the reader: the #n# reader macro now works for
2466 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2467 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2468 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2469 E. Brown and Tony Martinez respectively)
2470 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2471 SLOT-BOUNDP: in particular, it is now able to optimize them much
2472 better, and is now not vulnerable to having packages renamed.
2473 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2474 always called when a slot is not present in an instance. (thanks
2476 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2477 longer signal an error on unreading a character following EOF on
2478 the previous constituent stream. (thanks to Tony Martinez)
2479 * fixed some bugs revealed by Paul Dietz' test suite:
2480 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2481 not just nonnegative fixnums;
2482 ** the logical bit-array operators such as BIT-AND now accept an
2483 explicit NIL for their "opt-arg" argument (to indicate a
2484 freshly-consed result bit-array);
2485 ** ELT now signals an error on an invalid sequence index in safe
2487 ** the type system is now cleverer about negations of numeric
2488 types, and consequently understands the BIGNUM and RATIO types
2490 ** the type system is now cleverer about the interaction between
2491 INTEGER and RATIO types: while bugs still remain, many more
2492 cases are accurately computed;
2493 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2494 if it is in the last clause;
2495 ** CONSTANTLY now correctly returns a side-effect-free function in
2497 ** DECLARE is no longer treated as a special-operator; in
2498 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2499 * incremented fasl file version number due to the change in the
2500 DEFSTRUCT-SLOT-DESCRIPTION structure.
2502 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2503 * minor incompatible change: code processed by the "interpreter" or
2504 EVAL now has a compilation optimization policy of (DEBUG 2)
2505 (changed from (DEBUG 1)) to improve debuggability of interactive
2506 development, and to allow the use of the debug RETURN command in
2508 * an experimental implementation of the RETURN command for the
2509 debugger has been included. (thanks to Frederik Kuivinen)
2510 * fixed bug 62: constraints were not propagated into a loop.
2511 * fixed bug in embedded calls of SORT (reported and investigated by
2513 * fixed some bugs revealed by Paul Dietz' test suite:
2514 ** printing and reading of arrays with some dimensions having
2515 length 0 (thanks to Gerd Moellmann);
2516 ** BOA constructor with &AUX argument without a default value does
2517 not cause a type error;
2518 ** CONSTANTP now returns true for all self-evaluating objects.
2520 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2521 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2522 accessors that are related by inheritance, as specified in the
2523 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2524 Valtteri Vuorikoski)
2525 * The compiler is now able to inline functions that were defined in
2526 a complex lexical environment (e.g. inside a MACROLET).
2527 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2529 * Support for the upcoming FreeBSD-5.0 release has been included.
2530 (thanks to Dag-Erling Smorgrav)
2531 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2532 effect when it is not in a toplevel context.
2533 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2534 stand a better chance of being correct. (thanks to Gerd
2536 * fixed bug in COERCE, which now signals an error on coercing a
2537 rational to a bounded real type which excludes the expected
2539 * The compiler is now able to derive types more accurately from the
2540 COERCE and COMPILE functions.
2541 * fixed bug 223: functional binding is considered to be constant
2542 only for symbols in the CL package.
2543 * fixed bug 231: SETQ did not check the type of a variable being set
2544 (reported by Robert E. Brown)
2545 * A new optimization for MAKE-INSTANCE has been included, fixing
2546 various bugs (including relating to :ALLOCATION :CLASS slots and
2547 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2549 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2550 ** As required by ANSI, LOOP now disallows anonymous collection
2551 clauses such as COLLECT I in conjunction with aggregate boolean
2552 clauses such as THEREIS (= I 1);
2553 ** LOOP now signals an error when any variable is reused in the
2554 same loop (including the potentially useful construct analogous
2555 to WITH A = 1 WITH A = (1+ A);
2556 ** IT is only a special loop symbol within the first clause of a
2557 conditional loop clause;
2558 ** LOOP with a typed iteration variable over a hashtable now
2559 signals a type error iff it should.
2560 * fixed some other bugs revealed by Paul Dietz' test suite:
2561 ** FILE-STREAM now names the class previously known as FD-STREAM;
2562 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2563 argument) no longer signals an error;
2564 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2565 of the slot symbol, rather than using the current package
2566 ((:CONC-NAME "") continues to intern the slot's name in the
2568 * incremented fasl file version number, because of the incompatible
2569 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2570 of the new implementation of DEFINE-COMPILER-MACRO.
2572 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2573 * Support for building SBCL for MIPS platforms running in
2574 little-endian mode has now been checked in, and basic
2575 functionality on said platforms verified.
2576 * minor incompatible change: PCL now records the pathname of a file
2577 in which methods and the like are defined, rather than its
2579 * minor incompatible change: TRUENAME now considers the truename of
2580 a file naming a directory to be the pathname with :DIRECTORY
2581 component indicating that directory.
2582 * minor incompatible change: a NAMED clause in the extended form of
2583 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2584 reason for the previous behaviour is unclear.
2585 * more systematization and improvement of CLOS and MOP conformance
2586 in PCL (thanks to Gerd Moellman and Pierre Mai):
2587 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2589 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2590 primary methods with no specializers;
2591 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2593 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2594 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2595 FORWARD-REFERENCED-CLASSes; error reporting on
2596 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2598 ** SXHASH on CLOS instances now uses a slot internal to the
2599 instance to return different numbers on distinct instances,
2600 while preserving the same return value through invocations of
2602 ** DEFMETHOD signals errors when methods with longer incongruent
2603 lambda lists are added to generic functions;
2604 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2605 CLASS, as specified in AMOP;
2606 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2607 on the order returned by the primary method for classes of
2608 class STANDARD-CLASS;
2609 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2610 * fixed some bugs shown by Paul Dietz' test suite:
2611 ** DOLIST puts its body in TAGBODY;
2612 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2614 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2615 value producing form;
2616 ** if more variables are given to PROGV than values, extra
2617 variables are bound and made to have no value;
2618 ** NSUBSTITUTE on list arguments gets the right answer with
2620 ** ELT signals an error of type TYPE-ERROR when the index argument
2621 is not a valid sequence index;
2622 ** LOOP signals (at macroexpansion time) an error of type
2623 PROGRAM-ERROR when duplicate variable names are found;
2624 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2625 ** FUNCALL of special-operators now cause an error of type
2627 ** PSETQ now works as required in the presence of side-effecting
2628 symbol-macro places;
2629 ** NCONC accepts any object as its last argument;
2630 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2632 ** loop-for-as-package does not require a package to be explicitely
2634 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2636 * fixed bug 166: compiler preserves "there is a way to go"
2637 invariant when deleting code.
2638 * fixed bug 172: macro lambda lists with required arguments after
2639 &REST arguments now cause an error to be signalled. (thanks to
2641 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2642 bug. (thanks to Matthew Danish)
2643 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2645 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2646 arguments are lexically rebound. (thanks to Gerd Moellmann and
2648 * fixed bug 194: error messages are now more informative when there
2649 is no primary method applicable in a call to a generic
2650 function. (thanks to Gerd Moellmann)
2651 * fixed bug in command line argument checking (thanks to Julian
2653 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2654 (thanks to Matthew Danish)
2655 * incremented fasl file version number, because of the
2656 SXHASH-related changes in the layout of CLOS data structures
2658 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2659 * minor incompatible change: The runtime (the Unix executable named
2660 "sbcl") is now much pickier about the .core files it will load.
2661 Essentially it now requires .core files to descend from the same
2662 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2663 as the runtime does. (The intent is to prevent the crashes which
2664 can occur, and which can even be reported as mysterious failures,
2665 when people patch the sources or change the build parameters
2666 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2667 sbcl and .core files.)
2668 * fixed bug: VALUES-LIST is no longer optimized away.
2669 * fixed bug 142: The FFI conversion of C string values to Lisp
2670 string values no longer conses excessively. (thanks to Nathan
2671 Froyd porting Raymond Toy's fix to CMU CL)
2672 * began to systematize and improve MOP conformance in PCL (thanks to
2673 Nathan Froyd, Gerd Moellman and Pierre Mai):
2674 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2676 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2677 ** FINALIZE-INHERITANCE is now called on class finalization;
2678 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2679 argument precedence order.
2680 * fixed bug 202: The compiler no longer fails on functions whose
2681 derived types contradict their declared type.
2682 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2683 so it can be non-toplevel.
2684 * The fasl file version number has changed (because of the new
2685 implementation of DEFMACRO).
2686 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2687 safe code, that any length requirement by their type-specifier
2688 argument is valid. The exceptions to this are described in bug
2690 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2691 functions now check the types of their inputs as required by ANSI.
2692 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2693 introduced symbol is DECLAREd to be SPECIAL.
2694 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2695 * fixed bug: :COUNT argument to sequence functions may be negative
2696 * fixed bug: body of DO-SYMBOLS may contain declarations
2697 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2698 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2699 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2700 symbol macro only once
2701 * fixed printing of call frame when argument list is unavailable
2702 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2703 * compiler no longer signals WARNING on unknown keyword
2706 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2707 * A beta-quality port to the mips architecture running Linux,
2708 based on the old CMUCL backend, has been made. It has been tested
2709 on a big-endian kernel, and works sufficiently well to be able to
2710 rebuild itself; it has not been tested in little-endian mode.
2711 * fixed an inconsistency between gencgc.c and purify.c which made
2712 dumping/loading .core files unreliable
2713 * fixed bug 120a: The compiler now deals correctly with IFs where
2714 the consequent is the same as the alternative, instead of
2715 misderiving the return type. (thanks to Alexey Dejneka)
2716 * fixed bug 113: Logical pathnames are now dumpable (the logical
2717 host is resolved at load-time, throwing an error if it is not
2719 * fixed bug 174: FORMAT's error message is slightly clearer when a
2720 non-printing character is used in a format directive.
2721 * fixed several bugs in compiler checking of type declarations, i.e.
2722 violations of the Python "declarations are assertions" principle
2723 (thanks to Alexey Dejneka)
2724 * fixed several bugs in PCL's error checking (thanks to Gerd
2726 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2728 * fixed bug in compilation of functions as first class values
2729 (thanks to Antonio Martinez-Shotton)
2730 * The compiler's handling TYPE-ERRORs which it can prove will
2731 inevitably happen at runtime has been cleaned up and corrected
2732 in several ways. (thanks to Alexey Dejneka)
2733 * improved argument type checking for various basic arithmetic
2734 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2735 lavished on them in the past that they can be compiled in many
2736 ways in different special cases
2737 * fixed bug 181: compiler checks validity of user supplied type
2739 * cleaned up code flushing in optimization: Function calls which
2740 should signal errors for safety purposes (e.g. which ANSI says
2741 should signal errors when their arguments are of incorrect type)
2742 are no longer optimized away.
2743 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2744 * incremented fasl file version number, because changes in the
2745 implementation of sequence functions like COERCE caused
2746 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2747 in old inline expansions) to become undefined. (Actually these
2748 changes were later undone, so we might very well be binary
2749 compatible with 0.7.7 after all, but leaving the version number
2750 incremented seemed like the simplest and most conservative
2753 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2754 * An alpha-quality port to the parisc architecture running Linux,
2755 based on the old CMUCL backend, has been made. This, even more so
2756 than the other backends, should be considered still a work in
2757 progress; known problems include that the Linux kernel in 64-bit
2758 mode does not propagate the correct sigcontext structure to
2759 userspace, and consequently SBCL on a parisc64 kernel will not
2761 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2762 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2763 suppressed.) Also now that the compiler is looking at declarations
2764 in the environment, it checks optimization declarations as well,
2765 and suppresses inlining when (> DEBUG SPEED).
2766 * More fixes have been made to treatment of floating point exception
2767 treatment and other Unix signals. In particular, floating point
2768 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2769 * The detection and handling of control stack exhaustion (infinite
2770 or very deeply nested recursion) has changed. Stack exhaustion
2771 detection is now done by write-protecting pages at the OS level
2772 and applies at all optimization settings; when found, a
2773 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2774 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2775 be caught using IGNORE-ERRORS.
2776 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2777 correctly when transforming recursive LABELS functions to LETs.
2778 (thanks to Alexey Dejneka)
2779 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2780 that are names of constants or global variables.
2781 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2782 alien routines with docstrings.
2783 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2784 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2786 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2787 (thanks to Eric Marsden)
2788 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2789 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2790 * Bug fix: The SPARC backend can now compile functions involving
2791 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2792 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2793 to a functions expecting a PATHNAME-DESIGNATOR.
2794 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2795 lambda lists. (thanks to Alexey Dejneka)
2796 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2797 functions so that EXPORTing the name of the class doesn't cause
2798 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2799 to Antonio Martinez for reporting this)
2800 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2801 constant-folded without failing an assertion. (thanks to Einar
2802 Floystad Dorum for reporting this)
2803 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2804 types are now dealt with more correctly by the compiler.
2805 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2806 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2807 behaviour analogous to RENAME-FILE. This puts its behaviour more
2808 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2810 * The fasl file version number has changed again. (because of the
2811 bug fix involving the names of PCL MAKE-INSTANCE functions)
2813 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2814 * bug fix: Floating point exceptions are treated much more
2815 consistently on the x86/Linux and PPC/Linux platforms.
2816 * Array initialization with :INITIAL-ELEMENT is now much faster for
2817 cases when the compiler cannot open code the array creation, but
2818 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2819 array accesses have also seen a speed increase.
2820 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2821 specified. (This was at the root of some bad interactions between
2822 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2824 * bug fix: Internal error arguments for undefined functions are now
2825 computed correctly on the PPC/Linux platform.
2826 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2827 Raymond Toy's patch for CMU CL)
2828 * Support for the Solaris 9 operating environment has been included
2829 (thanks to Daniel Merritt)
2830 * A very ugly but hopefully complete draft of the missing FFI chapter
2831 of the manual has been created by reformatting the corresponding
2832 CMU CL manual chapter into (currently very ugly and incoherent)
2833 DocBook and bringing it up to date for SBCL behavior. Thus, the
2834 manual is now essentially complete, at least by my extreme
2835 once-and-only-once standards, whereby it's acceptable to refer to
2836 the doc strings of SB-EXT functions as the primary documentation.
2837 * The fasl file version number has changed again, due to cleanup of
2838 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2839 is no longer a static symbol.)
2841 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2842 * SBCL now builds with OpenMCL (version 0.12) as the
2843 cross-compilation host; also, more progress has been made toward
2844 bootstrapping under CLISP.
2845 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2847 * bug 158 fixed: The compiler can now deal with integer loop
2848 increments different from 1; fixing this turned out also to fix
2850 * bug 169 fixed: no more bogus warnings about using lexical bindings
2851 despite the presence of perfectly good SPECIAL declarations (thanks
2852 to David Lichteblau)
2853 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2854 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2855 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2857 * bug 180 fixed: Method combination specifications no longer ignore
2858 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2859 * bug fix: Structure type predicate functions now check their argument
2860 count as they should.
2861 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2862 correctly. (thanks to Pierre Mai)
2863 * minor incompatible change: The --noprogrammer option is deprecated
2864 in favor of the new --disable-debugger option, which is very similar.
2865 (The major difference is that it takes effect at a slightly different
2866 time at startup, causing handling of errors in --sysinit and
2867 --userinit files will be affected differently.) The
2868 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2869 been added to allow this functionality to be controlled from ordinary
2870 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2871 maintainers, who might want to run non-interactive scripts to
2872 build SBCL cores which will later be used interactively.)
2873 * minor incompatible change: The LOAD function no longer, when given
2874 a wild pathname to load, loads all files matching that pathname.
2875 Instead, an error of type FILE-ERROR is signalled.
2877 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2878 * bug 147 fixed: The compiler preserves its block link/count
2879 invariants more correctly now so that it doesn't crash. (thanks
2881 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2883 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2884 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2885 no-value-for-this-character as NIL (as specified by ANSI).
2886 * HOST-NAMESTRING on physical pathnames now returns a string that is
2887 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2888 (thanks to Christophe Rhodes)
2889 * The Alpha port handles icache flushing more correctly. (thanks to
2891 * More progress has been made toward bootstrapping under CLISP. (thanks
2892 to Christophe Rhodes)
2893 * The fasl file format has changed again, because dynamic loading
2894 on OpenBSD (which has non-ELF object files) motivated some cleanups
2895 in the way that foreign symbols are transformed and passed around.
2896 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2897 is no longer treated as whitespace by the reader, but instead as
2898 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2899 |A
\7fB|, instead of A as it used to.
2901 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2902 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2903 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2904 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2905 this as of 0.7.1.45, but was left out of the previous news section
2906 (thanks to Dan Barlow)
2907 * SBCL now runs on the Solaris operating system on SPARC architectures
2908 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2909 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2910 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2911 * SPARC backend cleanups, allowing builds of cores optimized for V8
2912 and V9 SPARCS, and also emission of code targeted to a particular
2913 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2915 * SBCL is closer to bootstrapping under CLISP, thanks to various
2916 fixes by Christophe Rhodes.
2917 * The fasl file format has changed again, to allow the compiler's
2918 INFO database to support symbol macros.
2919 * The user manual (in doc/) is formatted into HTML more nicely.
2920 (thanks to coreythomas)
2921 * The system is smarter about SUBTYPEP relationships, especially
2922 those involving NOT types (including types such as ATOM which are
2923 represented internally using NOT types). Thus SUBTYPEP is less
2924 likely to return (VALUES NIL NIL) in general, and in particular
2925 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2927 * The fasl file format has changed again, because the internal
2928 representation of types now includes a new slot to support the new
2929 SUBTYPEP-of-NOT-types logic.
2930 * (not a change in the main branch of SBCL, but a related prototype
2931 which can hopefully be merged into the main branch of SBCL in the
2932 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2933 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2934 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2935 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2936 dispatch macro character combinations. (thanks to Alexey Dejneka)
2937 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2938 that superficially look like logical namestrings correctly.
2939 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2941 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2942 default supplied-p) for &optional and &key arguments. (thanks to
2945 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2946 * incompatible change: The compiler is now less aggressive about
2947 tail call optimization, doing it only when (> SPACE DEBUG) or
2948 (> SPEED DEBUG). (This is an incompatible change because there are
2949 programs which relied on the old CMU-CL-style behavior to optimize
2950 away their unbounded recursion which will now die of stack overflow.)
2951 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2952 for non-GENCGC systems has been increased to 20M (since that
2953 seems much closer to the likely performance optimum for modern
2954 systems than the old 4M value was)
2955 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2956 and *DEBUG-PRINT-LEVEL*
2957 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2958 of CMU CL's support for SPARC, and various endianness and other
2959 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2960 * new syntactic sugar for the Unix command line: --load foo.bar is now
2961 an alternate notation for --eval '(load "foo.bar")'.
2963 ** The system now detects stack overflow and handles it gracefully,
2964 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2965 optimization settings. (This is a good thing in general, and
2966 its introduction in this version should be particularly timely
2967 for anyone whose code fails because of suppression of tail
2969 ** The system now hunts for the C variable "environ" in a more
2970 devious way, to avoid segfaults when the C library version differs
2971 between compile time and run time. (thanks to Christophe Rhodes)
2972 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2973 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2974 ** The compiler no longer issues bogus style warnings for undefined
2975 classes in the same source file as the DEFCLASSes which defined
2976 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2978 ** fixes in CONDITION class precedence list for undefined function
2979 errors (thanks to Alexei Dejneka)
2980 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2981 correctly. (thanks to Dan Barlow)
2982 ** portability fixes aiming at bootstrapping under CLISP (thanks
2983 to Dave McDonald and Christophe Rhodes)
2984 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2985 ** fixes in type translation and and type inference (thanks to
2987 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2988 ** various fixes in the new ports (thanks to Dan Barlow)
2989 * several changes related to debugging:
2990 ** suppression of tail recursion, as noted above
2991 ** stack overflow detection, as noted above
2992 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2993 is now the default. (For some time encapsulation has been more
2994 reliable than the breakpoint-based :ENCAPSULATE NIL
2995 implementation, at least on X86 systems; and I just noticed that
2996 encapsulation also seems closer to the spirit of the ANSI
2999 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3001 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3002 up properly again. (There was a packaging bug in 0.7.0 which
3003 left their definitions in SB-SYS::LOAD-FOREIGN and
3004 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3005 vital for most things which interface to C-level interfaces,
3006 like extensions working with sockets or databases or
3007 Perl-compatible regexes or whatever, and the need to fix
3008 this bug is the main reason that 0.7.1 was released so
3010 ** DEFGENERIC is now choosier about the methods it redefines, so that
3011 reLOADing a previously-LOADed file containing DEFGENERICs does
3012 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3013 cycle now works as it should. (thanks to Alexey Dejneka)
3014 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3015 fixed by Christophe Rhodes. (He actually submitted this patch
3016 months ago, and I delayed until after 0.7.0.)
3017 ** Bug 111 (internal compiler confusion about runtime checks on
3018 FUNCTION types) was fixed by Alexey Dejneka.
3019 * Some internal cleanups (getting rid of variables which aren't
3020 needed now that the byte interpreter is gone) caused the fasl
3021 file format number to change again.
3023 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3024 * major incompatible change: The default fasl file extension, i.e. the
3025 default extension for files produced by COMPILE-FILE, has changed
3026 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3028 ** There are many changes in the implementation of the compiler.
3029 SBCL is now essentially a compiler-only implementation of ANSI
3030 Common Lisp. EVAL still "interprets" a few special cases, but
3031 almost all the interesting cases are handled by creating
3032 a LAMBDA expression, calling COMPILE on it, then calling
3033 FUNCALL on the result.
3034 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3035 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3036 Since the code is newer, there might still be some new bugs
3037 (though not as many as before Martin Atzmueller's fixes:-). But
3038 the new code is substantially simpler and clearer, and hopefully
3039 any remaining bugs will be simpler, less fundamental, and more
3040 fixable then the bugs in the old code.
3041 ** The revised compiler is still a little unsteady on its feet.
3043 *** The debugging information it produces (particularly the names
3044 of FUNCTION objects) is sometimes much less useful than what
3045 the old compiler produced.
3046 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3047 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3048 within a MACROLET) has been temporarily weakened.
3049 ** There are new compiler optimizations for various functions:
3050 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3051 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3052 *** the math functions TRUNCATE, FLOOR, and CEILING
3053 *** the function-of-all-trades COERCE
3054 Mostly these should be transparent, but there's one
3055 potentially-annoying problem (bug 117): when the compiler
3056 inline-expands a function and does type analysis on the result,
3057 it can create control paths which have type mismatches, and
3058 when it can't prove that those control paths aren't taken,
3059 it will issue WARNINGs about the type mismatches. This is
3060 a particular problem in practice for the new sequence functions.
3061 It's not clear how this should be fixed, and for now, a
3062 workaround is given in the entry for 117 in the BUGS file.
3063 ** (Because of the interaction between the two previous items --
3064 occasional inlining problems and new inline expansions -- some
3065 of the new sequence function optimizations won't really kick in
3066 completely until debugging information, and then inlining, are
3067 straightened out in some future version.)
3068 * minor incompatible changes:
3069 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3070 DIRECTORY no longer implicitly promotes NIL slots of its
3071 pathname argument to :WILD. In particular, when you ask for the
3072 contents of a directory (which you used to be able to do without
3073 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3074 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3075 ** changes in behavior that ANSI explicitly defines to be
3076 implementation dependent:
3077 *** The new compiler-only implementation still conforms with ANSI,
3078 but acts a little different than before. Besides the obvious
3079 changes in performance tradeoffs (that the cost per form passed
3080 to EVAL has gone up, and the cost per form executed by EVAL
3081 has gone down), the behavior of the system changes a little
3082 because there are no longer any interpreted function objects.
3083 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3084 e.g. doing COMPILE on the output of interactive DEFUN is
3086 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3088 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3089 has changed from (:CL) to NIL.
3090 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3091 e.g. #\Newline and #\Tab, don't have names specified in the
3092 ANSI Common Lisp standard, is now based on their ASCII symbolic
3093 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3094 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3095 are no longer used for output.
3096 ** changes in internal implementation constants:
3097 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3098 4 million. (If your application spends a lot of time GCing and
3099 you have a lot of RAM, you might want to experiment with
3100 increasing it even more.)
3101 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3102 However, almost all old code should still continue to work without
3103 immediate update, as SB-C-CALL is now a (deprecated) nickname
3105 ** Old operator names in the style DEF-FOO are now deprecated in
3106 favor of new corresponding names DEFINE-FOO, for consistency with
3107 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3108 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3109 internal symbols, but a few supported extensions like
3110 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3111 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3112 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3113 etc. as you get deeper into recursive calls to the debugger
3114 command loop, instead of the old "5]", "5]]", "5]]]"
3115 sequence. (I was motivated to do this when squabbles between
3116 ILISP and SBCL left me very deeply nested in the debugger. In the
3117 short term, this change will probably provoke more ILISP/SBCL
3118 squabbles, but hopefully it will be an improvement in the long run.)
3119 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3120 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3121 function, since declaiming return types involving aliens is
3122 (1) annoyingly messy to do by hand and (2) vital to efficient
3123 compilation of code which calls such functions.
3124 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3125 longer reexported by the SB-EXT package. They're solely useful
3126 for alien code, so it seems more logical that you should get
3127 them from the SB-ALIEN package, not in SB-EXT.
3128 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3129 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3130 features. Instead, the code that they used to control is always
3131 built into the system.
3132 * many other bug fixes
3133 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3134 advantage of the new EVAL-WHEN stuff and to clean them up in
3135 general, and they are now more ANSI-compliant in a number of
3136 ways. Martin Atzmueller is responsible for a lot of this.
3137 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3139 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3140 *** correct ERROR type for various file operations
3141 *** some fixes for Lisp streams
3142 *** DEFMETHOD syntax checking
3143 *** changing old weird representation of debug information as
3144 strings (which, among their other deficiencies, don't transform
3145 correctly when you rename packages, and don't change their
3146 print representation when you change things like *PACKAGE*
3147 and *PRINT-LENGTH*) to symbols and lists of symbols
3148 He also made several improvements and fixed several bugs in DESCRIBE.
3149 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3151 *** misbehavior of WRITE-STRING/WRITE-LINE
3152 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3153 and several other LOOP problems as well
3154 *** DIRECTORY when similar filenames are present
3155 *** DEFGENERIC with :METHOD options
3156 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3157 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3158 *** argument ordering in FIND with :TEST option
3159 *** mishandled package designator argument in APROPOS-LIST
3160 *** various problems in the backquote readmacro
3161 *** a bug in APROPOS
3162 *** probably some others that I'm not describing very well here,
3163 since the CVS log documents them by reference to sbcl-devel
3164 messages, and the SourceForge archives aren't working well.:-(
3165 ** Dan Barlow improved the Alpha port (and is making progress on the
3166 PPC port, for those of you who think different).
3167 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3168 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3169 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3170 continues to work on the SPARC port (for those of you in a position
3171 to look down upon our little PC-compatible boxes from a great height).
3172 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3173 and uses that copy, rather than the current dynamic value, when
3174 it's trying to decide whether to truncate output. Thus e.g.
3175 (let ((*print-lines* 50))
3176 (pprint-logical-block (stream nil)
3178 (let ((*print-lines* 8))
3179 (print (aref possiblybigthings i) stream)))))
3180 should now truncate the logical block only at 50 lines, instead of
3181 often truncating it at 8 lines, as it did before.
3182 * The doc/cmucl/ directory, containing old CMU CL documentation
3183 from the time of the fork, is no longer part of the base system.
3184 SourceForge has shut down its anonymous FTP service, and with it
3185 my original plan for distributing the old CMU CL documentation
3186 there. For now, if you need these files you can download an old
3187 SBCL source release and extract them from it.
3188 * The fasl file version number changed again, for dozens of reasons,
3189 some of which are apparent above.
3191 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3192 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3193 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3194 patch, so that DISASSEMBLE output is much nicer.
3195 * The code in the SB-PROFILE package now seems reasonably stable.
3196 I still haven't decided what the final interface should look like
3197 (I'd like PROFILE to interact cleanly with TRACE, since both
3198 facilities use function encapsulation) but if you have a need
3199 for profiling now, you can probably use it successfully with
3200 the current CMU-CL-style interface.
3201 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3202 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3203 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3204 link now returns the dangling link itself, and for similar
3205 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3206 link itself. (In these cases the old code signalled an error and
3207 looped endlessly, respectively.) Thus, DIRECTORY now works even
3208 in the presence of dangling and cyclic symbolic links.
3209 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3210 is now a supported extension again, since the consensus on
3211 sbcl-devel was that it can be useful for ordinary development
3212 work, not just for debugging SBCL itself.
3213 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3214 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3215 that currently-defined functions might be redefined later with
3216 different return types.
3217 * Hash tables can be printed readably, as inspired by CMU CL code
3218 of Eric Marsden and SBCL code of Martin Atzmueller.
3219 * better error handling in CLOS method combination, thanks to
3220 Martin Atzmueller porting Pierre Mai's CMU CL patches
3221 * more overflow fixes for >16Mbyte I/O buffers
3222 * A bug in READ has been fixed, so that now a single Ctrl-D
3223 character suffices to cause end-of-file on character streams.
3224 In particular, now you only need one Ctrl-D at the command
3225 line (not two) to exit SBCL.
3226 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3228 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3229 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3230 does the right thing.
3231 * The implementation of some type tests, especially for CONDITION
3232 types, is now tidier and maybe faster, due to CMU CL code
3233 originally by Douglas Crosher, ported by Martin Atzmueller.
3234 * Some math functions have been fixed, and there are new
3235 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3236 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3237 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3238 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3239 likely to be built into sbcl-0.7.0.)
3240 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3241 CL, ported to SBCL by Martin Atzmueller)
3242 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3243 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3244 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3245 * There's a new slam.sh hack to shorten the edit/compile/debug
3246 cycle for low-level changes to SBCL itself, and a new
3247 :SB-AFTER-XC-CORE target feature to control the generation of
3248 the after-xc.core file needed by slam.sh.
3249 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3250 COMPILE-FILE is no longer supported, so that now every function
3251 gets an entry point, so that block compilation looks a little
3252 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3253 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3254 returns the number of bytes consed since the system started,
3255 rather than the number consed since the first time the function
3256 was called. (The new definition parallels ANSI functions like
3257 CL:GET-INTERNAL-RUN-TIME.)
3258 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3259 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3260 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3261 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3262 * Fasl file version numbers are now independent of the target CPU,
3263 since historically most system changes which required version
3264 number changes have affected all CPUs equally. Similarly,
3265 the byte fasl file version is now equal to the ordinary
3268 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3269 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3270 is no longer recognized. I apologize for this, because it was
3271 listed in SB-EXT as a supported extension, but I found that
3272 its existing behavior was poorly specified, as well as incorrectly
3273 specified, and it looked like too much of a mess to straighten it
3274 out. I have enough on my hands trying to get ANSI stuff to work..
3275 * many patches ported from CMU CL by Martin Atzmueller, with
3276 half a dozen bug fixes in pretty-printing and the debugger, and
3277 half a dozen others elsewhere
3278 * fixed bug 13: Floating point infinities are now supported again.
3279 They might still be a little bit flaky, but thanks to bug reports
3280 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3281 as flaky as they were.
3282 * The --noprogrammer command line option is now supported. (Its
3283 behavior is slightly different in detail from what the old man
3284 page claimed it would do, but it's still appropriate under the
3285 same circumstances that the man page talks about.)
3286 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3287 are now supported, and enabled by default. Thus, the compiler can
3288 handle many floating point and complex operations much less
3289 inefficiently. (Thus e.g. you can implement a complex FFT
3291 * The compiler now detects type mismatches between DECLAIM FTYPE
3292 and DEFUN better, and implements CHECK-TYPE more correctly, and
3293 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3294 thanks to patches from Martin Atzmueller.
3295 * various fixes to make the cross-compiler more portable to
3296 ANSI-conforming-but-different cross-compilation hosts (notably
3297 Lispworks for Windows, following bug reports from Arthur Lemmens)
3298 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3299 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3300 fixed thanks to Pierre Mai's CMU CL patches.
3301 * fixes to keep the system from overflowing internal counters when
3302 it tries to use i/o buffers larger than 16M bytes
3303 * fixed bug 45a: Various internal functions required to support
3304 complex special functions have been merged from CMU CL sources.
3305 (When I was first setting up SBCL, I misunderstood a compile-time
3306 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3307 * improved support for type intersection and union, fixing bug 12
3308 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3309 more obscure bugs as well
3310 * some steps toward byte-compiling non-performance-critical
3311 parts of the system, courtesy of patches from Martin Atzmueller
3312 * Christophe Rhodes has made some debian packages of sbcl at
3313 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3314 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3315 stable, but are nonetheless usable. When he's ready, I'd be happy
3316 to add them to the SourceForge "File Releases" section. (And if
3317 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3318 * new fasl file format version number (because of changes in
3319 internal representation of (OR ..) types to accommodate the new
3320 support for (AND ..) types, among other things)
3322 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3323 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3325 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3327 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3328 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3329 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3330 * DESCRIBE now works on structure objects again.
3331 * Most function call argument type mismatches are now handled as
3332 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3333 know whether the function will be redefined before the call is
3334 executed. (The compiler could flag local calls with full WARNINGs,
3335 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3336 it doesn't keep track of enough information to know whether calls
3337 are local in this sense.)
3338 * Compiler output is now more verbose, with messages truncated
3339 later than before. (There should be some supported way for users
3340 to override the default verbosity, but I haven't decided how to
3341 provide it yet, so this behavior is still controlled by the internal
3342 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3343 src/compiler/ir1util.lisp.)
3344 * Fasl file format version numbers have increased again, because
3345 support for the Gray streams extension changes the layout of the
3346 system's STREAM objects.
3347 * The Gray subclassable streams extension now works, thanks to a
3348 patch from Martin Atzmueller.
3349 * The full LOAD-FOREIGN extension (not just the primitive
3350 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3351 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3352 but like most other programs, it defaults to copying the Unix
3353 environment from the original process instead of starting the
3354 new process in an empty environment.
3355 * Extensions which manipulate the Unix environment now support
3356 an :ENVIRONMENT keyword option which doesn't smash case or
3357 do other bad things. The CMU-CL-style :ENV option is retained
3358 for porting convenience.
3359 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3360 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3362 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3364 * A patch from Martin Atzmueller seems to have solved the SIGINT
3365 problem, and as far as we know, signal-handling now works cleanly.
3366 (If you find any new bugs, please report them!)
3367 * The system no longer defaults Lisp source file names to types
3368 ".l", ".cl", or ".lsp", but only to ".lisp".
3369 * The compiler no longer uses special default file extensions for
3370 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3371 seems to expect a single default extension for all compiled code,
3372 and there's no compelling reason to try to stretch the standard
3373 to allow two different extensions.) Instead, byte-compiled files
3374 default to the same extension as native-compiled files.
3375 * Fasl file format version numbers have increased again, because
3376 a rearrangement of internal implementation packages made some
3377 dumped symbols in old fasl files unreadable in new cores.
3378 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3379 many fewer weird special cases.
3380 * Bug #17 (differing COMPILE-FILE behavior between logical and
3381 physical pathnames) has been fixed, and some related misbehavior too,
3382 thanks to a patch from Martin Atzmueller.
3383 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3384 by Tim Moore, ported to SBCL by Martin Atzmueller.
3385 * Martin Atzmueller fixed several filesystem-related problems,
3386 including bug #36, in part by porting CMU CL patches, which were
3387 written in part by Paul Werkowski.
3388 * More compiler warnings in src/runtime/ are gone, thanks to
3389 more patches from Martin Atzmueller.
3390 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3393 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3395 * DESCRIBE now works on CONDITION objects.
3396 * The debugger now handles errors which arise when trying to print
3397 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3399 * The build system now uses an additional file, customize-target-features.lisp,
3400 to allow local modifications to the target *FEATURES* list. (The point of
3401 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3402 debugging features enabled, without having to worry about propagating it
3403 into everyone's system when I do a "cvs update".) When no
3404 customize-target-features.lisp file exists, the target *FEATURES* list
3405 should be constructed the same way as before.
3406 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3407 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3408 warn about behavior which is undefined under the ANSI spec. Note: This
3409 is specified by ANSI, but it's not very popular with programmers.
3410 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3411 macro in the SBCL sources for an example of a workaround which you
3412 might use to make portable ANSI-standard code which does what you want.
3413 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3414 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3415 when it's used as a non-toplevel form. (This is required in order
3416 to implement the DEFCONSTANT-EQX macro.)
3417 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3418 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3419 funny things, due to bugs in the implementation of EVAL-WHEN
3420 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3421 non-toplevel DEFCONSTANTs are uncommon.)
3422 * The core file version number and fasl file version number have been
3423 incremented, because the old noncompliant DEFCONSTANT behavior involved
3424 calling functions which no longer exist, and because I also took the
3425 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3426 * fixed bug 1 (error handling before read-eval-print loop starts), and
3427 redid debugger restarts and related debugger commands somewhat while
3429 ** The QUIT debugger command is gone, since it did something
3430 rather different than the SB-EXT:QUIT command, and since it never
3431 worked properly outside the main toplevel read/eval/print loop.
3432 Invoking the new TOPLEVEL restart provides the same functionality.
3433 ** The GO debugger command is also gone, since you can just invoke
3434 the CONTINUE restart directly instead.
3435 ** The TOP debugger command is also gone, since it's redundant with the
3436 FRAME 0 command, and since it interfered with abbreviations for the
3438 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3440 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3441 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3442 some porting work by Martin Atzmueller)
3443 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3444 with a boatload of patches which he ported to SBCL. Now that those
3446 ** The system tries to make sure that its low-priority messages
3447 are prefixed by semicolons, to help people who like to use
3448 syntax highlighting in their ILISP buffer. (This patch
3449 was originally due to Raymond Toy.)
3450 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3451 patches originally written by Raymond Toy.
3452 ** The compiler understands coercion between single-value and
3453 multiple-VALUES type expressions better, getting rid of some very
3454 weird behavior, thanks to patches originally by Robert MacLachlan
3455 and Douglas Crosher.
3456 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3457 lambda lists, thanks to a patch originally by Pierre Mai.
3458 ** The system no longer bogusly warns about "abbreviated type
3460 ** The compiler gets less confused by inlining and RETURN-FROM,
3461 thanks to some patches originally by Tim Moore.
3462 ** The system no longer hangs when dumping circular lists to fasl
3463 files, thanks to a patch originally from Douglas Crosher.
3464 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3465 undefined function error.
3466 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3467 in structure slot initforms) for now, documented workaround instead:-|
3468 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3469 * fixed bug 5: added stubs for various Gray stream functions called
3470 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3471 installed, at least appropriate type errors are generated
3472 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3473 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3474 consistently in DEFMETHOD forms.
3475 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3476 it doesn't seem to affect SBCL after all
3477 * The C runtime system now builds with better optimization and many
3478 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3480 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3482 * The system is now under CVS at SourceForge (instead of the
3483 CVS repository on my home machine).
3484 * The new signal handling code has been tweaked to treat register
3485 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3486 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3487 downstream, e.g. in debug-int.lisp, has implicit dependencies
3488 on the unsignedness of integer representation of machine words,
3489 and that caused the system to bomb out with infinite regress
3490 when trying to recover from type errors involving signed values,
3491 e.g. (BUTLAST '(1 2 3) -1).)
3492 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3493 (This was one of the bugs Peter Van Eynde reported back in July.)
3494 * The system now uses code inspired by Colin Walters' O(N)
3495 implementation of MAP (from the cmucl-imp@cons.org mailing
3496 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3497 inline the MAP operation, and there is more than one
3498 sequence argument to the MAP call (so that it can't just
3499 do ETYPECASE once and for all based on the type of the
3500 single sequence argument). (The old non-inline implementation
3501 of the general M-argument sequence-of-length-N case required
3502 O(M*N*N) time when any of the sequence arguments were LISTs.)
3503 * The QUIT :UNIX-CODE keyword argument has been renamed to
3504 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3505 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3507 * Raymond Wiker's patches to port dynamic loading from Linux to
3508 FreeBSD have been added.
3509 * The BUGS file is now more nearly up to date, thanks in large part
3510 to Martin Atzmueller's review of it.
3511 * The debugger now flushes standard output streams before it begins
3512 its output ("debugger invoked" and so forth).
3513 * The core version number and fasl file version number have both
3514 been incremented, because of incompatible changes in the layout
3516 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3517 used to not be called for a saved Lisp image.)
3518 * Martin Atzmueller's version of a patch to fix a compiler crash,
3519 as posted on sbcl-devel 13 September 2000, has been installed.
3520 * Instead of installing Martin Atzmueller's patch for the
3521 compiler transform for SUBSEQ, I deleted the compiler transform,
3522 and transforms for some similar consing operations.
3523 * A bug in signal handling which kept TRACE from working on OpenBSD
3525 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3526 away by constant folding
3527 * The system now defines its address space constants in one place
3528 (in the Lisp sources), and propagates them automatically elsewhere
3529 (through GENESIS and the sbcl.h file). Therefore, patching the
3530 address map is less unnecessarily tedious and error-prone. The
3531 Lisp names of address space constants have also been systematized.
3532 * CVS tags like dollar-Header-dollar have been removed from
3533 the sources, because they have never saved me trouble and
3534 they've been source of trouble working with patches and other
3535 diff-related operations.
3536 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3537 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3539 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3541 * The system has been ported to OpenBSD.
3542 * The system now compiles with a simple "sh make.sh" on the systems
3543 that it's supported on. I.e., now you no longer need to tweak
3544 text in the target-features.lisp-expr and symlinks in src/runtime/
3545 by hand, the make.sh takes care of it for you.
3546 * The system is no longer so grossly inefficient when compiling code
3547 involving vectors implemented as general (not simple) vectors (VECTOR T),
3548 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3549 which dares to use the various sequence functions on non-simple
3550 vectors, takes less of a performance hit.
3551 * There is now a primitive type predicate VECTOR-T-P
3552 to test for the (VECTOR T) type, so that e.g.
3553 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3554 can now be compiled with some semblance of efficiency. (The old code
3555 turned the type declaration into a full call to %TYPEP at runtime!)
3556 * AREF on (VECTOR T) is still not fast, since it's still compiled
3557 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3558 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3559 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3560 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3561 without making full calls to SUBTYPEP at runtime.
3562 (Some analogous efficiency issues for non-simple vectors specialized to
3563 element types other than T, or for non-simple multidimensional arrays,
3564 have not been addressed. They could almost certainly be handled the
3565 same way if anyone is motivated to do so.)
3566 * The changes in array handling break binary compatibility, so
3567 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3568 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3569 instead of (VALUES T T).
3570 * By following the instructions that Dan Barlow posted to sbcl-devel
3571 on 2 July 2000, I was able to enable primitive dynamic object
3572 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3573 functionality is not implemented (since it calls ld to resolve
3574 library references automatically, requiring RUN-PROGRAM for its
3575 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3576 try to resolve library references) is now supported.
3577 * The system now flushes the standard output streams when it terminates,
3578 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3579 them at several other probably-convenient times, e.g. in each pass of
3580 the toplevel read-eval-print loop, and after evaluating a form given
3581 as an "--eval" command-line option. (These changes were motivated by a
3582 discussion of stream flushing issues on cmucl-imp in August 2000.)
3583 * The source transform for TYPEP of array types no longer assumes
3584 that an array whose element type is a not-yet-defined type
3585 is implemented as an array of T, but instead punts, so that the
3586 type will be interpreted at runtime.
3587 * There is now some support for cross-compiling in make.sh: each of
3588 the phases of make.sh has its own script. (This should be transparent
3589 to people doing ordinary, non-cross-compile builds.)
3590 * Since my laptop doesn't have hundreds of megabytes of memory like
3591 my desktop machine, I became more motivated to do some items on
3592 my to-do list in order to reduce the size of the system a little:
3593 ** Arrange for various needed-only-at-cold-init things to be
3594 uninterned after cold init. To support this, those things have
3595 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3596 symbols with such names are now uninterned after cold init).
3597 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3598 fasl files for cold load.
3599 ** Remove the old compiler structure pooling code (which used to
3600 be conditional on the target feature :SB-ALLOC) completely.
3601 ** Redo the representation of some data in cold init to be more compact.
3602 (I also looked into supporting byte compiled code at bootstrap time,
3603 which would probably reduce the size of the system a lot, but that
3604 looked too complicated, so I punted for now.)
3605 * The maximum signal nesting depth in the src/runtime/ support code has
3606 been reduced from 4096 to 256. (I don't know any reason for the very
3607 large old value. If the new smaller value turns out to break something,
3608 I'll probably just bump it back up.)
3609 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3611 * Many, many bugs reported by Peter Van Eynde have been added to
3612 the BUGS list; some have even been fixed.
3613 * While enabling dynamic object file loading, I tried to make the
3614 code easier to understand, renaming various functions and variables
3615 with less ambiguous names, and changing some function calling
3616 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3617 * While trying to figure out how to do the OpenBSD port, I tried to
3618 clean up some of the code in src/runtime/. In particular, I dropped
3619 support for non-POSIX signal handling, added various comments,
3620 tweaked the code to reduce the number of compilation warnings, and
3621 renamed some files to increase consistency.
3622 * To support the new automatic configuration functionality in make.sh,
3623 the source file target-features.lisp-expr has been replaced with the
3624 source file base-target-features.lisp-expr and the machine-generated
3625 file local-target-features.lisp-expr.
3626 * fixed a stupid quoting error in make.sh so that using CMU CL
3627 "lisp -batch" as cross-compilation host works again
3629 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3631 * DESCRIBE no longer tries to call itself recursively to describe
3632 bound/fbound values, so that it no longer fails on symbols which are
3633 bound to themselves (like keywords, T, and NIL).
3634 * DESCRIBE now works on generic functions.
3635 * The printer now prints less-screwed-up representations of closures
3636 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3637 * A private symbol is used instead of the :EMPTY keyword previously
3638 used to mark empty slots in hash tables. Thus
3639 (DEFVAR *HT* (MAKE-HASH-TABLE))
3640 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3641 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3642 now does what ANSI says that it should. (You can still get
3643 similar noncompliant behavior if bang on the hash table
3644 implementation with all the symbols you get back from
3645 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3646 This breaks binary compatibility, since tests for equality to
3647 :EMPTY are wired into things like the macroexpansion of
3648 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3650 * There's now a minimal placeholder implementation for CL:STEP,
3651 as required by ANSI.
3652 * An obscure bug in the interaction of the normal compiler, the byte
3653 compiler, inlining, and structure predicates has been patched
3654 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3655 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3657 * Missing ordinary arguments in a macro call are now detected even
3658 when the macro lambda list contains &KEY or &REST.
3659 * The debugger no longer complains about encountering the top of the
3660 stack when you type "FRAME 0" to explicitly instruct it to go to
3661 the top of the stack. And it now prints the frame you request even
3662 if it's the current frame (instead of saying "You are here.").
3663 * As specified by ANSI, the system now always prints keywords
3664 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3666 * The default initial SIZE of HASH-TABLEs is now smaller.
3667 * Type information from CLOS class dispatch is now propagated
3668 into DEFMETHOD bodies, so that e.g.
3669 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3671 is now basically equivalent to
3672 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3673 (DECLARE (TYPE SINGLE-FLOAT X))
3675 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3676 operation, without having to do run-time type dispatch.
3677 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3678 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3679 * Since I don't seem to be making big file reorganizations very often
3680 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3681 is overflowing my ability to conveniently back them up), I've finally
3682 checked the system into CVS. (The CVS repository is on my home system,
3683 not at SourceForge -- putting it on SourceForge might come later.)
3684 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3685 high-level GC-NOTIFY-FOO functions send their output. (There's
3686 still very little control of where low-level verbose GC functions
3687 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3688 less than it used to -- the GC-NOTIFY-FOO functions are now under
3689 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3690 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3691 in only one place in the source code, and propagates it automatically
3692 everywhere that it's needed. Thus e.g. when I bump the version from
3693 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3694 * The C source files now include boilerplate legalese and documentation
3695 at the head of each file (just as the Lisp source files already did).
3696 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3697 to his page will be replaced with a link to his new CLiki service.
3699 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3701 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3702 * The build process now looks for GNU make under the default name "gmake",
3703 instead of "make" as it used to. If GNU make is not available as "gmake"
3704 on your system, you can change this default behavior by setting the
3705 GNUMAKE environment variable.
3706 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3707 can build without error under CMU CL.
3709 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3711 * There is now a partial SBCL user manual (with some new text and some
3712 text cribbed from the CMU CL manual).
3713 * The beginnings of a profiler have been added (starting with the
3714 CMU CL profiler and simplifying and cleaning up). Eventually the
3715 main interface should be through the TRACE macro, but for now,
3716 it's still accessed through vaguely CMU-CL-style functions and macros
3717 exported from the package SB-PROFILE.
3718 * Some problems left over from porting CMU CL to the new
3719 cross-compilation bootstrap process have been cleaned up:
3720 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3721 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3722 part when getting rid of the compiler *BACKEND* stuff.)
3723 ** The value of *NULL-TYPE* was screwed up, because it was
3724 being initialized before the type system knew the final
3725 definition of the 'NULL type. This screwed up several key
3726 optimizations in the compiler, causing inefficiency in all sorts
3727 of places. (I found it because I wanted to understand why
3728 GET-INTERNAL-RUN-TIME was consing.)
3729 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3730 PROCLAIM FTYPE information. Unfortunately this broke binary
3731 compatibility again, since now the forms output by DEFGENERIC
3732 to refer to functions which didn't exist in 0.6.3.
3733 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3734 can use the new (as of 0.6.3) transform for SOME into MAP into
3736 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3737 values to (INTEGER 0 1000000), so that the time code will no longer
3738 occasionally get blown up by Linux returning 1000000 microseconds
3739 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3740 its output conform to the ANSI spec. (Alas, this makes its output
3741 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3742 * A full call to MAP NIL with a single sequence argument no longer conses.
3743 * fixes to problems pointed out by Martin Atzmueller:
3744 * The manual page no longer talks about multiprocessing as though
3745 it were currently supported.
3746 * The ILISP support patches have been removed from the distribution,
3747 because as of version 5.10.1, ILISP now supports SBCL without us
3748 having to maintain patches.
3749 * added a modified version of Raymond Toy's recent CMU CL patch for
3750 EQUALP comparison of HASH-TABLE
3752 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3754 * The system still can't cross-compile itself with
3755 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3756 entails), but at least it can compile more of itself that way
3757 than it used to be able to, and various buglets which were uncovered
3758 by trying to cross-compile itself that way have now been fixed.
3759 * This release breaks binary compatibility again. This time
3760 at least I've incremented the FASL file format version to 2, so that the
3761 problem can be detected reliably instead of just causing weird errors.
3762 * various new style warnings:
3763 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3764 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3765 ** using the lexical binding of a variable named in the *FOO* style
3766 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3767 as specified by ANSI.
3768 * *RANDOM-STATE* is no longer automatically initialized from
3769 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3770 default behavior of the system is to repeat its behavior every time
3771 it's run. If you'd like to change this behavior, you can always
3772 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3773 old convention there was no comparably easy way to get the system to
3774 repeat its behavior every time it was run.
3775 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3776 FTYPE declarations has been removed, in favor of their ANSI
3777 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3778 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3779 the types of their sequence arguments can be determined at compile time.
3780 This is done through a new open code expansion for MAP which eliminates
3781 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3782 argument types can be determined at compile time.
3783 * The optimizer now transforms COERCE into an identity operation when it
3784 can prove that the coerced object is already of the correct type. (This
3785 can be a win for machine generated code, including the output of other
3786 optimization transforms, such as the MAP transform above.)
3787 * Credit information has been moved from source file headers into CREDITS.
3788 * Source file headers have been made more standard.
3789 * The CASE macro now compiles without complaining even when it has
3792 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3794 * (Note that the way that the PCL macroexpansions were rewritten
3795 to accommodate the change in DEFGENERIC below breaks binary
3796 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3797 not run under sbcl-0.6.2. Once we get out of alpha releases,
3798 i.e. hit release 1.0.0, we'll probably try to maintain binary
3799 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3800 and sbcl-1.4.4. Until then, however, it might be fairly common
3801 for maintenance releases to break binary compatibility.)
3802 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3804 * The reporting of unhandled signals has been changed to print some
3805 explanatory text as well as the report form. (Previously only
3806 the report form was printed.)
3807 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3808 it defines, so that the compiler no longer issues undefined function
3809 warnings for compiled-but-not-yet-loaded generic functions.
3810 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3811 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3812 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3813 section of the ANSI spec are supported.
3814 * The "" nickname for the "KEYWORD" package has been removed.
3815 The reader still handles symbol tokens which begin with a package marker
3816 as keywords, but it doesn't expose its mechanism for doing so in the
3817 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3818 * The system now issues STYLE-WARNINGs for contradictory TYPE
3819 proclamations. (Warnings for contradictory FTYPE proclamations would
3820 be nice too, but those can't be done usefully unless the type system
3821 is made smarter about FUNCTION types.)
3822 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3823 systematized, so that "*target-*.lisp is supposed to exist only on the
3824 target and imply that there's a related file which exists on the
3825 host, and *host-*.lisp is supposed to exist only on the host and imply
3826 that there's a related file which exists on the target. This involves a
3827 lot of renaming. Hopefully the acute confusion caused by the renaming
3828 will be justified by the reduction in chronic confusion..
3829 ** runtime-type.lisp -> early-target-type.lisp
3830 ** target-type.lisp -> late-target-type.lisp
3831 ** early-host-format.lisp -> early-format.lisp
3832 ** late-host-format.lisp -> late-format.lisp
3833 ** host-error.lisp -> misc-error.lisp
3834 ** early-error.lisp -> early-target-error.lisp
3835 ** late-error.lisp -> late-target-error.lisp
3836 ** host-defboot.lisp -> early-defboot.lisp
3837 ** code/misc.lisp -> code/target-misc.lisp
3838 ** code/host-misc.lisp -> code/misc.lisp
3839 ** code/numbers.lisp -> code/target-numbers.lisp
3840 ** code/early-numbers.lisp -> numbers.lisp
3841 ** early-host-type.lisp -> early-type.lisp
3842 ** late-host-type.lisp -> late-type.lisp
3843 ** host-typep.lisp -> typep.lisp
3844 ** load.lisp -> target-load.lisp
3845 ** host-load.lisp -> load.lisp
3846 ** host-disassem.lisp -> disassem.lisp
3847 ** host-insts.lisp -> insts.lisp
3848 ** byte-comp.lisp -> target-byte-comp.lisp
3849 ** host-byte-comp.lisp -> byte-comp.lisp
3850 ** host-signal.lisp -> signal.lisp
3851 ** host-defstruct.lisp -> defstruct.lisp
3852 ** late-target-type.lisp -> deftypes-for-target.lisp
3853 Furthermore, several other previously target-only files foo.lisp (e.g.
3854 hash-table.lisp and random.lisp) have been split into a target-and-host
3855 foo.lisp file and a target-only target-foo.lisp file, with their key type
3856 definitions in the target-and-host part, so that the cross-compiler will
3857 know more about target types.
3858 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3859 gone away. In their place are various *BACKEND-FOO* variables
3860 corresponding to the slots of the old structure.
3861 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3862 into the target SBCL has been fixed.
3863 * The chill.lisp system for loading cold code into a running SBCL
3865 * Support for the CMU CL "scavenger hook" extension has been removed.
3866 (It was undocumented and unused in the CMU CL sources that SBCL was
3867 derived from, and stale in sbcl-0.6.1.)
3868 * Various errors in the cross-compiler type system were detected
3869 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3870 (enabling various consistency checks). Many of them were fixed,
3871 but some hard problems remain, so the compiler is back to
3872 running without *TYPE-SYSTEM-INITIALIZED* for now.
3873 * As part of the cross-compiler type system cleanup, I implemented
3874 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3875 * I have started adding UNCROSS calls throughout the type system
3876 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3877 on cross-compiler input files will be able to go away, and various
3879 * CONSTANTP now returns true for quoted forms (as explicitly required
3882 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3884 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3885 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3886 (MAKE-LIST -1). In the longer term, it ought to become true
3887 that declarations are assertions even at SAFETY 1. For now, it's not
3888 quite true even at SAFETY 3, but it's at least more nearly true..
3889 (Note that this change seems to increases the size of the system by
3890 O(5%) and to decrease the speed of the compiler by 20% or more.)
3891 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3892 for the problem of printing dozens of lines of distracting information
3893 about low-level system machinery as part of the top stack frame
3894 on entry to the debugger when an undefined function was called.
3895 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3896 is not reset to COMMON-LISP-USER.
3897 * Compilation of stuff related to dyncount.lisp has been made conditional
3898 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3899 smaller. The various dyncount-related symbols have been moved into
3900 a new "SB-DYNCOUNT" package.
3901 * tty-inspect.lisp has been renamed to inspect.lisp.
3902 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3903 feature has gone away. (When we eventually port to other flavors of
3904 libc and/or Unix, we'll try to make the differences between flavors
3905 invisible at the user level.)
3906 * Various other *FEATURES* tags, and/or their associated conditionals,
3907 have been removed if obsolescent, or given better documentation, or
3908 sometimes given more-mnemonic names.
3910 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3912 * tidied up "make.sh" script
3913 * tidied up system directory structure
3914 * better "clean.sh" behavior
3915 * added doc/FOR-CMUCL-DEVELOPERS
3916 * many many small tweaks to output format, e.g. removing possibly-confusing
3917 trailing #\. character in DESCRIBE-INSTANCE
3918 * (EQUALP #\A 'A) no longer signals an error.
3919 * new hashing code, including EQUALP hashing
3920 * tidied up Lisp initialization and toplevel
3921 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3922 * command line argument processing
3923 * added POSIX-GETENV function to deal with Unix-ish environment variables
3924 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3925 terminating SBCL on EOF
3926 * non-verbose GC by default
3927 * There is no more "sbcl" shell script; the sbcl file is now the C
3928 runtime executable (just like CMU CL).
3929 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3931 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3932 symbol and package information as Lisp native symbol and package objects
3933 instead of strings naming symbols and strings naming packages. This way,
3934 whenever packages are renamed (as in warm init), debug information is
3935 transformed along with everything else.
3936 * tweaked the optimization policy declarations which control the building
3937 of SBCL itself. Now, among other things, the system no longer saves
3938 source location debugging information. (This helps two problems at once
3939 by reducing SBCL size and by keeping SBCL from trying to look for its
3940 sources -- which may not exist -- when reporting errors.)
3941 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3942 debugging and testing purposes
3943 * cleaned up printing, making the printer call PRINT-OBJECT for
3944 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3945 methods, giving nearly-ANSI behavior
3946 * converted almost all special variables to use *FOO* naming convention
3947 * deleted PARSE-TIME functionality, since it can be done portably
3948 * moved some files out of cold init into warm init
3949 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3951 * regularized formatting of source files
3952 * added an install.sh script
3953 * fixed ridiculous memory usage of cross-compiler by making
3954 compiler/alloc.lisp not try to do pooling unless it can hook
3955 itself into the GC of the cross-compilation host. Now the system
3956 builds nicely on my old laptop.
3957 * added :SB-ALLOC in target-features.lisp-expr
3958 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3959 was not implemented)
3960 * re-did condition handling and note reporting in the compiler. Notes
3961 are no longer handled by signalling conditions. Style warnings
3962 and warnings are handled more correctly and reported in such a way
3963 that it's easy to find one or the other in your output (so that you
3964 can e.g. figure out which of many problems caused COMPILE-FILE to
3966 * changed the severity of several compiler warnings from full WARNING
3967 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3968 compiler note reporting so that it doesn't use the condition system
3969 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3970 COMPILE-FILE command)
3971 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3972 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3973 no longer supported, since their implementation was deeply intertwingled
3974 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3975 * removed bogus "support" for compiler macros named (SETF FOO), and
3976 removed the compiler macro for SETF INFO (but only after making a fool
3977 of myself on the cmucl-imp mailing list by posting a bogus patch for
3978 DEFINE-COMPILER-MACRO..)
3979 * Compiled files containing forms which have side effects on the Lisp
3980 reader (such as DEFPACKAGE forms) are now handled more correctly.
3981 (Compiler queuing of top level lambdas has been suppressed by setting
3982 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3983 * deleted various currently-unused source files, e.g. gengc.lisp. They
3984 may be added back at some point e.g. when porting to other architectures,
3985 but until they are it's distracting to distribute them and to try to
3987 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3988 there were so many of them they're just distractions, and UNCROSS is
3989 known to be able to handle the current sources
3990 * moved PROFILE functionality into TRACE, so that it will be clear
3991 how the wrapping and unwrapping of functions when you profile them
3992 interacts with the wrapping and unwrapping of functions when you
3993 trace them. (Actually, the functionality isn't there yet, but at least
3994 the interface specification is there. Hopefully, the functionality will
3995 arrive with some future maintenance release.)
3996 * removed host-oops.lisp
3997 * changed signature of QUIT function to allow UNIX-CODE argument
3998 * fixed READ-SEQUENCE bug
3999 * tweaked verbose GC output so that it looks more like the progress
4000 output that ANSI specifies for functions like LOAD
4001 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4002 * added <http://sbcl.sourceforge.com> to the banner information printed by