1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.1 relative to sbcl-1.0:
3 * improvement: sb-sprof traces call stacks to an arbitrary depth on
4 x86/x86-64, rather than the previous fixed depth of 8
5 * bug fix: fix handling of non-ascii command-line arguments (thanks to
7 * bug fix: TRACE :ENCAPSULATE NIL (and function end breakpoints)
10 changes in sbcl-1.0 relative to sbcl-0.9.18:
11 * improvement: experimental support for threading on FreeBSD/x86.
12 (thanks to NIIMI Satoshi)
13 * improvement: runtime option --dynamic-space-size can be used
14 to set the size of the dynamic space reserved on startup.
15 * improvement: floating point modes in effect are now saved in
16 core, and restored on startup.
17 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
18 startup, not time since first call to GET-INTERNAL-REAL-TIME.
19 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
20 threads are not running after *SAVE-HOOKS* have run.
21 * improvement: writes to CLOS instance slots are type-checked in code
22 compiled with (SAFETY 3)
23 * improvement: floating-point exception handling on FreeBSD (thanks to
25 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
26 (thanks to Zach Beane)
27 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
29 * improvement: added support for the Shift-JIS external format.
30 (contributed by NIIMI Satoshi)
31 * improvement: callbacks are supported on Linux/PPC. (thanks to
33 * bug fix: compiler bug triggered by a (non-standard) VALUES
34 declaration in a LET* was fixed. (reported by Kaersten Poeck)
35 * bug fix: file compiler no longer confuses validated and already
36 dumped structurres. (reported by Kaersten Poeck)
37 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
38 fill-pointer signals a type-error as required. (thanks to
40 * bug fix: disassemly of funcallable instances works.
41 * bug fix: single stepping on PPC.
42 * bug fix: fix thread-safety problems in the type system (generally
43 manifesting as nonsensical errors like "STRING is a bad type specifier
44 for sequences" or "The value 1 is not of type FIXNUM")
45 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
46 (reported by Josip Gracin)
47 * bug fix: an error is signaled for attempts to displace arrays with
48 incompatible element types (thanks to Mario Mommer)
49 * bug fix: more correct handling of wide characters in debug info
50 (bug reported by Attila Lendvai and fixed by Juho Snellman)
51 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
52 and don't cause extra consing
53 * optimization: MAP and MAP-INTO are significantly faster on vectors
54 whose elements types have been declared.
55 * Improvements to SB-SPROF:
56 ** Support for allocation profiling
57 ** Reduced profiling overhead, especially for long profiling runs
58 * Improvements to the Windows port:
59 ** floating point exceptions are now reported correctly.
60 ** stack exhaustion detection works partially.
61 ** more accurate GET-INTERNAL-REAL-TIME.
62 ** better RUN-PROGRAM behaviour in case of errors in the child
64 ** PROBE-FILE now simplifies pathnames correctly.
65 ** DIRECTORY now works correctly with :WILD-INFERIORS.
66 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
67 1st 1970 (although time timezone data will be unreliable).
68 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
70 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
71 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
72 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
74 * bug fix: two potential GC deadlocks affecting threaded builds.
75 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
76 returns T (reported by Anton Kazennikov)
77 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
78 with non-variable places
79 * bug fix: remove a race condition in the setting of
80 funcallable-instance functions, this should make threaded CLOS
81 code more stable against memory faults.
82 * bug fix: corruption of specials when unbinding is interrupted by an
83 asynchronous unwind (reported by Hannu Koivisto)
84 * improvement: the debugger will now also display local variables that
85 are only used once, for code compiled with a DEBUG optimization quality
88 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
89 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
90 * incompatible change: External-format support for FFI calls. The
91 SB-ALIEN:C-STRING no longer implies an ASCII
92 external-format. Instead the string is subject to external-format
93 conversion. Additionally return values of type SB-ALIEN:C-STRING
94 are by default of type (SIMPLE-ARRAY CHARACTER), not
95 SIMPLE-BASE-STRING. For an alien type that behaves like the old
96 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
97 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
98 * incompatible change: SB-EXT package no longer contains the
99 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
100 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
102 * incompatible change: the single-stepper is no longer available
103 on Alpha, Mips and Sparc platforms.
104 * minor incompatible change: the direct superclasses of
105 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
106 not (STANDARD-OBJECT FUNCTION). This makes the
107 class-precedence-lists of GENERIC-FUNCTION and
108 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
110 * minor incompatible change: the default stream external format on
111 non-unicode SBCL is no longer detected from the locale, but is always
113 * new feature: Add a version of evaluator that uses an interpreter instead
114 of the compiler. EVAL still uses the compiler by default, to switch it
115 to use the interpreter, set the value of the variable
116 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
117 * minor incompatible change: the single-stepper REPL has been merged
118 with the normal debugger (see the "Stepping" heading of the debugger help
119 for more details). The debugger command STEP will no longer switch
120 to the single-stepper REPL.
121 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
122 for a type now works.
123 * bug fix: loading of default sysinit file works. (thanks to Leonid
125 * bug fix: better detection of circularities in the file-compiler.
126 (reported by Marco Monteiro)
127 * bug fix: the CL pathname functions now work with files that have
128 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
129 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
130 systems with Unix98 pty semantics.
131 * bug fix: ASDF-INSTALL will now work with bsd tar.
132 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
134 * bug fix: timers expiring in dead threads no longer cause a
135 type-error (reported by Paul "Nonny Mouse").
136 * bug fix: thanks to more lightweight single-stepper instrumentation,
137 code compiled with (DEBUG 3) will compile and execute significantly faster,
138 and will have more accurate type-inferencing than before
139 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
140 whose bindings are modified
141 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
142 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
143 ** replace dummy implementations of CL:MACHINE-INSTANCE and
144 CL:SOFTWARE-VERSION with proper version
146 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
147 * feature: implemented the READER-METHOD-CLASS and
148 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
149 as specified by AMOP.
150 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
152 * optimization: faster LOGCOUNT implementation on x86 and x86-64
153 (thanks to Lutz Euler)
154 * optimization: hashing of general arrays and vectors has been
155 improved. (reported by Any Fingerhut)
156 * enhancement: SB-INTROSPECT is now able to find definitions of
157 profiled functions. (thanks to Troels Henriksen)
158 * enhancement: compiler-macro expansion applies now to FUNCALL forms
159 as well, allowing compiler-macros for SETF-functions to expand.
160 * enhancement: step-instrumentation no longer wraps known
161 single-value functions with multiple-value context, allowing
162 better type inference.
163 * fixed bug #337: use of MAKE-METHOD in method combination now works
164 even in the presence of user-defined method classes. (reported by
165 Bruno Haible and Pascal Costanza)
166 * fixed bug #339(c): if there are applicable methods not part of any
167 long-form method-combination group, call INVALID-METHOD-ERROR.
168 (reported by Bruno Haible)
169 * fixed bug #361: the :FUNCTION initarg in the protocol for
170 initialization of methods can now be used to override
171 internally-produced optimized functions. (reported by Bruno
173 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
174 system-provided lambda expression no longer cause warnings about
175 unbound #:|pv-table| symbols.
176 * bug fix: improved the handling of type declarations and the
177 detection of violations for keyword arguments with non-constant
179 * bug fix: potentially erronous calls to PATHNAME and
180 MERGE-PATHNAMES were being flushed in some cases.
181 (reported by Richard Kreuter)
182 * bug fix: compiled calls to TYPEP were mishandling obsolete
183 instances. (reported by James Bielman and Attila Lendvai)
184 * bug fix: format strings with ~> without matching ~< no longer
185 trigger an AVER, but signal an understandable error instead.
186 (reported by Antonio Martinez)
187 * bug fix: specifying an output-file without a pathname-type for
188 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
189 of input-file instead of "fasl". (reported by Robert Dodier)
190 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
191 (reported by James Y Knight).
192 * bug fix: compiler-macros-function did not consider the environment
193 argument for shadowing by local functions.
194 * bug fix: compiler-macros expansion was inhibited by local INLINE
196 * bug fix: inline expansions of known functions were subject to
197 step-instrumentation in high debug policies, leading to problems
199 * bug fix: compiler failed to differentiate between different CONS
201 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
202 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
203 * bug fix: #368: incorrect use of expressed vs. upgraded array
205 * bug fix: #306a: more precise unions of array types.
206 * thread-safety improvements:
207 ** CONDITION-WAIT could return early on Linux, if the thread was
208 interrupted and subsequently continued with SIGCONT.
209 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
211 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
212 * added support for the ucs-2 external format. (contributed by Ivan
214 * minor incompatible change: pretty printing of objects of type
215 (cons symbol) is, in the default pprint-dispatch-table, now
216 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
218 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
219 called later in a class's lifetime, possibly as late as when the
220 first instance of the class is created. Previously,
221 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
222 class became finalizeable.
223 * fixed bug: FILE-POSITION sometimes returned inconsistent results
224 for multibyte external-format streams. (thanks to "vbzoli")
225 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
226 with :ALLOCATION :CLASS inherited from superclasses of the
228 * fixed bug: anonymous classes can now be created using the :NAME
229 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
230 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
231 * fixed bug: core-files saved with :EXECUTABLE T can again be
232 executed when SBCL_HOME isn't set. (reported by James Knight)
233 * fixed bug: toplevel LOCALLY forms with declarations could
234 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
235 * fixed bug: printing from several different threads using different
236 values of *print-case* could cause invalid output, due to
237 some internal special variables of the printer not being bound
238 thread-locally (reported by Max Mikhanosha)
239 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
240 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
241 (reported by Pascal Costanza)
242 * fixed bug: FINALIZE-INHERITANCE is called from
243 REINITIALIZE-INSTANCE on classes when the class has previously
244 been finalized, as required by AMOP.
245 * minor code generation optimizations:
246 ** better register allocation in CLOS dispatching functions
247 ** overflow detection when coercing signed bytes to fixnums on x86-64
248 ** is now implemented with one IMUL instruction instead of three shifts
249 ** more efficient bit-vector access on x86 and x86-64
250 ** more efficient access to raw structure slots on x86 and x86-64
251 * fixed some bugs revealed by Paul Dietz' test suite:
252 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
255 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
256 * feature: thread support on Solaris/x86, and experimental thread support
258 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
260 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
261 default method for SLOT-UNBOUND.
262 * minor incompatible change: prevent the user from specializing the
263 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
264 somewhat counter to the intent of the protocol, I (CSR) think, and
265 additionally it just doesn't work in SBCL as currently
266 implemented, thanks to optimizations (that are always valid for
267 the other three SLOT-VALUEish functions, but not for the setter).
268 * minor incompatible change: the :SB-LDB feature is now enabled by
269 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
270 the low-level debugger.
271 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
272 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
273 down to an SB-SYS:FD-STREAM, allowing direct communication in
274 more cases. Temporary files and pipes are still used as a fallback
276 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
277 on the non-reentrant C functions localtime(3) and gmtime(3).
278 * bug fix: global optimization policy was not visible in LOCALLY and
280 * bug fix: class objects can be used as specializers in methods.
281 (reported by Pascal Costanza)
282 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
283 with a type error. (reported by blitz_ on #lisp)
284 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
285 failed with a type error. (reported by Pascal Bourguignon)
286 * bug fix: merging pathnames against defaults with :DIRECTORY
287 starting with '(:RELATIVE :BACK) should preserve the :BACK.
288 (reported by James Y Knight)
289 * bug fix: saving large (>2GB) cores on x86-64 now works
290 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
291 constant index and a (simple-array (signed-byte 32)) array
292 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
293 error (patch by Robert J. Macomber)
294 * feature: TIME now displays the amount of run-time spent in GC
295 * bug fix: The debugger now does a better job of respecting
296 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
297 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
298 workaround for bug 403.)
299 * bug fix: floating point exception handling now works on FreeBSD
300 (thanks to NIIMI Satoshi)
301 * fixed some bugs revealed by Paul Dietz' test suite:
302 ** MISC.641: LET-conversion were not supposed to work in late
304 * improvements to the Win32/x86 port:
305 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
306 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
308 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
309 * new feature: source path information is generated for macro-expansion
310 errors for use in IDE's like Slime (thanks to Helmut Eller)
311 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
312 * bug fix: compiler does not loop forever on an invalid type in
314 * improvement: compilation of most CLOS applications is significantly
316 * optimization: added a limited bytecode compiler for simple toplevel
317 forms, speeding up compilation and FASL loading
318 * bug fix: the statistical profiler now properly distinguishes anonymous
321 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
322 * minor incompatible change: in sbcl-0.9.11 (but not earlier
323 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
324 into subclasses of the finalized class. Now user calls to
325 FINALIZE-INHERITANCE finalize just the one class, and calls by the
326 system return before any subclasses are finalized.
327 * minor incompatible change: The reader no longer ignores errors
328 regarding non-existent packages in #+ and #- feature tests.
329 * new feature: command line options --no-sysinit, --no-userinit to
330 inhibit loading the corresponding init files
331 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
332 for sending data through UDP sockets (thanks to François-René Rideau)
333 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
334 error is signalled instead (thanks to François-René Rideau)
335 * minor incompatible change: Error signalling behaviour of lexical
336 operations violating package locks has changed slightly. Refer to
337 documentation on package locks for details.
338 * enhancement: EVAL can process IF-expressions without resorting to the
340 * bug fix: Some operations involving SIN, COS, or TAN weren't being
341 constant-folded properly due to missing out-of-line functions.
342 (reported by Mika Pihlajamäki)
343 * bug fix: LISTEN sometimes returned T even in cases where no data was
344 immediately available from the stream
345 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
346 were reversed. (reported by Levente Mészáros)
347 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
348 when given a vector displaced to an adjustable array. (reported
350 * bug fix: derivation of float boundaries from numbers outside the
351 appropriate float range (reported by John Wiseman)
352 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
354 * bug fix: precomputing cache entries for generic functions with
355 some subclasses of specializers as yet invalid does not attempt to
356 fill a cache line with a negative offset. (reported by Levente
358 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
359 allow condition class objects as arguments to DOCUMENTATION and
360 (SETF DOCUMENTATION); only find and set documentation for
361 structure names for the STRUCTURE doc-type. (suggested by Gary
363 * improvements to the Win32/x86 port:
364 ** (user-homedir-pathname) and default initialization file
365 locations now know about the user's "Documents and Settings"
366 directory (thanks to Yaroslav Kavenchuk)
367 ** run-program is implemented (thanks to Mike Thomas)
368 ** sockets support (thanks to Timothy Ritchey)
369 ** better backtrace support (thanks to Alastair Bridgewater)
370 ** sb-grovel supported
371 ** asdf-install and sb-posix work somewhat
372 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
373 * improvements to the Solaris/x86 port:
374 ** works on Solaris 11/Solaris Express
375 ** floating-point exception handling support
376 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
377 * fixed some bugs revealed by Paul Dietz' test suite:
378 ** REMOVE-METHOD returns its generic function argument even when
379 the method is not one of the generic functions' methods.
380 ** objects declared as MEMBER types can be admissible arguments to
382 ** printing characters should simply be printed by the FORMAT ~:C
384 ** compiler failure when compiling functions with hairy constant
385 defaults for optional parameters.
386 ** compiler produces wrong code when MAYBE-INLINE-expanding a
387 function, which is already optimized.
389 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
390 * new platform: experimental support for SBCL x86/Darwin, including
391 MacOS X 10.4.5 on Intel.
392 * new feature: Unicode character names are now known to the system
393 (through CHAR-NAME and NAME-CHAR).
394 * new feature: the filesystem location of SBCL's core file is
395 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
396 * minor incompatible change: the contrib modules SB-POSIX and
397 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
398 this change is to make it easier to distribute
399 location-independent binaries.
400 * bug fix: as implied by AMOP, standardized classes no longer have
401 slots named by external symbols of public packages. (reported by
403 * optimization: calling functions via constant symbols -- as in
404 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
405 via the function object as in (FUNCALL #'FOO).
406 * optimization: on x86 and x86-64, the calling convention for the
407 case of dealing with an unknown number of values has been altered
408 to be friendlier to the prediction heuristics implemented,
409 particularly with reference to CALL and RET pairing. (thanks to
410 Alastair Bridgewater)
411 * optimization: on x86, the code for access to thread-local symbol
412 values and binding/unbinding of thread-local symbols is smaller.
413 (thanks to Alastair Bridgewater)
414 * enhancement: CONSTANTP is now able to determine constantness of
415 more complex forms, including calls to constant-foldable standardized
416 functions and some special forms beyond QUOTE.
417 * fixed bug: occasional GC crashes on Solaris/x86
418 * optimization: x86-64 supports stack allocation of results of simple
419 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
420 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
421 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
422 (thanks to James Knight)
423 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
424 by the standard (thanks to Stephen Compall)
426 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
427 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
428 be used for bundling the runtime and the core file into one
429 executable binary. This feature is not currently supported on all SBCL
430 platforms. (thanks to James Bielman and NIIMI Satoshi)
431 * new feature: a generational or ephemeral garbage collector is now
432 the default on the PowerPC platform (both Linux and Darwin). The
433 old Cheney (stop and copy) collector is a build-time option.
434 (thanks to Cyrus Harmon, after Raymond Toy)
435 * minor incompatible change: the method by which SBCL finds its
436 contributed modules has changed; it no longer relies on symbolic
437 links from an $SBCL_HOME/systems directory, but searches directly
438 in the subdirectories of $SBCL_HOME.
439 * enhancement: the dynamic heap size on the Linux/PPC platform is
440 markedly larger, even using the older Cheney garbage collector.
441 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
442 certain complicated string types. (reported by Gary King)
443 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
444 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
445 * fixed bug: DOCUMENTATION on structures no longer signals an
446 error if no structure type of the right name exists. (reported by
448 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
449 applicable non-standard primary or :AROUND method on
450 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
451 no longer cause unbound variable errors. (reported by Kevin Reid)
452 * fixed bug: metacircle resolution in cases where methods have slots
453 added before the slots from STANDARD-METHOD. (reported by Jean
455 * fixed bug: the Power PC assembly code for calling into lisp
456 sometimes computed its return address wrongly (depending on the
457 15th bit of the address, fixed at link time). This used to
458 manifest itself in a segmentation violation while building PCL.
459 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
460 many others over the years)
461 * fixed bug: ROOM no longer reports silly numbers for stack usage if
462 the stack pointer (treated as a signed integer) is negative.
463 (thanks to Peter van Eynde)
465 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
466 * new platform: experimental support for the Windows operating
467 system has been added. (thanks to Alastair Bridgewater)
468 * fixed several bugs in and robustified the PPC FFI (including
469 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
470 * bug fix: allow non-simple string symbol names (reported by Paul
472 * bug fix: interrupt handling on NetBSD (thanks to Richard M
474 * bug fix: saving a core corrupted callbacks on x86/x86-64
475 * bug fix: closed a loophole in metacircularity detection and
476 grounding in the PCL implementation of CLOS.
477 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
478 (reported by tomppa on #lisp)
479 * bug fix: clear the direction flag on Lisp -> C transitions, as
480 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
481 (reported by Andrej Grozin and Hendrik Maryns)
482 * optimization: major improvements to GC efficiency on GENCGC platforms
483 * optimization: faster implementation of EQUAL
484 * optimization: emit more efficient opcodes for some common
485 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
486 * optimization: several other minor code-generation improvements on x86-64
487 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
489 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
490 * minor incompatible change: (SETF CLASS-NAME) and (SETF
491 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
492 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
493 by ANSI as a generic function, and for consistency (SETF
494 GENERIC-FUNCTION-NAME) is treated likewise.
495 * fixed bug #233.b: make constraint propagation notice when a variable
496 value is changed after it is referenced but before it is used
497 * fixed bug #296: no more arbitrary behaviour from filenames with
498 odd characters as --load arguments. (reported by Adam Warner)
499 * fixed bug #390: :CHARACTER-SET pathname components now work as
500 expected. (reported by Tim Daly Jr)
501 * fixed bug #391: complicated :TYPE intersections in slot
502 definitions no longer cause an error in PCL internals.
503 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
504 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
505 returning the number of octets which would be written to the
506 file-stream. (thanks to Robert J. Macomber)
507 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
508 arguments to the internal stream functions. (thanks to David
510 * bug fix: stack exhaustion is now handled on the x86/SunOS
511 platform. (thanks to Daisuke Homma)
512 * bug fix: more accurate ROOM results on GENCGC platforms
513 * optimization: improved type inference for arithmetic-for
514 index variables in LOOP
515 * optimization: faster floating-point SQRT on x86-64
516 * fixed some bugs revealed by Paul Dietz' test suite:
517 ** DOCUMENTATION returns NIL instead of "" for method combinations
518 that don't have a docstring
520 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
521 * minor incompatible change: (SETF CLASS-NAME) and (SETF
522 GENERIC-FUNCTION-NAME) are no longer generic functions, and
523 therefore are not customizeable by user code (as seems to be at
524 least permitted and maybe required by AMOP). As a consolation,
525 however, the SBCL implementation of these functions now calls
526 REINITIALIZE-INSTANCE as specified by AMOP.
527 * bug fix: slot-definition documentation is propagated as per ANSI
528 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
529 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
530 argument, as do eslotd initialization methods. (from Pascal
531 Costanza's "Closer" project)
532 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
533 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
535 * bug fix: it is now possible to have more than one subclass of
536 STANDARD-GENERIC-FUNCTION without causing stack overflow.
537 (reported by Bruno Haible, Pascal Costanza and others)
538 * bug fix: the dependent update protocol now works for generic
539 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
541 * bug fix: condition-class instances corresponding to
542 DEFINE-CONDITION forms are now created eagerly. (reported by
543 Kalle Olavi Niemitalo on comp.lang.lisp)
544 * bug fix: floating point printing is more accurate in some
545 circumstances. (thanks to Simon Alexander)
546 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
547 merged with *DEFAULT-PATHNAME-DEFAULTS*.
548 * bug fix: callbacks on OS X now preserve stack-alignment.
549 (thanks to Cyrus Harmon)
550 * enhancement: the x86-64 disassembler is much better at
551 disassembling SSE instructions. (thanks to Lutz Euler)
552 * enhancement: CHECK-TYPE on a variable now allows type-inference
553 (thanks to Vincent Arkesteijn)
554 * optimization: improved performance of EUC-JP external format.
555 (thanks to NIIMI Satoshi)
556 * optimization: performance improvements to IO on file streams of
557 :ELEMENT-TYPE CHARACTER
558 * optimization: much faster memory allocation on x86-64
559 * optimization: faster garbage collections (latency and throughput)
561 * optimization: faster bignum subtraction and division on x86 and x86-64
562 (thanks to Lutz Euler)
563 * optimization: more accurate type-inference for loops that use a
564 floating point index variable or a negative step.
566 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
567 * bug fix: add a workaround to SBCL looping infinitely at startup on
568 Linux kernels with apparently buggy implementations of personality().
569 (thanks to Svein Ove Aas)
570 * bug fix: Unicode symbols are correctly printed in LDB backtraces
571 (thanks to David Lichteblau)
572 * bug fix: local bindings shadow global symbol macros for
573 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
574 * optimization: non-open coded uses of numeric comparison operators
575 (e.g. >) no longer cons when called with more than one parameter
576 on platforms supporting dynamic-extent allocation.
577 * enhancement: saving cores with foreign code loaded is now supported
578 on MIPS/Linux in addition to the previously supported platforms.
579 * bug fix: FIND-RESTART now tests for activity, not applicability when given
580 a restart object as identifier. (reported by Helmut Eller for CMUCL)
581 * bug fix: division by zero in sb-sprof when no samples were collected
582 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
583 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
584 (thanks to Alexander Kjeldaas)
585 * new feature: ignoring the return values of destructive functions
586 like NREVERSE signals a compile-time style-warning.
587 (thanks to Kevin Reid)
588 * new feature: the alignment of alien structure fields can be
589 explicitly specified. (thanks to Cyrus Harmon)
590 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
591 and dump core on SIGQUIT
593 ** incompatible change: threads do not inherit values of specials
594 from their parents (see manual)
595 ** bug fix: threads stacks belonging to dead threads are freed by the
596 next exiting thread, no need to gc to collect thread stacks anymore
597 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
598 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
599 ** bug fix: dynamic variable and thread start related gc lossage
600 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
602 * fixed some bugs revealed by Paul Dietz' test suite:
603 ** SUBTYPEP is slightly more accurate on heinously complicated
604 CONS types where some of the members have uncertain (in the
605 NIL, NIL sense) type relationships to each other.
607 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
608 no longer purified unless :PURIFY T is explicitly specified.
609 ** Non-purified cores are significantly smaller than before
611 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
612 * new feature: timers based on Zach Beane's excellent timer package
613 * added support for the following external formats: koi8-u,
614 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
615 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
616 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
617 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
618 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
619 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
620 * incompatible change: a threaded SBCL will no longer revert to
621 non-threaded mode on non-NPTL systems, but refuse to start entirely.
622 * bug fix: interrupts are disabled until startup is complete; no
623 more sigsegvs when receiving a signal to soon
624 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
626 * bug fix: add a workaround for the memory randomization features in
627 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
628 workaround will only be in effect on systems with the proc filesystem
630 * bug fix: printing objects of type HASH-TABLE signals a
631 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
633 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
634 have been running for over 50 days. (reported by Gilbert Baumann)
635 * bug fix: the logic for getting names of functions gets less
636 confused when confronded with alternate-metaclass
637 funcallable-instances. (reported by Cyrus Harmon)
638 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
639 consistent, even on internal alternate-metaclass objects.
640 * bug fix: flush closure information collected by physical
641 environment analysis prepass before the main pass. (bug reported
643 * bug fix: compiler pack phase does not modify a hashtable, which is
644 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
645 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
646 may be expanded. (reported by Alexander <malishev> on c.l.l)
647 * bug fix: no more style-warnings when compiling code using LAST
648 under high-SPEED low-SPACE settings. (reported by David Wallin
650 * fixed some incompatibilities between SBCL's MOP and the MOP
652 ** the METAOBJECT class is now implemented;
653 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
654 STANDARD-OBJECT, as required;
655 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
656 now compatible (as required by VALIDATE-SUPERCLASS); there
657 remains a consistency requirement over the presence or absence
658 of the FUNCTION class in the superclass list of finalized
659 classes; see the manual for more details;
660 ** the :AROUND method for COMPUTE-SLOTS on
661 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
662 requested slot ordering.
664 ** bug fix: parent thread now can be gc'ed even with a live
666 ** bug fix: binding a special with PROGV to no value is not
667 the same as that symbol not having been bound (thanks to
669 ** bug fix: binding specials is thread safe (thanks to
671 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
672 and SB-SPROF do not die with 'no handler for signal XX in
673 interrupt_handle_now(..)' anymore
674 ** bug fix: WITH-TIMEOUT works with multiple running threads
675 * fixed some bugs revealed by Paul Dietz' test suite:
676 ** ENSURE-GENERIC-FUNCTION should take a method class object for
677 the :method-class keyword argument.
679 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
680 * new port: the Solaris operating system on x86 processors is now
681 mostly supported, though some rough edges in the environment
682 remain. (thanks to Daisuke Homma)
683 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
684 space for its heap. (thanks to Thiemo Seufer)
685 * optimization: SBCL on MIPS platforms now supports dynamic-extent
686 closures. (thanks to Thiemo Seufer)
687 * minor incompatible change: eof selects abort in the debugger.
688 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
689 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
691 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
692 deprecated, and will go away in a future revision of SBCL.
693 * minor incompatible change: GC-ON and GC-OFF are no longer
694 implemented with a counter, it does not matter how many times gc
695 is switched on or off
696 * bug fix: discriminating functions for generic function classes
697 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
698 make invalid assumptions about method precedence order. (reported
700 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
701 platforms. (thanks to Thiemo Seufer)
702 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
703 * bug fix: correct pathname printing with printer escaping is on.
704 (thanks to Kevin Reid)
705 * bug fix: complex VOP definitions in "user-space" no longer trigger
706 package locks. (reported by Zach Beane)
707 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
708 for funcallable-instance functions, and is no different from
710 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
711 characters in the current readtable. (reported by Nicholas Neuss)
712 * bug fix: SUBTYPEP on various CONS types returns more a more
713 accurate acknowledgment of its certainty.
714 * optimizations: REMOVE-DUPLICATES now runs in linear time on
715 lists in some cases. This partially fixes bug 384.
716 * flush all standard streams before prompting in the REPL and the
718 * bug fix: signal handling and triggering gc do not conflict
719 directly anymore, in particular a high frequency sb-sprof does
720 not prevent gc from running
721 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
722 approximation for timezone and DST information between the
723 universal time epoch and the smallest negative 32-bit time_t.
724 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
725 year 1899 so long as the encoded time is non-negative.
726 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
727 the correct answer for non-constant rotate amounts when performing
728 an inline 32-bit rotation.
730 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
731 there is only one thread in the session
732 ** bug fix: memory leak for streams created in one thread and
733 written to in another
734 ** bug fix: lockup when compiled with gcc4
735 ** bug fix: race that allows the gc to be triggered when gc is
737 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
738 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
740 ** bug fix: finished threads are gc'ed properly
741 * fixed some bugs revealed by Paul Dietz' test suite:
742 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
743 has arguments with a different set of applicable methods from
744 the orignal arguments.
745 ** The type error thrown by MAP now has a correct expected-type
747 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
748 name a compiled function.
749 ** Three MISC tests where a large bignum was improperly coerced to
750 a float that couldn't represent that bignum during type
751 derivation were fixed.
752 ** SUBTYPEP can now handle types involving the negation of a
753 list-form FUNCTION type.
754 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
755 as (COMPLEX (AND RATIO (NOT FIXNUM))).
756 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
758 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
759 * New feature: Experimental support for bivalent streams: streams
760 opened with :element-type :default now allow character and binary
761 (unsigned-byte 8) I/O
762 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
763 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
765 * bug fix: no more highly sporadic "couldn't check whether ~S is
766 readable" when reading a stream and an interrupt hits in the middle
767 of a select system call
768 * compiler better recognizes complex arrays (reported by Eduardo
770 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
771 for variables on SBCL built with linkage-tables. (reported by Luis
773 * various error reporting improvements.
774 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
775 (Thanks to Thiemo Seufer)
776 * bug fix: nested reader invokations work correctly
777 * bug fix: it is possible to have simultaneous references to foreign
778 code and foreign data with the same name.
780 ** added x86-64 support
781 ** incompatible change: the threading api now works with thread
782 objects instead of thread ids
783 ** bug fix: threads are protected from signals and interruption when
784 starting up or going down
785 ** bug fix: a race where an exiting thread could lose its stack to gc
786 ** bug fix: don't halt on infinite error in threads if possible
787 ** fixed numerous gc deadlocks introduced in the pthread merge
788 ** bug fix: fixed thread safety issues in read and print
789 ** bug fix: debugger doesn't hang on session lock if interrupted at
790 an inappropriate moment
791 ** bug fix: run-program is now thread safe(r)
792 ** bug fix: inner with-recursive-lock no longer releases the mutex
793 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
794 * fixed some bugs revealed by Paul Dietz' test suite:
795 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
796 EXPECTED-TYPE slots filled.
797 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
799 ** MAKE-CONDITION accepts classes as type-designators.
800 ** COMPILE may never return NIL.
801 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
802 range before calling Unix time functions
804 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
805 * numerous signal handling fixes to increase stability
806 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
807 * minor incompatible change: we now correctly canonize default
808 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
809 as per the MOP, rather than the historical (INITARG INITFUNCTION
810 INITFORM). (reported by Bruno Haible)
811 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
812 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
813 for more information.
814 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
815 pathname is a directory pathname.
816 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
817 * dynamic space size on PPC has been increased to 768Mb. (thanks to
819 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
820 :METACLASS argument in addition to a class name. (reported by
821 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
822 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
823 (reported by Bruno Haible; patch from Gerd Moellmann)
824 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
826 * bug fix: more cleanups to the floating point exception handling on
827 x86-64 (thanks to James Knight)
828 * bug fix: the compiler does not try to constant fold calls of
829 COERCE and BIT-* functions when they return freshly constructed
830 objects. (reported by Nikodemus Siivola and Paul Dietz)
831 * optimization: improved the MIPS versions of generic (in the
832 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
833 * optimization: direct conversion of (unsigned-byte 32) to floats on
834 the PowerPC platform.
835 * optimization: structure instances with raw slots now use less
836 memory, and probably show better memory locality. (thanks to
838 * optimization: DYNAMIC-EXTENT declarations for lists and closures
839 are treated as requests for stack allocation on the x86-64,
840 Alpha-32, PPC and SPARC platforms.
841 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
842 the wrong file descriptor; implementation of SOCKET-OPEN-P.
843 (thanks to Tony Martinez)
845 ** gcing a dead thread can no longer lead to lockups
846 ** threads block signals until they are set up properly
847 ** errno is no longer shared by threads
848 ** interrupt-thread restores the eflags register on x86
849 ** fixed some lockups due to gc/thread interaction
850 * fixed some bugs revealed by Paul Dietz' test suite:
851 ** invalid dotted lists no longer raise a read error when
853 ** signal an error if a symbol that names a declaration is used
854 as the name of a type, or vice versa
855 ** allow using the (declare (typespec var*)) abbreviation for
856 (declare (type typespec var*)) with all type specifiers
857 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
858 ** always return NIL from PPRINT-POP when OBJECT is NIL
859 ** don't signal errors when pretty-printing malformed LABELS,
860 FLET or MACROLET forms
861 ** declarations in a DOLIST body are also in scope for the
863 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
865 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
868 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
869 * fixed cross-compiler leakages that prevented building a 32-bit
870 target with a 64-bit host compiler.
871 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
872 opened with :IF-EXISTS :OVERWRITE.
873 * fixed bug 281: error for an invalid qualifier in a short-form method
874 combination method is not signalled until the faulty method is called.
875 * bug fix: iteration variable type inferrer failed to deal with open
876 intervals. (reported by Alan Shields)
877 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
878 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
879 or more runtime options were provided to the sbcl binary.
880 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
881 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
882 * optimization: there's now a fast-path for fixnum arguments in the
883 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
884 * optimization: the code generated on x86-64 is more compact thanks
885 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
886 * bug fix: floating-point exception handling now partly works on
887 x86-64. (Thanks to James Knight)
888 * improvement to the MIPS backend from Thiemo Seufer: C-style
889 64-bit long long arguments and return values to alien functions
891 * fixed some bugs revealed by Paul Dietz' test suite:
892 ** the type-error signalled from WARN has a filled-in DATUM slot.
893 ** the type-error required when a stream is not associated with
894 a file has the stream as its datum.
895 ** type-errors on single-floats on x86-64 no longer have
896 :INVALID-OBJECT as the datum
897 ** the type-errors signalled for invalid function names now have
898 a correct expected type
899 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
900 for structures defined with a :TYPE. Documentation strings for
901 typed structures are no longer immediately discarded
902 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
903 broadcast streams and synonym streams. FILE-LENGTH now also works
904 on broadcast streams.
906 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
907 * incompatible change: the --noprogrammer option, deprecated since
908 version 0.7.5, has been removed. Please use the equivalent
909 --disable-debugger option instead.
910 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
912 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
913 inoperational for a while now) has been completely removed.
914 * null lexical environments are now printed as #<NULL-LEXENV>,
915 significantly reducing the amount of clutter in typical backtraces.
916 * documentation on weak pointers, finalization, and after GC hooks
917 has been added to the manual.
918 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
919 as other specialized array types, is much faster. SUBSEQ and
920 COPY-SEQ on such arrays have also been sped up.
921 * optimization: EQL is now more efficient when at least other argument
922 is known to be of type (OR FIXNUM (NOT NUMBER)).
923 * fixed bug: the runtime is now less vulnerable to changes in the
924 size of the SBCL object on OS X, and virtual memory is reserved for
925 all spaces that need to be at a fixed address.
926 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
927 * fixed bug: finalizers and after GC hooks that cause consing are now
929 * fixed bug: compiler error messages and summaries are now printed to
930 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
931 * fixed inference of the upper bound of an iteration variable.
932 (reported by Rajat Datta).
933 * fixed bug 211e: calling local functions with duplicated constant
934 keyword argument no longer causes a bogus style warning about an
936 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
937 ftype declaration to be disregarded. (reported by Dave Roberts)
938 * fixed bug 373: caused by erronous compilation of references to alien
939 variables in the runtime on ppc/darwin.
940 * fixed bug 376: CONJUGATE type deriver.
941 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
942 (reported by Baughn on #lisp)
943 * a cross-compiler bug on non-x86 platforms has been identified and
944 fixed. (thanks to Bruno Haible)
945 * improvements to the MIPS runtime code for increased stability.
946 (thanks to Thiemo Seufer)
947 * increased the maximimum compact environment size to allow
948 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
949 * partial workaround for bug 135: don't name the function we're
950 calling for hairy cases of EVAL, so as not to accumulate one environment
951 entry per call to EVAL. (reported by Kevin Reid)
952 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
953 * optimization: type testing for non-vector arrays should be faster.
954 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
956 * bug fix: setting 31st element of a bit vector to zero did not work
958 * bug fix: redefining a class definition which failed due to a
959 previous accessor / function clash now works (but see BUGS entry
960 #380 for more problems in this area). (thanks to Zach Beane)
961 * the long form of DEFINE-METHOD-COMBINATION disables method group
962 checking when given a single method group with pattern *. (thanks
964 * on x86 compiler supports stack allocation of results of simple
965 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
966 * support for building and running on Mac OS X 10.4 "Tiger" has been added
967 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
968 * fixed some bugs related to Unicode integration:
969 ** the restarts for recovering from input and output encoding
970 errors only appear when there is in fact such an error to
972 * fixed some bugs revealed by Paul Dietz' test suite:
973 ** MISC.549 and similar: late transformation of unsafe type
974 assertions into derived types caused unexpected code
976 ** SCALE-FLOAT type deriver is less wrong.
977 ** type derivers for EXP, LOG and similar functions compute result
978 types for complex arguments better.
979 ** (MISC.563) CONJUGATE type deriver works for very restricted
981 ** out-of-line type testers for character strings are available.
982 ** EQUAL compiler transform understands specialness of objects
984 ** accessing double-floats stored on the stack now works on x86-64.
985 ** debugger internals could sometimes create invalid lispobjs,
986 resulting in GC crashes.
987 ** MISC.548: type check weakening can convert required type into
989 ** initialization forms for bindings are not in scope of free special
992 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
993 * incompatible change: thread support for non-NPTL systems has
994 been removed - locking is buggy and unreliable. A threaded
995 SBCL build will now warn at startup and refuse to create
996 new threads, unless futex support is detected in the kernel
997 * incompatible change: the top level REPL now has only an ABORT
998 restart associated with it, not TOPLEVEL and ABORT as it used to.
999 TOP and TOPLEVEL are now available as debugger commands for
1000 returning to the top level.
1001 * incompatible change: forms evaluated in the REPL now use the
1002 global optimization policy.
1003 * incompatible change: user- and system-initialization files are
1004 no longer processed with LOAD, but by READ and EVAL; hence the
1005 global optimization policy, startup package, readtable, etc,
1007 * The .fasl file format number has been incremented because of
1008 various incompatible changes.
1009 * internal entry point details and argument counts no longer appear
1010 in backtraces unless explicitly requested by setting
1011 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1012 * built-in and standard functions no longer have names like "top
1013 level local call to FOO".
1014 * fixed bug 32: functions defined in non-null lexical environments
1015 now have more legible printed representation
1016 * fixed bug 33: functions defined in non-null lexical environemnts
1017 are now more amenable to inspection by INSPECT.
1018 * workaround for bug 354: XEPs no longer appear in backtraces unless
1019 explicitly requested.
1020 * fixed bug: receiving the signal which results from attempting to
1021 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1022 SIGBUS on other *BSDs) on architectures where the C stack is also
1023 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1024 to debugger confusion or wild instability. (reported by Cheuksan
1026 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1027 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1028 (reported by Lutz Euler)
1029 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1030 compiled code. (reported by Paul Dietz)
1031 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1032 specialized parameter to a method to something that is not TYPEP
1033 the specializer is now possible.
1034 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1035 face of package deletion.
1036 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1037 pathname. (reported by David Tolpin on c.l.l)
1038 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1039 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1040 STRING-STREAM and FILE-STREAM, respectively.
1041 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1042 than previously; in addition, its test suite is now run on build.
1043 (thanks to Luke Gorrie)
1044 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1045 * optimization: added a immediate representation for single-floats
1047 * fixed some bugs revealed by Paul Dietz' test suite:
1048 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1049 ** attempting to create a package with a colliding nickname causes
1050 correctable errors to be signalled.
1051 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1052 ** several x86-64 backend bugs related to sign-extension of immediate
1055 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1056 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1057 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1059 * fixed loading of multiply forward-referenced layouts.
1060 (thanks to Cheuksan Wang)
1061 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1062 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1063 (see COMPILE-FILE documentation for details of :PRINT option)
1064 * fixed bugs 19 and 317: fixed-format floating point printing is
1065 more accurate. This also fixes a bug reported by Adam Warner
1066 related to the ~@F format directive.
1067 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1069 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1070 dispatch table if the from-char is a dispatch macro character.
1071 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1072 by the wordsize no longer ignore the last word. (reported by Lutz
1074 * fixed bug in type checking of dynamic-extent variables. (reported
1076 * optimization: sequence traversal functions use their freedom to
1077 coerce function designators to functions.
1078 * optimization: code with many calls to CLOS methods specialized on
1079 CLOS classes has had redundant type checks removed.
1080 * optimization: added declarations to speed up operations that access
1081 the internal character database (for example STRING-UPCASE)
1082 * optimization: comparison operations between floats and sufficiently small
1083 fixnums no longer create extra rationals
1084 * fixed some bugs related to Unicode integration:
1085 ** portions of multibyte characters at the end of buffers for
1086 character-based file input are correctly transferred to the
1087 start of the buffer at the next read.
1088 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1089 passing it through to OPEN.
1090 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1091 argument given to internal calls to OPEN.
1092 ** the built-in comment readers (introduced by character sequences
1093 ";" and "#|") are more forgiving to encoding errors; they will
1094 STYLE-WARN and then attempt to resync the stream at a character
1095 boundary. (thanks to Teemu Kalvas)
1096 * fixed some bugs in the x86-64 port:
1097 ** Negative short int return values from c-calls are sign-extended
1099 ** The stack is aligned to 16-bytes for c-calls, as required by
1100 the ABI. (thanks to Cheuksan Wang)
1101 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1102 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1103 for enums. (thanks to Vincent Arkesteijn)
1104 ** Multiple small optimizations and bugfixes for floating point
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** Space, Tab, Linefeed, Return and Page have the invalid
1108 secondary constituent character trait.
1109 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1111 ** WITH-INPUT-FROM-STRING should only update the index place on
1113 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1114 works more reliably.
1115 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1116 with one-dimensional arrays with fill pointers.
1117 ** TYPE-OF failed on a complex with an integer realpart and a
1119 ** compiler failure during type inference for the code of form
1120 (IF (EQL X (THE ...)) ...) (MISC.535).
1122 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1123 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1124 platform. The port supports 61-bit fixnums, large memory spaces
1125 and reloading shared object files.
1126 * enhancement: saving cores with foreign code loaded is now
1127 supported on ppc/Darwin in addition to the previously supported
1129 * enhancement: the statistical profiler now walks deeper into the
1130 call stack for more meaningful call-graphs and accrued time
1131 reports (x86/x86-64 only). It also now reports time spent in
1133 * enhancement: it is now possible to trace most individual methods
1134 of a generic function in addition to tracing the generic function
1136 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1138 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1139 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1140 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1141 produces an error. (thanks to Vincent Arkesteijn)
1142 * bug fix: NAMESTRING on pathnames with :WILD components in their
1143 directories works correctly. (thanks to Artem V. Andreev)
1144 * fixed bug 125: compiler preserves identity of closures. (reported
1146 * bug fixed: functions with &REST arguments sometimes failed with
1147 "Undefined function" when compiled with (DEBUG 3). (reported by
1149 * bug fix: overflow during compiling of setting element of a bit
1150 vector with constant index and value. (reported by Timmy Douglas)
1151 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1152 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1153 * on x86 compiler partially supports stack allocation of dynamic-extent
1155 * GO and RETURN-FROM do not check the extent of their exit points
1156 when compiled with SAFETY 0.
1157 * fixed some bugs related to Unicode integration:
1158 ** encoding and decoding errors are now much more robustly
1159 handled; it should now be possible to recover even from invalid
1160 input or output to the terminal. (thanks to Teemu Kalvas)
1161 ** provided a first cut at implementing STRING-TO-OCTETS and
1162 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1163 ** altered the SB-MD5 contributed module slightly, changing the
1164 interface just enough for it to be supportable for builds where
1165 lisp characters are not eight bits.
1166 * fixed some bugs revealed by Paul Dietz' test suite:
1167 ** the FORMATTER-generated functions for ~V[ conditionals require
1168 the correct number of arguments.
1169 ** READ-FROM-STRING returns the mandated second value when applied
1170 to displaced strings.
1171 ** the #\Rubout and #\Backspace characters are treated as invalid
1172 constituent characters by the tokenizer.
1174 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1175 * new feature: reloading changed shared object files with
1176 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1177 * new feature: references to foreign variables and functions
1178 can now be compiled and loaded before the corresponding shared
1179 object file is loaded, as long as the foreign definitions are
1180 available at runtime.
1181 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1182 supported. (thanks to Dan Debertin)
1183 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1184 just keywords. (thanks to Vincent Arkesteijn)
1185 * fixed bug #331: structure-class instances corresponding to
1186 DEFSTRUCT forms are now created eagerly.
1187 * fixed bug #345: backtraces from calls to undefined functions work
1188 on x86 as well. Related bug #61 is now also partially fixed on x86
1189 (backtraces from throws to unknown catch tags.)
1190 * bug fix: lambda-list parsing is now stricter vrt. order and number
1191 of lambda-list keywords.
1192 * bug fix: as specified by AMOP, an error is signalled if a
1193 class-option appears multiple times in a DEFCLASS form. (reported
1195 * bug fix: RANDOM can be compiled when the compiler derives the type
1196 of its numeric argument as a disjoint set of small integers.
1197 (reported by Paul Dietz)
1198 * bug fix: starting a core saved with shared objects loaded when
1199 those objects are not available no longer causes threaded SBCL to
1200 hang. (reported by Sean Ross)
1201 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1203 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1204 argument, which must always be NIL. (reported by Kalle Niemitalo)
1205 * bug fix: printing 1.0d+23 no longer results in an error.
1206 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1207 * bug fix: structure slot setters preserve evaluation order. (thanks
1209 * fixed some bugs related to Unicode integration:
1210 ** RUN-PROGRAM can allow its child to take input from a Lisp
1211 stream. (reported by Stefan Scholl)
1212 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1213 locales. (reported by Ken Causey)
1214 * fixed some bugs revealed by Paul Dietz' test suite:
1215 ** INCF, DECF and REMF evaluate their place form as specified in
1217 ** FORMATTER expands ~{ ~} iteration directives with V or #
1218 parameters correctly.
1219 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1220 consequent uses no arguments correctly.
1221 ** the system has a partial understanding of the (COMPLEX RATIO)
1223 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1226 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1227 * new feature: a build-time option (controlled by the :SB-UNICODE
1228 keyword feature, enabled by default) for building the system with
1229 support for the entire 21-bit character space defined by the
1231 * new feature: the system now has rudimentary external-format
1232 support; the primary user-visible change at this time is that
1233 characters with the high bit set (such as lower-case-e-acute) will
1234 print correctly to a terminal in a UTF-8 environment.
1235 * minor incompatible change: BASE-CHAR no longer names a class;
1236 however, CHARACTER continues to do so, as required by ANSI.
1237 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1238 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1239 has been moved to the SB-EXT package (temporarily re-exported from
1241 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1242 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1244 * fixed bug #351: better error-handlind and reporting for malformed
1246 * fixed bug #350: bignum-printing is now more memory-efficient,
1247 allowing printing of very large bignums, eg. (expt 2 10000000).
1248 (reported by Bruno Haible)
1249 * fixed bug #302: better primitive-type selection for intersection
1251 * fixed bug #308: non-graphic characters now all have names, as
1252 required. (reported by Bruno Haible)
1253 * bug fix: redefining a class with different superclasses now correctly
1254 removes it from the direct-subclasses of its previous superclasses.
1255 (reported by David Morse)
1256 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1257 new value now works. (reported by Bruno Haible)
1258 * bug fix: correct canonicalization of multiple non-standard slot
1259 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1260 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1261 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1262 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1263 now exists, an signals an error.
1264 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1265 during macroexpansion for non-iteration variables as well. (reported
1266 by Bruno Haible for CMUCL)
1267 * bug fix: Cyclic structures and unprintable objects in compiler
1268 messages no longer cause errors. (reported by Bruno Haible)
1269 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1270 and READ-FROM-STRING all now return a primary value of NIL if
1271 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1272 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1273 (reported by Bruno Haible for CMUCL)
1274 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1275 do not designate a condition. (reported by Bruno Haible for
1277 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1278 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1279 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1280 permitting resolution in favour of any of the conflicting symbols.
1281 (reported by Bruno Haible for CMUCL)
1282 * bug fix: EQUAL compiler optimizations is less aggressive on
1283 strings which can potentially compare true despite having distinct
1284 specialized array element types.
1285 * bug fix: unit enumerations can be defined without dividing by
1286 zero. (thanks to Vincent Arkesteijn)
1287 * FORMAT compile-time argument count checking has been enhanced.
1288 (report from Bruno Haible for CMUCL)
1289 * a partial workaround for the bug 262: the compiler does not try to
1290 inline-expand a local function doing RETURN-FROM from a deleted
1291 BLOCK. (thanks to Peter Denno for the bug report and to David
1292 Wragg for the simple test case)
1293 * fixed some bugs revealed by Paul Dietz' test suite:
1294 ** RENAME-PACKAGE allows all package designators as new package
1296 ** constraint propagation and lambda variable substitution are
1297 more cautious in dealing with partially deleted code.
1298 ** compiler sometimes lost reoptimization passes.
1299 ** CERROR, given a condition as condition designator, passes its
1300 remaining arguments to the continue format control without
1302 ** Case-altering FORMAT directives work correctly on non-ASCII
1304 ** The REAL type specifier handles bounds outside the floating
1305 point ranges without signalling FLOATING-POINT-OVERFLOW.
1306 ** Functions with IR1-transformations can create intercomponent
1307 references to global functions.
1308 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1310 ** FORMAT ~R treats a nil value for its first parameter correctly.
1312 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1313 * enhancement: saving cores with foreign code loaded is now
1314 supported on x86/NetBSD and sparc/Linux in addition to the previously
1315 supported platforms.
1316 * bug fix: on some platforms repeated installations caused multiple
1317 copies of HTML documentation to be installed -- should not happen
1318 any more. (reported by Stefan Scholl)
1319 * bug fix: parsing self-recursive alien record types multiple times
1320 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1321 original patch by Helmut Eller for CMUCL)
1322 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1323 (thanks to Richard Kreuter)
1324 * bug fix: defining classes whose accessors are methods on existing
1325 generic functions in other (locked) packages no longer signals
1326 bogus package lock violations. (reported by François-René Rideau)
1327 * bug fix: special variables as DEFMETHOD parameters no longer have
1328 associated bogus type declarations. (reported by David Wragg and
1330 * bug fix: read-write consistency on streams of element-type
1331 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1332 * bug fix: redefiniton of the only method of a generic function with
1333 no DEFGENERIC no longer emits a full WARNING. In addition,
1334 redefinition of generic functions with no DEFGENERIC to an
1335 incompatible lambda list now signals an error. (thanks to Zach
1337 * bug fix: DEFGENERIC now works even when there's a function of the
1338 same name in an enclosing lexical environment. (thanks to Zach
1340 * fixed compiler failure, caused by instrumenting code during
1341 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1342 * optimization: added loop analysis and improved register allocation
1343 * fixed some bugs revealed by Paul Dietz' test suite:
1344 ** POSITION on displaced vectors with non-zero displacement
1345 returns the right answer.
1346 ** (SIMPLE-STRING) is a valid type specifier for sequence
1348 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1350 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1351 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1353 ** The justification version of the FORMAT ~< directive treats
1354 non-zero minpad parameter correctly.
1356 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1357 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1358 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1359 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1360 the supported interface.
1361 * new feature: Single-stepping of code compiled with DEBUG 2 or
1362 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1363 * new feature: saving cores with foreign code loaded is now
1364 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1365 Timothy Moore's work for CMUCL)
1366 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1367 parameters to * instead of NIL if no initform is supplied.
1368 (reported by Johan Bockgaard)
1369 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1370 correctly when FUNCALL appears as the car of the form. Note:
1371 despite this FUNCALL forms are not currently subject to
1372 compiler-macro expansion. (port of Raymond Toy's fix for the
1373 same from CMUCL, reported by Johan Bockgaard)
1374 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1375 (thanks for Teemu Kalvas)
1376 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1377 case instead of terminating the iteration (thanks for Julian
1378 Squires, Sean Champ and Raymond Toy)
1379 * bug fix: incorrect expansion of defgeneric that caused a style
1380 warning. (thanks for Zach Beane)
1381 * bug fix: slot accessor effective method computation works properly
1382 for classes with multiple non-standard applicable methods on
1383 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1384 * on x86 compiler supports stack allocation of results of LIST and
1385 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1386 CMUCL implementation by Gerd Moellmann)
1387 * fixed some bugs revealed by Paul Dietz' test suite:
1388 ** FORMAT strings with both the ~<~:;~> form of the justification
1389 directive and pretty-printing directives cause an error.
1391 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1392 * incompatible change: the internal functions
1393 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1394 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1395 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1396 instead of the old functions.
1397 * new feature: on platforms where "dladdr" is available foreign
1398 function names now appear in backtraces. (based on Helmut Eller's
1400 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1401 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1403 * documentation: ASDF manual is now distributed with SBCL.
1404 * bug fix: SBCL can now load its contributed modules with REQUIRE
1405 even if the system-provided entries have been removed from
1406 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1408 * bug fix: backtraces involving undefined functions or assembly
1409 routines are more informative. (thanks to Brian Downing)
1410 * bug fix: mutually referent alien structures now work correctly.
1411 (reported by Rick Taube)
1412 * bug fix: structures defined by WITH-ALIEN can be referred to
1413 within other definitions in the same WITH-ALIEN.
1414 * bug fix: division operators (MOD, TRUNCATE and the like) with
1415 constant zero divisors and integer dividends no longer generate
1417 * bug fix: provide default methods for INPUT-STREAM-P and
1418 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1419 * bug fix: improve the detection and resolution of MOP metacycles
1420 (where effective method computation depends on the generic function
1421 having its effective method computed). (reported by Bruno Haible)
1422 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1423 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1424 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1425 * bug fix: pathnames with relative directory components can be
1426 represented relative to default pathnames.
1427 * optimization: in taking the GCD of bignums, reduce the two bignums
1428 to approximately the same size (using Euclid's algorithm) before
1429 applying the more sophisticated binary GCD. (thanks to Juho
1431 * optimization: COUNT on bitvectors now operates word-at-a-time.
1432 * optimization: ASH with a positive, but not necessarily constant,
1433 (leftwards) shift, when in a modular context, is compiled to a
1435 * fixed some bugs revealed by Paul Dietz' test suite:
1436 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1437 if the corresponding argument is NIL.
1438 ** FORMAT directives accepting a minpad parameter treat negative
1440 ** literal commas as character parameters to FORMAT directives are
1442 ** literal spaces directly after ~<Newline> directives within a
1443 format-logical-block (~:< ~@:>) do not induce :FILL-style
1444 conditional newlines.
1445 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1446 as specified: it no longer includes conditional newlines).
1447 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1449 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1450 :PER-LINE-PREFIX argument does not evaluate to a string.
1452 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1453 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1454 the manual for details; for now, package locks can be disabled by
1455 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1456 if no major problems are found then it is likely that they will be
1457 compiled in unconditionally.
1458 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1459 unsupported operators on all platforms. To load a shared library
1460 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1461 object file, link it into a shared library outside of SBCL and
1462 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1464 * bug workaround: ROOM T and the GC were not working together
1465 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1466 reporting facility aren't preserved. That reporting has been
1467 disabled, so now until and unless someone figures out how to make it
1468 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1469 * minor incompatible change: as threatened around sbcl-0.8.0, the
1470 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1471 The symbols are also exported from SB-PCL for backwards
1472 compatibility, but more so than before SB-PCL should be treated as
1473 an implementation-internal package.
1474 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1476 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1477 It is known as of this release not to work on the Alpha, however.
1478 * fixed bug #167: errors signalled due to illegal syntax in method
1479 bodies are now more legible.
1480 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1481 designators and can hence be used with TYPEP.
1482 * fixed bug #333: CHECK-TYPE now ensures that the type error
1483 signalled, if any, has the right object to be accessed by
1484 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1485 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1486 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1488 * fixed bug #334: programmatic addition of slots using specialized
1489 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1490 and :ALLOCATION :CLASS slots.
1491 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1492 float exponents. (rereported by Peter Seibel)
1493 * fixed a bug: #\Space (and other whitespace characters) are no
1494 longer considered to be macro characters in standard syntax by
1495 GET-MACRO-CHARACTER.
1496 * fixed bug: initialization of condition class metaobjects no longer
1497 causes an instance of the condition to be created. (reported by
1499 * fixed bug: it is now possible to have slots such that
1500 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1501 description is neither :INSTANCE nor :CLASS.
1502 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1503 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1504 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1505 non-standard SLOT-DEFINITION-ALLOCATION values.
1506 * fixed bugs: various race conditions handling exiting threads.
1507 CL-PPCRE's thread torture test now passes.
1508 * fixed arguably-a-bug: GC time proportional to number of threads in
1509 system even when most of them are idle
1510 * optimization: improved performance of BIT and SBIT on bit-vectors.
1511 * .fasl file incompatibility: The fasl file version number has
1512 been incremented because of changes associated with package locks.
1514 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1515 * minor incompatible change: the system no longer provides
1516 optimization hints (conditions which are TYPEP
1517 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1518 mode; these hints are still emitted when higher SPEED optimization
1520 * new contrib module: a sampling profiler (profiling by statistical
1521 sampling, rather than by instrumenting functions) is available as
1522 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1523 * the behaviour of the standard function ED is now customizeable by
1524 third parties through a hook variable: see ED's documentation
1525 string for information on the protocol.
1526 * the compiler no longer emits efficiency notes for (FUNCALL X)
1527 when the type of X is uncertain under default optimization
1529 * fixed bug 276: mutating a binding of a specialized parameter to a
1530 method to something that is not TYPEP the specializer is now
1532 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1533 exhibit the required behaviour on the x86 platform. (thanks to
1534 Peter van Eynde, Eric Marsden and Bruno Haible)
1535 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1536 even for difficult arguments. (reported by Peter Graves)
1537 * fixed bug 141a: the backquote printer now descends quoted
1539 * fixed another bug in backquote printing: no more destructive
1540 modification of the form's list structure. (reported by Brian
1542 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1543 move between its address being taken and the call to
1544 interrupt_thread, fixing a crashing race condition.
1545 * the SB-POSIX contrib implementation has been adjusted so that it
1546 no longer exhibits ridiculously poor performance when constructing
1547 instances corresponding to C structs.
1549 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1550 * minor incompatible change: the sb-grovel contrib now treats C
1551 structures as alien (in the sense of SB-ALIEN) objects rather than
1552 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1553 has implications for memory management of client code
1554 (sb-grovel-returned objects must now be manually managed) and for
1555 type safety (alien objects now have full types).
1556 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1557 used to control emission of compiler diagnostics, rather than the
1558 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1559 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1560 quality should be considered deprecated.
1561 * install.sh now installs the user manual as well
1562 * (not quite a new documentable feature, but worth considering in
1563 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1564 beginnings of a semantically meaningful condition hierarchy is
1565 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1567 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1568 designator as the defaults argument.
1569 * fixed bug: Displaced arrays whose displaced-to array has become
1570 too small now cause ARRAY-DIMENSION to signal an error, providing
1571 for safer bounds-checking. (reported by Bruno Haible)
1572 * fixed bug: DEFCLASS slot definitions with identical :READER and
1573 :WRITER names now signal a reasonable error. (reported by Thomas
1575 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1577 * fixed bug: Invalid slot specification errors now print correctly.
1578 (thanks to Zach Beane)
1579 * fixed bug 320: Shared to local slot value transfers in class
1580 redefinitions now happen corrently from superclasses as
1581 well. (reported by Bruno Haible)
1582 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1584 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1585 improper lists correctly. (reported by Bruno Haible)
1586 * fixed bug 313: source-transform for <fun-name> was erroneously
1587 applied to a call of a value of a variable with name <fun-name>.
1588 (reported by Antonio Menezes Leitao)
1589 * fixed bug 307: The obsolete instance protocol ensures that
1590 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1591 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1592 remove dead unknown-values globs from the middle of the stack.
1593 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1594 easier to resume long computations after using *BREAK-ON-SIGNALS*
1595 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1596 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1597 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1598 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1599 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1600 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1602 * fixed a (fairly theoretical) bug in string printing: if
1603 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1604 does not have array-element-type equal to the most general string
1606 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1607 function name. (thanks to Nikodemus Siivola)
1608 * fixed bug: compile-time format string checker failed on
1609 non-closed ~{. (reported by Thomas F Burdick)
1610 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1611 don't warn on legal loop code involving more than one aggregate
1612 boolean. (thanks to Nikodemus Siivola)
1613 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1614 in keyword position to MAKE-PACKAGE was not being reported as
1616 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1617 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1619 * fixed bugs 280 and 312: the checking for multiple definitions in a
1620 file is less likely to become confused by uses of inline
1622 * fixed bug: the #S reader macro performs the keyword coercion
1623 specified for slot names. (reported by Kalle Niemitalo)
1624 * fixed bug: lambda lists may contain symbols whose names start with
1625 & but are not lambda-list-keywords; their occurrence triggers a
1627 * fixed bug 321: define-method-combination argument lambda lists do
1628 not cause invalid code to be generated when &optional or &aux
1629 variables have default values. (reported by Bruno Haible)
1630 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1631 have CLOS classes; this appears also to have fixed reported
1632 instability in stack exhaustion detection.
1633 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1634 MUFFLE-WARNING when no associated restart is present is now
1636 * optimization: rearranged the expansion of various defining macros
1637 so that each expands into only one top-level form in a
1638 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1640 * optimization: used a previously little-used slot in symbols to
1641 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1643 * fixed some bugs revealed by Paul Dietz' test suite:
1644 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1646 ** Fixed an optimization bug related to inheritance of initforms
1647 from local to shared slots.
1648 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1649 ** CAST optimizer forgot to flush argument derived type.
1650 ** print/read consistency on floats is now orders of magnitude
1651 more likely. (thanks also to Bruno Haible for a similar report
1653 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1654 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1655 for objects of type REAL. Make it so.
1656 ** SXHASH is specified (infelicitously) to respect similarity,
1657 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1658 it so. (thanks to Markus Ziegler)
1659 ** on the Alpha, the compiler succeeds in compiling functions
1660 returning a known number of arguments greater than 63.
1661 ** fixed handling of invalid NIL arguments in keyword position
1663 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1665 ** provided a readably-printable representation for RANDOM-STATE
1667 ** ensured that pathnames, where they have a namestring, always
1668 print using #P"..." syntax.
1670 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1671 * Support for the forthcoming 2.0 version of the NetBSD kernel
1672 running on x86 hardware has been added. (thanks to Perry
1673 E. Metzger most immediately, and others for their past work)
1674 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1675 current release version; SBCL's OpenBSD support had been broken
1676 since about the time of OpenBSD's switch to ELF binary format.)
1677 * [placeholder for DX summary]
1678 ** user code with &REST lists declared dynamic-extent, under high
1679 speed or space and low safety and debug optimization policy.
1680 * The manual has been converted to Texinfo format and the debugger
1681 chapter from the cmucl manual has been added.
1682 * A facility has been added to extract documentation strings from
1683 sbcl and store them as Texinfo-formatted snippets for inclusion in
1684 the manual (via Texinfo's @include directive)
1685 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1686 the test case to Dave Roberts)
1687 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1688 pointers. (reported by Sean Ross)
1689 * bug fix: PROFILE output is printed nicely even for large numerical
1690 values. (thanks to Zach Beane)
1691 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1692 greater than 32 handle EOF correctly.
1693 * bug fix: on X86 an immediate argument of the IMUL instruction is
1694 correctly printed in disassembly. (thanks to Lutz Euler)
1695 * bug fix: class slots in redefined classes preserve their old
1696 values. (thanks to Bruno Haible and Nikodemus Siivola)
1697 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1698 succeeds. (reported by Marco Baringer)
1699 * fixed some bugs revealed by Paul Dietz' test suite:
1700 ** READ-SEQUENCE now works on ECHO-STREAMs.
1701 ** RATIONALIZE works more according to its specification. (thanks
1704 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1705 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1706 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1707 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1708 should matter to you only if you rebind the printer control
1709 variables and then find you want different bindings in the
1710 debugger than in the ordinary execution of your program.)
1711 * The runtime build system has been tweaked to support building
1712 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1713 assembler and linker. (thanks to Nikodemus Siivola)
1714 * Unbound, undefined, undeclared variables now trigger full
1715 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1716 more often programmer error than deliberate exploitation of undefined
1718 * optimization: the hash algorithm for strings has changed to one
1719 that is less vulnerable to spurious collisions. (thanks to Juho
1721 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1722 needless bounds checking. (thanks to Juho Snellman)
1723 * optimization: implemented multiplication as a modular
1724 (UNSIGNED-BYTE 32) operation on the PPC backend.
1725 * fixed some bugs revealed by Paul Dietz' test suite:
1726 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1728 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1729 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1730 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1732 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1734 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1736 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1737 their output stream on EOF from read.
1738 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1739 have been read to end-of-file.
1740 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1742 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1743 description of determination of which consecutive characters
1745 ** Printing the "Space" character with escaping on now yields "#\\ ",
1746 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1747 ** Reading floating-point numbers with *READ-BASE* set to a number
1748 less than 10 works correctly.
1749 ** Reading floating-point numbers with *READ-BASE* set to a number
1750 more than 10 works correctly.
1751 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1752 the readtable currently in effect.
1754 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1755 * minor incompatible change: parsing of namestrings on a physical
1756 (Unix) host has changed; numbers after the final #\. in a
1757 namestring are no longer interpreted as a version field. This is
1758 intented to be largely invisible to the user, except that the
1759 meaning of the namestring "*.*.*" has changed: it now refers to a
1760 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1761 should usually be replaced by
1762 (make-pathname :name :wild :type :wild :version :wild)
1763 with the added benefit that this is more likely to be portable.
1764 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1765 option to OPEN now signals an error if the file being opened
1766 exists; this may have an impact on existing code.
1767 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1768 signals received, on the PowerPC platforms (both Linux and
1769 Darwin). (thanks to Pierre Mai for pointing out the location of
1771 * several fixes on OS X: The system now builds and runs cleanly on
1772 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1773 (thanks to Brian Mastenbrook)
1774 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1775 second-resolution: integer multiples of 1/3600 between -24 and 24.
1776 (thanks to Vincent Arkesteijn)
1777 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1778 type when called with 1 argument; PEEK-CHAR checked type of
1779 PEEK-TYPE only after having read first character from a
1780 stream. (reported by Peter Graves)
1781 * bug fix: the garbage collector now has much better locality
1782 behaviour, and in particular no longer treats all memory as being
1783 exhausted when many small objects point to each other in a deeply
1785 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1786 recognized as being TYPEP their class.
1787 * bug fix: the PUSHNEW documentation string has been corrected.
1788 (thanks to Vincent Arkesteijn)
1789 * bug fix: defaulting of the value for the last of an atypically
1790 large number of multiple values being bound was not being
1791 performed correctly on the Alpha or PPC platforms
1792 * optimization: implemented multiplication as a modular
1793 (UNSIGNED-BYTE 32) operation on the x86 backend.
1794 * optimization: SEARCH on simple-base-strings can now be open-coded.
1795 (see also contrib/compiler-extras.lisp for inspiration for
1796 teaching the compiler about the Boyer-Moore algorithm).
1797 * value, returned by MAX (and MIN) called with several EQUALP, but
1798 not EQL, arguments now does not depend on compiler settings.
1799 * fixed some bugs revealed by Paul Dietz' test suite:
1800 ** in stack analysis liveness information is propagated from
1801 non-local entry points.
1802 ** pathwise CAST removing failed when the CAST node did not start
1804 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1805 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1807 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1808 host is already defined.
1809 ** RENAME-FILE works on streams instead of signalling an internal
1811 ** PEEK-CHAR uses the current readtable when determining whether
1812 or not a character is whitespace.
1813 ** MERGE-PATHNAMES handles the case when the pathname does not
1814 specify a name while the default-pathname specifies a version.
1815 ** Pathnames now stand a better chance of respecting print/read
1817 ** Attempting to use standardized file system operators with a
1818 pathname with invalid :DIRECTORY components signals a
1820 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1822 ** DIRECTORY on logical pathnames is more correct.
1823 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1824 signal a TYPE-ERROR if their argument is not a stream.
1825 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1826 designator argument does not designate a stream.
1827 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1828 examining the synonym.
1829 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1831 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1832 element-type larger than ([UN]SIGNED-BYTE 32).
1834 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1835 * When built with the :SB-FUTEX feature, threaded builds now take
1836 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1837 for faster/more reliable mutex and condition variable support.
1838 * Incompatible change (but one you probably shouldn't have been using
1839 anyway): the interface and code for arbitrating between multiple
1840 threads in the same user session has been redesigned.
1841 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1842 error when called without an explicit environment argument.
1843 (thanks to Nikodemus Siivola)
1844 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1845 confused about their position. (thanks to Adam Warner and Gerd
1847 * bug fix: Pretty printing backquoted forms with unquotations in the
1848 argument list position of various code constructs such as LAMBDA
1849 now works correctly. (reported by Paul Dietz)
1850 * bug fix: Pretty printing unquotations no longer loses all
1851 stream position information.
1852 * optimization: performance of string output streams is now less
1853 poor for multiple small sequence writes.
1854 * optimization: performance of CSUBTYPEP in the presence of complex
1855 expressions involving CONS and NOT many times has been improved.
1856 (reported by Paul Dietz)
1857 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1859 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1860 streams for the REPL, for the debugger, and so on. (thanks to
1862 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1863 * fixed some bugs revealed by Paul Dietz' test suite:
1864 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1865 is ignored if it is too small, rather than propagating through
1866 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1867 ** extremely complex negations of CONS types were not being
1868 sufficiently canonicalized, leading to inconsistencies in
1870 ** VALUES tranformer lost derived type.
1872 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1873 * fixed a bootstrapping bug: the build process no longer assumes
1874 that the various BOOLE-related constants have the same value in
1875 host and target lisps. (noted by Paul Dietz' test suite on an
1876 SBCL binary built from CLISP)
1877 * The system can now be dynamically linked on the MIPS platform,
1878 which enables dynamic loading of foreign code from Lisp. (thanks
1879 to Ralf Baechle for discussions on the MIPS ABI)
1880 * The system now records debugging information for its own source
1881 files in a filesystem-position-independent manner, relative to
1882 the "SYS" logical host.
1883 * fixed a compiler bug: MV-LET convertion did not check references
1884 to the "max args" entry point. (reported by Brian Downing)
1885 * tweaked disassembly notes to be less confident about proclaiming
1886 some instruction as an LRA. (thanks to Brian Downing)
1887 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1888 now each have their own history, command character, and other
1889 characteristics. (thanks to David Lichteblau)
1890 * fixed some bugs revealed by Paul Dietz' test suite:
1891 ** compiler failure in compiling LOGAND expressions including a
1893 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1894 shift greater than 32.
1895 ** FLUSH-DEST did not mark blocks for type check regeneration.
1896 ** HANDLER-CASE failed to accept declarations in handler clauses
1897 in some circumstances.
1899 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1900 * New code in contrib/sb-introspect (still probably not entirely
1901 stable yet) provides some support for smart Lisp development
1902 environments like SLIME.
1903 * The conditions signalled for errors occurring when loading .fasl
1904 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1905 in a way which should help ASDF recover gracefully.
1906 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1907 stuff has been cleaned up. If you code contrib/ stuff, this might
1908 affect you, and you can look at contrib/README, contrib/STANDARDS,
1909 and/or the 0.8.4.27 diff to check.
1910 * In full calls the compiler now does not generate checks for declared
1911 argument types for all arguments.
1912 * various threading fixes
1913 ** and some experimental patches which didn't make it into
1914 the main tree for this release, but which are shipped in
1915 contrib/experimental-thread.patch as a possible fix for some
1916 failures (deadlock, spinning...) in GC-intensive multithreaded
1918 * fixed PPC build problem (source code incompatibility of different
1919 library versions): added offsetof() hackery which attempts to divine
1920 where glibc maintainers put uc_mcontext today
1921 * fixed bug 282: compiler does not trust type assertions while passing
1922 arguments to a full call.
1923 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1924 &OPTIONAL VALUES type parameter.
1925 * fix bug 214: algorithm for noting rejected templates is now more
1926 similar to that of template seletion. (also reported by rydis on
1928 * fixed bug 141b: printing backquoted information readably and prettily
1929 inserts a space where necessary.
1930 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1931 (CHARACTER 10) are now reported as errors, rather than propagated
1932 as unknown types. (reported by piso on #lisp)
1933 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1934 with values NIL and :ERROR. (thanks to Milan Zamazal)
1935 * fixed bug 191c: CLOS now does proper keyword argument checking as
1936 described in CLHS 7.6.5 and 7.6.5.1.
1937 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1938 longer raise an error; further, using a list as a for-as-arithmetic
1939 counter now raises a meaningful error.
1940 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1941 now understood by sequence creation functions such as MAKE-SEQUENCE
1943 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1944 a STRING-INPUT-STREAM.
1945 * compiler enhancement: SIGNUM is now better able to derive the type
1947 * type declarations inside WITH-SLOTS are checked. (reported by
1949 * fixed some bugs revealed by Paul Dietz' test suite:
1950 ** incorrect optimization of TRUNCATE for a positive first
1951 argument and negative second.
1952 ** compiler failure in let-convertion during flushing dead code.
1953 ** compiler failure while deriving type of TRUNCATE on an
1954 interval, containing 0.
1955 ** ASH of a negative bignum by a negative bignum count now returns
1957 ** intersection of CONS types now canonicalizes properly, fixing
1958 inconsistencies in SUBTYPEP.
1960 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1961 * incompatible change: The --disable-debugger command line
1962 option now clobbers the debugger at a more fundamental
1963 level, by redefining #'INVOKE-DEBUGGER instead of by
1964 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1965 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1966 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1967 enter the debugger REPL and then suffer recursive errors
1968 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1969 while under the new system, BREAK in --disable-debugger mode
1970 terminates the system just as an unhandled error would.
1971 * fixed compiler performance when processing loops with a step >1;
1972 * bug fix: DOCUMENTATION now retrieves generic function
1973 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1974 support has been systematized, and now supports the methods
1975 specified by ANSI, along with a default method and a method for
1976 slot documentation. (reported by Nathan Froyd)
1977 * bug fix: effective methods associated with a generic function are
1978 no longer cached over a change of that generic function's method
1979 combination. (reported by Andreas Fuchs)
1980 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1981 names a non-existent directory. (thanks to Andreas Fuchs)
1982 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1983 the PPC platform, lead to stack corruption; this has been fixed.
1984 (reported by Rainer Joswig)
1985 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1986 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1987 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1988 arithmetic in general, is now much more reliable on the PPC
1990 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1991 platform now returns the right answer.
1992 * optimization: restored some effective method precomputation in
1993 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1994 precomputation is now tunable.
1995 * optimization: compiler-internal data structure use has been
1996 reviewed, and changes have been made that should improve the
1997 performance of the compiler by about 20%.
1998 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1999 simple-base-strings and simple-bit-vectors is improved.
2000 * optimization: the optimization of 32-bit logical and arithmetic
2001 functions introduced in version 0.8.3 on the x86 has been
2002 implemented on the mips, ppc and sparc platforms; an
2003 implementation of the same facility, but for 64-bit arithmetic,
2004 has been added for the alpha.
2005 * microoptimization: the compiler is better able to make use of the
2006 x86 LEA instruction for multiplication by constants.
2007 * bug fix: in some situations compiler did not report usage of
2008 generic arithmetic in (SPEED 3) policy.
2009 * bug 145b fix: compiler used wrong type specifier while converting
2010 MEMBER-types to numeric.
2011 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2013 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2014 index argument now works properly on the Alpha platform.
2015 * bug fix: floating point exception treatment on the Alpha platform
2017 * bug fix: FILE-POSITION works much better on string input and
2018 output streams. (thanks to Nikodemus Siivola)
2019 * bug fix: many threading/garbage collection symptoms sorted.
2020 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2021 might be pseudo-atomic.
2022 * internal change: Stopping for GC is now done with signals not
2023 ptrace. GC is now done in whichever thread wanted it, instead of
2025 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2027 * bug fix: result form in DO is not contained in the implicit
2029 * incompatible change: ICR structure is changed; the value part of
2030 CONTINUATION is now called LVAR; corresponding functions are
2031 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2032 * added type deriver for ISQRT (thanks to Robert E. Brown).
2033 * bug fix: better support for loading from the command line when an
2034 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2036 * fixed some bugs revealed by Paul Dietz' test suite:
2037 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2038 ** ROUND and FROUND now give the right answer when given very
2039 small float arguments.
2040 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2042 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2043 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2044 ** LCM should return a non-negative integer.
2045 ** PARSE-INTEGER returned the index of a terminator instead of the
2046 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2047 ** PARSE-INTEGER returned an incorrect index being applied to a
2049 ** LCM with two arguments of 0 returns 0 rather than signalling
2051 ** unsigned addition of a 32-bit constant with the high bit set no
2052 longer causes an internal compiler error on the x86.
2053 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2054 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2055 with negative last argument.
2056 ** byte specifiers with zero size and position no longer cause
2057 an error during type derivation.
2058 ** bignum multiplication on the Alpha platform now returns the
2060 * porting: The system now builds on SuSE AMD64, although it still
2061 generates a 32-bit binary.
2062 * .fasl file incompatibility: The fasl file version number has
2063 been incremented (because of the changes to internal compiler
2064 data structures referred to above).
2066 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2067 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2068 more accurately, on the Darwin kernel running on PowerPC hardware.
2069 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2070 * Compiler code deletion notes now signal a condition of type
2071 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2072 an associated MUFFLE-WARNING restart.
2073 * The compiler now performs limited argument count validation of
2074 constant format strings in FORMAT, and where appropriate in ERROR,
2075 CERROR and WARN. (thanks to Gerd Moellmann)
2076 * New ASDF-INSTALL contrib can be used for automatic download and
2077 installation of third-party Lisp code from CCLAN or other sites
2079 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2080 which forces another thread to execute a function supplied by the
2082 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2083 now accept and act upon their :ELEMENT-TYPE keyword argument.
2084 (reported by Martin Atzmueller, Edi Weitz)
2085 * bug fix: FILE-POSITION now accepts position designators up to
2086 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2087 is the greater. (thanks to Patrik Nordebo)
2088 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2089 Kalle Olavi Niemitalo)
2090 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2091 printable. (reported by Eric Marsden)
2092 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2093 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2094 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2095 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2096 (thanks to Stig Sandoe)
2097 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2098 UNREAD-CHAR now correctly includes the unread character in the
2099 target sequence. (thanks to Gerd Moellmann)
2100 * bug fix in threaded builds: the system can now be suspended and
2101 resumed by shell job control with minimal disruption.
2102 * bug fixes in times and timezones >2038AD
2103 * better handling of "where is GNU make?" problem in build scripts
2104 (thanks to Nikodemus Siivola)
2105 * new optimization: inside a named function any reference to a
2106 function with the same name is considered to be a self-reference;
2107 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2108 optimization quality.
2109 * new optimization on x86: logical functions and + now have
2110 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2111 used when the result is truncated to 32 bits.
2112 * VALUES declaration is partially enabled.
2113 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2114 * fixed some bugs revealed by Paul Dietz' test suite:
2115 ** The system now obeys the constraint imposed by
2116 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2117 types form a lattice under type intersection.
2118 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2119 ** ASSOC now ignores NIL elements in an alist.
2120 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2121 and (1+ MOST-POSITIVE-FIXNUM) answers.
2122 ** The addition of a method with invalid qualifiers to a generic
2123 function does not cause an error to be signalled immediately;
2124 a warning is signalled, and the error is generated only on
2125 calling the generic function.
2126 * changed .fasl file version number, in part to add type codes for
2127 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2128 obscure ANSI requirements
2130 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2131 * fixed bug 148: failure to inline-expand a local function left
2132 garbage, confusing the compiler.
2133 * fixed bugs 3cd: structure slot readers perform type check if the
2134 slot can have an invalid value (i.e. it is either not initialized
2135 or can be written with a less specific slot writer).
2136 * bug fix: the compiler now traps array references to elements off
2137 the end of an array; previously, the bounds checking in some
2138 circumstances could go off-by-one.
2139 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2141 * type declarations for array element types now obey the description
2142 on the CLHS page "Declaration TYPE", as per discussions on
2143 sbcl-help around 2003-05-08. This means that a declaration
2144 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2145 declaration, all references to BAR will be asserted or assumed
2146 (with THE, so dependent on compiler policy) to involve objects of
2147 type FOO. Note that no such declaration is implied in
2148 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2149 * declared types of functions from the "Conditions"
2150 chapter. (reported by Paul Dietz)
2151 * bug fix: CERROR accepts a function as its first argument.
2152 * bug fix: NTH an NTHCDR accept a bignum as index
2153 arguments. (reported by Adam Warner)
2154 * optimization: character compare routines now optimize comparing
2155 against a constant character. (reported by Gilbert Baumann)
2156 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2157 argument now works properly on the MIPS platform.
2158 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2159 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2160 declared to have a functional type, does not cause an error
2162 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2163 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2164 * fixed bug: if last continuation of a deleted block has a
2165 destination, this destination should be deleted too. (reported by
2167 * fixed a bug in the bootstrap process: the host compiler's values
2168 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2169 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2170 test case from Patrik Nordebo)
2171 * improved the ability of the disassembler on the PPC platform to
2172 provide helpful disassembly notes.
2173 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2174 the class in more cases than previously.
2175 * bug fix: FILE-POSITION now understands :START and :END for
2176 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2177 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2178 8) data. (thanks to David Lichteblau)
2179 * bug fix: it is possible to add a method to a generic function
2180 without lambda list.
2181 * bug fix: reader failed to signal END-OF-FILE inside an
2182 object representation. (reported by Nikodemus Siivola)
2183 * fixed some bugs revealed by Paul Dietz' test suite:
2184 ** LAST and [N]BUTLAST should accept a bignum.
2185 ** condition slot accessors are methods.
2186 ** (VECTOR NIL) is a subtype of STRING.
2188 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2189 * minor incompatible change: some nonsensical specialized lambda
2190 lists (used in DEFMETHOD) which were previously ignored now signal
2192 * minor incompatible change: the system is now aware of the types of
2193 variables in the COMMON-LISP package, and will signal errors for
2194 most violations of these type constraints (where previously they
2195 were silently accepted).
2196 * minor incompatible change: COMPILE-FILE now uses the freedom
2197 afforded (ANSI 3.2.2.3) to use derived function types for
2198 functions defined in the same file. This also permits the system
2199 to warn on static type mismatches and function
2200 redefinition. (Currently it does not work with high DEBUG level.)
2201 * minor incompatible change: VALUES declaration is disabled.
2202 * When issuing notes, the compiler now signals a condition of type
2203 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2204 restart for use in user handlers. It is expected that the
2205 COMPILER-NOTE condition will eventually become a condition
2206 supertype to a hierarchy of note types, which will then be
2207 handleable in a similar fashion. However, at the moment, no such
2208 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2209 but it's an implementation detail, not a classification for the
2211 * Changes in type checking closed the following bugs:
2212 ** type checking of unused values (192b, 194d, 203);
2213 ** template selection based on unsafe type assertions (192c, 236);
2214 ** type checking in branches (194bc).
2215 * A short form of VALUES type specifier has ANSI meaning (it has
2216 increased the number of situations when SBCL cannot perform type
2218 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2219 keywords or constants is permissible.
2220 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2221 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2222 classes. (thanks to Antonio Martinez)
2223 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2224 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2225 argument to a pretty-print stream. Instead, it leaves any such
2226 operations to DESCRIBE-OBJECT methods.
2227 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2229 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2230 lambda lists. (thanks to David Lichteblau)
2231 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2232 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2233 (thanks to Gerd Moellmann)
2234 * fixed bug 246: increased compilation speed of long
2235 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2237 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2239 * DEFCONSTANT now throws a condition of type
2240 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2241 constant to a non-EQL value; CONTINUE and ABORT restarts
2242 respectively change and preserve the value.
2243 * fixed bug 63: The code walker, part of the implementation of CLOS,
2244 is now better at handling symbol macros.
2245 * bug fix: There is no longer an internal implementation type named
2246 CL:LENGTH. (reported by Raymond Toy)
2247 * bug fix: In macro-like defining macros/special operators the
2248 implicit block does not enclose the lambda list.
2249 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2250 enclosed in parentheses) are not suitable as type specifiers, and
2251 their use properly signals an error now.
2252 * bug fix: An argument count mismatch for a type specifier in code
2253 being compiled no longer causes an unhandled error at compile
2254 time, but signals a compile-time warning.
2255 * fixed simple vector readable printing
2256 * bug fix: DESCRIBE takes more care over whether the class
2257 precedence list slot of a class is bound before accessing it.
2258 (reported by Markus Krummenacker)
2259 * bug fix: FORMATTER can successfully compile pretty-printer format
2260 strings which use variants of the ~* directive inside.
2261 * bug fix: SEARCH now applies its TEST predicate to the elements of
2262 the arguments in the correct order. (thanks to Wolfhard Buss)
2263 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2264 in type inference. (thanks to Robert E. Brown)
2265 * bug fix: Reading in symbols with an explicit package name of ""
2266 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2267 * fixed some bugs revealed by Paul Dietz' test suite:
2268 ** NIL is now allowed as a structure slot name.
2269 ** Arbitrary numbers, not just REALs, are allowed in certain
2270 circumstances in LOOP for-as-arithmetic clauses.
2271 ** Multiple class redefinitions before slot access no longer
2272 causes a type error.
2273 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2274 association between the name and a class.
2275 ** Generic functions with non-standard method-combination and over
2276 five methods all of which return constants no longer return NIL
2277 after the first few invocations. (thanks to Gerd Moellmann)
2278 ** CALL-NEXT-METHOD with no arguments now passes the original
2279 values of the arguments, even in the presence of assignment.
2280 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2282 ** DISASSEMBLE works with closures and funcallable instances.
2283 ** ADD-METHOD now returns the generic function, not the new method.
2284 ** FIND-METHOD signals an error if the lengths of the specializers
2285 is incompatible with the generic function, even if the ERRORP
2287 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2288 which its argument is a member.
2289 ** DEFCLASS only redefines the class named by its class-name
2290 argument if that name is the proper name of the class;
2291 otherwise, it creates a new class.
2292 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2293 of the UNBOUND-SLOT condition to the name of the slot.
2294 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2295 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2296 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2297 SLOT-MAKUNBOUND in the specified fashion.
2299 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2300 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2301 cross-compilation host. As a consequence, we can now bootstrap our
2302 way up to SBCL starting with a bare gcc toolchain and human-readable
2303 source code (first the source to CLISP, then the source to SBCL).
2304 * A contributed module containing a partial implementation of the
2305 simple-streams interface has been included. (thanks to Rudi
2307 * A contributed module implementing the RFC1321 Message Digest
2308 Algorithm, known as MD5, has been included.
2309 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2310 no longer has any effect, as the code controlled by this feature
2311 has been deleted. (As far as we know, no-one has ever built using
2312 this feature, and its semantics were confused in any case).
2313 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2314 work on conditions (as required by the ANSI specification),
2315 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2316 expected behaviour on conditions. Users should note, however,
2317 that such behaviour is not required by the ANSI specification,
2318 and so use of this behaviour may render their code unportable.
2319 * fixed some bugs revealed by Paul Dietz' test suite:
2320 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2322 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2323 ** SXHASH on condition objects no longer returns NIL.
2324 ** :ALLOCATION :CLASS slots are better treated; their values are
2325 updated on class redefinition, and initforms inherited from
2326 superclasses are applied.
2327 ** REMOVE-METHOD returns its generic function argument even when
2328 no method was removed.
2329 ** SHARED-INITIALIZE now initializes the values of the requested
2330 slots, including those with :ALLOCATION :CLASS.
2331 ** ALLOCATE-INSTANCE now works on structure classes defined via
2332 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2334 ** SLOT-EXISTS-P now works on conditions, as well as structures
2336 ** MAKE-LOAD-FORM now has the required methods on
2337 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2338 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2339 keyword, and now implements the SLOT-NAMES argument.
2340 ** methods with &OPTIONAL arguments no longer allow too many
2341 arguments to be passed in the call without error.
2342 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2343 option is consistent with the required arguments of the generic
2344 function lambda list.
2345 * bug fix: REQUIRE accepts a string designator. (Thanks to
2347 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2348 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2349 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2350 * bug fix: The compiler now checks for duplicated variables in macro
2352 * bug fix: SETQ on globals returns the correct value.
2353 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2354 not a COMPILER-ERROR (followed by some other strange error on
2355 choosing the CONTINUE restart).
2356 * bug fix: make.sh and friends are now more consistent in the way that
2357 they look for GNU "make".
2359 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2360 * experimental native threads support (on x86 Linux >=2.4 only).
2361 This is not compiled in by default: you need to add :SB-THREAD to
2362 the target features. See the "Beyond ANSI" chapter of the manual
2364 * fix for longstanding nonANSIism: The old distinction between
2365 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2366 The return value from CL:FIND-CLASS is now a CLOS class, and
2367 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2368 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2369 * An interface to the MetaObject Protocol, as described in Kiczales,
2370 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2371 MIT Press, 1991, is available from the SB-MOP package.
2372 * incompatible change: the SB-PCL package should now be considered
2373 a private implementation detail, and no longer a semi-private MOP
2375 * minor incompatible change: due to rearrangement for threads, the
2376 control stack and binding stack are now allocated at arbitrary
2377 addresses instead of being hardcoded per-port. Users affected by
2378 this probably have to be doing advanced things with shared
2379 libraries, and will know who they are.
2380 * minor incompatible change: Previously, all --eval forms used were
2381 processed with READ before any of them were processed with EVAL.
2382 Now each --eval form is processed with both READ and EVAL before
2383 the next --eval form is processed. (Thus package operations like
2384 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2385 work as the user might reasonably expect.)
2386 * minor incompatible change: *STANDARD-INPUT* is now only an
2387 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2389 * minor incompatible change: Y-OR-N-P is now character-oriented,
2390 not line oriented. Also, YES-OR-NO-P now works without errors.
2391 (thanks to Antonio Martinez)
2392 * sb-aclrepl module improvements: an integrated inspector, added
2393 repl features, and a bug fix to :trace command.
2394 * Known functions, which cannot be open coded by the backend, are
2395 considered to be able to check types of their arguments. (fixing
2396 a bug report by Nathan J. Froyd)
2397 * fixed a bug in computing method discriminating functions: It is
2398 now possible to define methods specialized on classes which have
2399 forward-referenced superclasses. (thanks to Gerd Moellmann)
2400 * fixed evaluation order in optional entries (reported by Gilbert
2402 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2403 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2404 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2405 required-by-AMOP NAME argument, as well as CLASS and
2406 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2407 * fixed bug 20: DEFMETHOD can define methods using names that are
2408 not the proper names of classes to designate class specializers.
2409 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2410 Unix file descriptors, instead of blowing up. (thanks to Antonio
2412 * Garbage collection refactoring: user-visible change is that a
2413 call to the GC function during WITHOUT-GCING will not do garbage
2414 collection until the end of the WITHOUT-GCING. If you were doing
2415 this you were probably losing anyway.
2416 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2417 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2418 TYPEP the latter but not the former.
2419 * The compiler issues a full WARNING for calls to undefined functions
2420 with names from the CL package.
2421 * MAP-INTO for a vector destination is open coded. (reported by
2422 Brian Downing on c.l.l)
2423 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2424 documentation string.
2425 * fixed some bugs revealed by Paul Dietz' test suite:
2426 ** COPY-ALIST now signals an error if its argument is a dotted
2428 ** Condition slots are now accessed more correctly in the presence
2429 of multiple initargs for a given slot.
2430 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2431 exclude restarts of the same name associated with a different
2433 ** DEFCLASS of forward-referenced classes with another
2434 forward-referenced class in the superclasses list no longer
2436 ** Condition slots are now initialized once each, not multiple
2437 times. (thanks to Gerd Moellmann)
2438 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2440 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2441 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2442 arguments contain duplicated elements.
2443 ** RESTART-CASE understands local macros.
2444 ** RESTART-CASE associates exactly its own restarts with a condition.
2445 ** ENDP in safe mode checks its argument to be of type LIST.
2446 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2447 in question is unbound.
2448 ** Optimized MAKE-INSTANCE functions no longer cause internal
2449 assertion failures in the presence of duplicate initargs.
2450 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2451 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2452 propagated even in the ctor optimized implementation of
2454 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2456 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2457 * a better implementation of SXHASH on (simple) bit vectors,
2458 measured both in execution speed and in distribution of results
2459 over the positive fixnums, has been installed. Likewise, a better
2460 implementation of EQUAL for simple bit vectors is now available.
2461 * fixed CEILING optimization for a divisor of form 2^k.
2462 * fixed bug 240 (emitting extra style warnings "using the lexical
2463 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2464 by Antonio Martinez)
2465 * fixed SXHASH, giving different results for NIL depending on type
2466 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2467 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2468 variable. (found by Rolf Wester)
2469 * SBCL does not ignore type declarations for special
2470 variables. (reported by rif on c.l.l 2003-03-05)
2471 * some bug fixes in contrib/sb-aclrepl/
2472 * fixed some bugs revealed by Paul Dietz' test suite:
2473 ** a bug in the CONS type specifier, whereby the CAR and CDR
2474 types got intertwined, has been fixed;
2475 ** the type system is now able to reason about the interaction
2476 between INTEGER and RATIO types more completely;
2477 ** APPEND, [N]REVERSE and NRECONC check that those their
2478 arguments, which must be proper lists, are really so;
2479 ** An array specialized to be unable to hold elements has been
2480 implemented, as required -- yes, really -- by ANSI;
2481 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2482 on malformed property lists;
2484 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2485 * incompatible packaging change: in line with Unix convention,
2486 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2487 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2488 * REQUIRE and PROVIDE are now optionally capable of doing something
2489 useful. See the documentation string for REQUIRE.
2490 * infrastructure for a managed SBCL contrib system: contributed
2491 modules in this release include:
2492 ** the ASDF system definition facility;
2493 ** an interface to the BSD Sockets API;
2494 ** an ACL-like convenience interface to the repl;
2495 (thanks to Kevin Rosenberg)
2496 ** an implementation of ROTATE-BYTE, with efficient implementation
2498 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2499 gives a closer approximation to the right answer than previously.
2500 (thanks to Raymond Toy)
2501 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2502 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2503 optional environment arguments, as required by ANSI.
2504 * fixed bugs in other functions taking environment objects, allowing
2505 calls with an explicit NIL environment argument to be compiled
2507 * fixed bug 228: primary return values from
2508 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2509 COMPILE or FUNCTION.
2510 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2511 :TYPE will no longer signal errors on innocuous objects.
2512 * fixed bug 231b: SETQ is better at respecting type declarations in
2513 the lexical environment.
2514 * fixed a bug in DEFCLASS: classes named by symbols with no or
2515 unprintable packages can now be defined.
2516 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2517 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2518 * fixed compiler failure related to checking types of functions.
2519 (reported by Robert E. Brown)
2520 * the compiler is now much more consistent in its error-checking
2521 treatment of bounding index arguments to sequence functions: in
2522 (SAFETY 3) code, errors will be signalled in almost all cases if
2523 invalid sequence bounding indices are passed to functions defined
2524 by ANSI to operate on sequences.
2525 * fixed a bug in the build procedure: documentation of SBCL-specific
2526 packages is now preserved and available in the final Lisp image.
2527 * lifted FDEFINITION lookup out of loops in the implementation of
2528 many list operations. (thanks to Robert E. Brown)
2529 * fixed a bug in the reader: the #n# reader macro now works for
2530 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2531 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2532 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2533 E. Brown and Tony Martinez respectively)
2534 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2535 SLOT-BOUNDP: in particular, it is now able to optimize them much
2536 better, and is now not vulnerable to having packages renamed.
2537 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2538 always called when a slot is not present in an instance. (thanks
2540 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2541 longer signal an error on unreading a character following EOF on
2542 the previous constituent stream. (thanks to Tony Martinez)
2543 * fixed some bugs revealed by Paul Dietz' test suite:
2544 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2545 not just nonnegative fixnums;
2546 ** the logical bit-array operators such as BIT-AND now accept an
2547 explicit NIL for their "opt-arg" argument (to indicate a
2548 freshly-consed result bit-array);
2549 ** ELT now signals an error on an invalid sequence index in safe
2551 ** the type system is now cleverer about negations of numeric
2552 types, and consequently understands the BIGNUM and RATIO types
2554 ** the type system is now cleverer about the interaction between
2555 INTEGER and RATIO types: while bugs still remain, many more
2556 cases are accurately computed;
2557 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2558 if it is in the last clause;
2559 ** CONSTANTLY now correctly returns a side-effect-free function in
2561 ** DECLARE is no longer treated as a special-operator; in
2562 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2563 * incremented fasl file version number due to the change in the
2564 DEFSTRUCT-SLOT-DESCRIPTION structure.
2566 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2567 * minor incompatible change: code processed by the "interpreter" or
2568 EVAL now has a compilation optimization policy of (DEBUG 2)
2569 (changed from (DEBUG 1)) to improve debuggability of interactive
2570 development, and to allow the use of the debug RETURN command in
2572 * an experimental implementation of the RETURN command for the
2573 debugger has been included. (thanks to Frederik Kuivinen)
2574 * fixed bug 62: constraints were not propagated into a loop.
2575 * fixed bug in embedded calls of SORT (reported and investigated by
2577 * fixed some bugs revealed by Paul Dietz' test suite:
2578 ** printing and reading of arrays with some dimensions having
2579 length 0 (thanks to Gerd Moellmann);
2580 ** BOA constructor with &AUX argument without a default value does
2581 not cause a type error;
2582 ** CONSTANTP now returns true for all self-evaluating objects.
2584 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2585 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2586 accessors that are related by inheritance, as specified in the
2587 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2588 Valtteri Vuorikoski)
2589 * The compiler is now able to inline functions that were defined in
2590 a complex lexical environment (e.g. inside a MACROLET).
2591 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2593 * Support for the upcoming FreeBSD-5.0 release has been included.
2594 (thanks to Dag-Erling Smorgrav)
2595 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2596 effect when it is not in a toplevel context.
2597 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2598 stand a better chance of being correct. (thanks to Gerd
2600 * fixed bug in COERCE, which now signals an error on coercing a
2601 rational to a bounded real type which excludes the expected
2603 * The compiler is now able to derive types more accurately from the
2604 COERCE and COMPILE functions.
2605 * fixed bug 223: functional binding is considered to be constant
2606 only for symbols in the CL package.
2607 * fixed bug 231: SETQ did not check the type of a variable being set
2608 (reported by Robert E. Brown)
2609 * A new optimization for MAKE-INSTANCE has been included, fixing
2610 various bugs (including relating to :ALLOCATION :CLASS slots and
2611 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2613 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2614 ** As required by ANSI, LOOP now disallows anonymous collection
2615 clauses such as COLLECT I in conjunction with aggregate boolean
2616 clauses such as THEREIS (= I 1);
2617 ** LOOP now signals an error when any variable is reused in the
2618 same loop (including the potentially useful construct analogous
2619 to WITH A = 1 WITH A = (1+ A);
2620 ** IT is only a special loop symbol within the first clause of a
2621 conditional loop clause;
2622 ** LOOP with a typed iteration variable over a hashtable now
2623 signals a type error iff it should.
2624 * fixed some other bugs revealed by Paul Dietz' test suite:
2625 ** FILE-STREAM now names the class previously known as FD-STREAM;
2626 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2627 argument) no longer signals an error;
2628 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2629 of the slot symbol, rather than using the current package
2630 ((:CONC-NAME "") continues to intern the slot's name in the
2632 * incremented fasl file version number, because of the incompatible
2633 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2634 of the new implementation of DEFINE-COMPILER-MACRO.
2636 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2637 * Support for building SBCL for MIPS platforms running in
2638 little-endian mode has now been checked in, and basic
2639 functionality on said platforms verified.
2640 * minor incompatible change: PCL now records the pathname of a file
2641 in which methods and the like are defined, rather than its
2643 * minor incompatible change: TRUENAME now considers the truename of
2644 a file naming a directory to be the pathname with :DIRECTORY
2645 component indicating that directory.
2646 * minor incompatible change: a NAMED clause in the extended form of
2647 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2648 reason for the previous behaviour is unclear.
2649 * more systematization and improvement of CLOS and MOP conformance
2650 in PCL (thanks to Gerd Moellman and Pierre Mai):
2651 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2653 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2654 primary methods with no specializers;
2655 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2657 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2658 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2659 FORWARD-REFERENCED-CLASSes; error reporting on
2660 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2662 ** SXHASH on CLOS instances now uses a slot internal to the
2663 instance to return different numbers on distinct instances,
2664 while preserving the same return value through invocations of
2666 ** DEFMETHOD signals errors when methods with longer incongruent
2667 lambda lists are added to generic functions;
2668 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2669 CLASS, as specified in AMOP;
2670 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2671 on the order returned by the primary method for classes of
2672 class STANDARD-CLASS;
2673 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2674 * fixed some bugs shown by Paul Dietz' test suite:
2675 ** DOLIST puts its body in TAGBODY;
2676 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2678 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2679 value producing form;
2680 ** if more variables are given to PROGV than values, extra
2681 variables are bound and made to have no value;
2682 ** NSUBSTITUTE on list arguments gets the right answer with
2684 ** ELT signals an error of type TYPE-ERROR when the index argument
2685 is not a valid sequence index;
2686 ** LOOP signals (at macroexpansion time) an error of type
2687 PROGRAM-ERROR when duplicate variable names are found;
2688 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2689 ** FUNCALL of special-operators now cause an error of type
2691 ** PSETQ now works as required in the presence of side-effecting
2692 symbol-macro places;
2693 ** NCONC accepts any object as its last argument;
2694 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2696 ** loop-for-as-package does not require a package to be explicitely
2698 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2700 * fixed bug 166: compiler preserves "there is a way to go"
2701 invariant when deleting code.
2702 * fixed bug 172: macro lambda lists with required arguments after
2703 &REST arguments now cause an error to be signalled. (thanks to
2705 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2706 bug. (thanks to Matthew Danish)
2707 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2709 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2710 arguments are lexically rebound. (thanks to Gerd Moellmann and
2712 * fixed bug 194: error messages are now more informative when there
2713 is no primary method applicable in a call to a generic
2714 function. (thanks to Gerd Moellmann)
2715 * fixed bug in command line argument checking (thanks to Julian
2717 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2718 (thanks to Matthew Danish)
2719 * incremented fasl file version number, because of the
2720 SXHASH-related changes in the layout of CLOS data structures
2722 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2723 * minor incompatible change: The runtime (the Unix executable named
2724 "sbcl") is now much pickier about the .core files it will load.
2725 Essentially it now requires .core files to descend from the same
2726 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2727 as the runtime does. (The intent is to prevent the crashes which
2728 can occur, and which can even be reported as mysterious failures,
2729 when people patch the sources or change the build parameters
2730 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2731 sbcl and .core files.)
2732 * fixed bug: VALUES-LIST is no longer optimized away.
2733 * fixed bug 142: The FFI conversion of C string values to Lisp
2734 string values no longer conses excessively. (thanks to Nathan
2735 Froyd porting Raymond Toy's fix to CMU CL)
2736 * began to systematize and improve MOP conformance in PCL (thanks to
2737 Nathan Froyd, Gerd Moellman and Pierre Mai):
2738 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2740 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2741 ** FINALIZE-INHERITANCE is now called on class finalization;
2742 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2743 argument precedence order.
2744 * fixed bug 202: The compiler no longer fails on functions whose
2745 derived types contradict their declared type.
2746 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2747 so it can be non-toplevel.
2748 * The fasl file version number has changed (because of the new
2749 implementation of DEFMACRO).
2750 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2751 safe code, that any length requirement by their type-specifier
2752 argument is valid. The exceptions to this are described in bug
2754 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2755 functions now check the types of their inputs as required by ANSI.
2756 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2757 introduced symbol is DECLAREd to be SPECIAL.
2758 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2759 * fixed bug: :COUNT argument to sequence functions may be negative
2760 * fixed bug: body of DO-SYMBOLS may contain declarations
2761 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2762 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2763 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2764 symbol macro only once
2765 * fixed printing of call frame when argument list is unavailable
2766 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2767 * compiler no longer signals WARNING on unknown keyword
2770 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2771 * A beta-quality port to the mips architecture running Linux,
2772 based on the old CMUCL backend, has been made. It has been tested
2773 on a big-endian kernel, and works sufficiently well to be able to
2774 rebuild itself; it has not been tested in little-endian mode.
2775 * fixed an inconsistency between gencgc.c and purify.c which made
2776 dumping/loading .core files unreliable
2777 * fixed bug 120a: The compiler now deals correctly with IFs where
2778 the consequent is the same as the alternative, instead of
2779 misderiving the return type. (thanks to Alexey Dejneka)
2780 * fixed bug 113: Logical pathnames are now dumpable (the logical
2781 host is resolved at load-time, throwing an error if it is not
2783 * fixed bug 174: FORMAT's error message is slightly clearer when a
2784 non-printing character is used in a format directive.
2785 * fixed several bugs in compiler checking of type declarations, i.e.
2786 violations of the Python "declarations are assertions" principle
2787 (thanks to Alexey Dejneka)
2788 * fixed several bugs in PCL's error checking (thanks to Gerd
2790 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2792 * fixed bug in compilation of functions as first class values
2793 (thanks to Antonio Martinez-Shotton)
2794 * The compiler's handling TYPE-ERRORs which it can prove will
2795 inevitably happen at runtime has been cleaned up and corrected
2796 in several ways. (thanks to Alexey Dejneka)
2797 * improved argument type checking for various basic arithmetic
2798 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2799 lavished on them in the past that they can be compiled in many
2800 ways in different special cases
2801 * fixed bug 181: compiler checks validity of user supplied type
2803 * cleaned up code flushing in optimization: Function calls which
2804 should signal errors for safety purposes (e.g. which ANSI says
2805 should signal errors when their arguments are of incorrect type)
2806 are no longer optimized away.
2807 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2808 * incremented fasl file version number, because changes in the
2809 implementation of sequence functions like COERCE caused
2810 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2811 in old inline expansions) to become undefined. (Actually these
2812 changes were later undone, so we might very well be binary
2813 compatible with 0.7.7 after all, but leaving the version number
2814 incremented seemed like the simplest and most conservative
2817 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2818 * An alpha-quality port to the parisc architecture running Linux,
2819 based on the old CMUCL backend, has been made. This, even more so
2820 than the other backends, should be considered still a work in
2821 progress; known problems include that the Linux kernel in 64-bit
2822 mode does not propagate the correct sigcontext structure to
2823 userspace, and consequently SBCL on a parisc64 kernel will not
2825 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2826 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2827 suppressed.) Also now that the compiler is looking at declarations
2828 in the environment, it checks optimization declarations as well,
2829 and suppresses inlining when (> DEBUG SPEED).
2830 * More fixes have been made to treatment of floating point exception
2831 treatment and other Unix signals. In particular, floating point
2832 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2833 * The detection and handling of control stack exhaustion (infinite
2834 or very deeply nested recursion) has changed. Stack exhaustion
2835 detection is now done by write-protecting pages at the OS level
2836 and applies at all optimization settings; when found, a
2837 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2838 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2839 be caught using IGNORE-ERRORS.
2840 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2841 correctly when transforming recursive LABELS functions to LETs.
2842 (thanks to Alexey Dejneka)
2843 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2844 that are names of constants or global variables.
2845 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2846 alien routines with docstrings.
2847 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2848 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2850 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2851 (thanks to Eric Marsden)
2852 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2853 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2854 * Bug fix: The SPARC backend can now compile functions involving
2855 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2856 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2857 to a functions expecting a PATHNAME-DESIGNATOR.
2858 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2859 lambda lists. (thanks to Alexey Dejneka)
2860 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2861 functions so that EXPORTing the name of the class doesn't cause
2862 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2863 to Antonio Martinez for reporting this)
2864 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2865 constant-folded without failing an assertion. (thanks to Einar
2866 Floystad Dorum for reporting this)
2867 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2868 types are now dealt with more correctly by the compiler.
2869 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2870 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2871 behaviour analogous to RENAME-FILE. This puts its behaviour more
2872 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2874 * The fasl file version number has changed again. (because of the
2875 bug fix involving the names of PCL MAKE-INSTANCE functions)
2877 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2878 * bug fix: Floating point exceptions are treated much more
2879 consistently on the x86/Linux and PPC/Linux platforms.
2880 * Array initialization with :INITIAL-ELEMENT is now much faster for
2881 cases when the compiler cannot open code the array creation, but
2882 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2883 array accesses have also seen a speed increase.
2884 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2885 specified. (This was at the root of some bad interactions between
2886 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2888 * bug fix: Internal error arguments for undefined functions are now
2889 computed correctly on the PPC/Linux platform.
2890 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2891 Raymond Toy's patch for CMU CL)
2892 * Support for the Solaris 9 operating environment has been included
2893 (thanks to Daniel Merritt)
2894 * A very ugly but hopefully complete draft of the missing FFI chapter
2895 of the manual has been created by reformatting the corresponding
2896 CMU CL manual chapter into (currently very ugly and incoherent)
2897 DocBook and bringing it up to date for SBCL behavior. Thus, the
2898 manual is now essentially complete, at least by my extreme
2899 once-and-only-once standards, whereby it's acceptable to refer to
2900 the doc strings of SB-EXT functions as the primary documentation.
2901 * The fasl file version number has changed again, due to cleanup of
2902 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2903 is no longer a static symbol.)
2905 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2906 * SBCL now builds with OpenMCL (version 0.12) as the
2907 cross-compilation host; also, more progress has been made toward
2908 bootstrapping under CLISP.
2909 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2911 * bug 158 fixed: The compiler can now deal with integer loop
2912 increments different from 1; fixing this turned out also to fix
2914 * bug 169 fixed: no more bogus warnings about using lexical bindings
2915 despite the presence of perfectly good SPECIAL declarations (thanks
2916 to David Lichteblau)
2917 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2918 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2919 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2921 * bug 180 fixed: Method combination specifications no longer ignore
2922 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2923 * bug fix: Structure type predicate functions now check their argument
2924 count as they should.
2925 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2926 correctly. (thanks to Pierre Mai)
2927 * minor incompatible change: The --noprogrammer option is deprecated
2928 in favor of the new --disable-debugger option, which is very similar.
2929 (The major difference is that it takes effect at a slightly different
2930 time at startup, causing handling of errors in --sysinit and
2931 --userinit files will be affected differently.) The
2932 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2933 been added to allow this functionality to be controlled from ordinary
2934 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2935 maintainers, who might want to run non-interactive scripts to
2936 build SBCL cores which will later be used interactively.)
2937 * minor incompatible change: The LOAD function no longer, when given
2938 a wild pathname to load, loads all files matching that pathname.
2939 Instead, an error of type FILE-ERROR is signalled.
2941 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2942 * bug 147 fixed: The compiler preserves its block link/count
2943 invariants more correctly now so that it doesn't crash. (thanks
2945 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2947 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2948 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2949 no-value-for-this-character as NIL (as specified by ANSI).
2950 * HOST-NAMESTRING on physical pathnames now returns a string that is
2951 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2952 (thanks to Christophe Rhodes)
2953 * The Alpha port handles icache flushing more correctly. (thanks to
2955 * More progress has been made toward bootstrapping under CLISP. (thanks
2956 to Christophe Rhodes)
2957 * The fasl file format has changed again, because dynamic loading
2958 on OpenBSD (which has non-ELF object files) motivated some cleanups
2959 in the way that foreign symbols are transformed and passed around.
2960 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2961 is no longer treated as whitespace by the reader, but instead as
2962 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2963 |A
\7fB|, instead of A as it used to.
2965 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2966 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2967 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2968 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2969 this as of 0.7.1.45, but was left out of the previous news section
2970 (thanks to Dan Barlow)
2971 * SBCL now runs on the Solaris operating system on SPARC architectures
2972 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2973 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2974 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2975 * SPARC backend cleanups, allowing builds of cores optimized for V8
2976 and V9 SPARCS, and also emission of code targeted to a particular
2977 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2979 * SBCL is closer to bootstrapping under CLISP, thanks to various
2980 fixes by Christophe Rhodes.
2981 * The fasl file format has changed again, to allow the compiler's
2982 INFO database to support symbol macros.
2983 * The user manual (in doc/) is formatted into HTML more nicely.
2984 (thanks to coreythomas)
2985 * The system is smarter about SUBTYPEP relationships, especially
2986 those involving NOT types (including types such as ATOM which are
2987 represented internally using NOT types). Thus SUBTYPEP is less
2988 likely to return (VALUES NIL NIL) in general, and in particular
2989 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2991 * The fasl file format has changed again, because the internal
2992 representation of types now includes a new slot to support the new
2993 SUBTYPEP-of-NOT-types logic.
2994 * (not a change in the main branch of SBCL, but a related prototype
2995 which can hopefully be merged into the main branch of SBCL in the
2996 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2997 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2998 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2999 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3000 dispatch macro character combinations. (thanks to Alexey Dejneka)
3001 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3002 that superficially look like logical namestrings correctly.
3003 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3005 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3006 default supplied-p) for &optional and &key arguments. (thanks to
3009 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3010 * incompatible change: The compiler is now less aggressive about
3011 tail call optimization, doing it only when (> SPACE DEBUG) or
3012 (> SPEED DEBUG). (This is an incompatible change because there are
3013 programs which relied on the old CMU-CL-style behavior to optimize
3014 away their unbounded recursion which will now die of stack overflow.)
3015 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3016 for non-GENCGC systems has been increased to 20M (since that
3017 seems much closer to the likely performance optimum for modern
3018 systems than the old 4M value was)
3019 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3020 and *DEBUG-PRINT-LEVEL*
3021 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3022 of CMU CL's support for SPARC, and various endianness and other
3023 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3024 * new syntactic sugar for the Unix command line: --load foo.bar is now
3025 an alternate notation for --eval '(load "foo.bar")'.
3027 ** The system now detects stack overflow and handles it gracefully,
3028 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3029 optimization settings. (This is a good thing in general, and
3030 its introduction in this version should be particularly timely
3031 for anyone whose code fails because of suppression of tail
3033 ** The system now hunts for the C variable "environ" in a more
3034 devious way, to avoid segfaults when the C library version differs
3035 between compile time and run time. (thanks to Christophe Rhodes)
3036 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3037 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3038 ** The compiler no longer issues bogus style warnings for undefined
3039 classes in the same source file as the DEFCLASSes which defined
3040 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3042 ** fixes in CONDITION class precedence list for undefined function
3043 errors (thanks to Alexei Dejneka)
3044 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3045 correctly. (thanks to Dan Barlow)
3046 ** portability fixes aiming at bootstrapping under CLISP (thanks
3047 to Dave McDonald and Christophe Rhodes)
3048 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3049 ** fixes in type translation and and type inference (thanks to
3051 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3052 ** various fixes in the new ports (thanks to Dan Barlow)
3053 * several changes related to debugging:
3054 ** suppression of tail recursion, as noted above
3055 ** stack overflow detection, as noted above
3056 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3057 is now the default. (For some time encapsulation has been more
3058 reliable than the breakpoint-based :ENCAPSULATE NIL
3059 implementation, at least on X86 systems; and I just noticed that
3060 encapsulation also seems closer to the spirit of the ANSI
3063 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3065 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3066 up properly again. (There was a packaging bug in 0.7.0 which
3067 left their definitions in SB-SYS::LOAD-FOREIGN and
3068 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3069 vital for most things which interface to C-level interfaces,
3070 like extensions working with sockets or databases or
3071 Perl-compatible regexes or whatever, and the need to fix
3072 this bug is the main reason that 0.7.1 was released so
3074 ** DEFGENERIC is now choosier about the methods it redefines, so that
3075 reLOADing a previously-LOADed file containing DEFGENERICs does
3076 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3077 cycle now works as it should. (thanks to Alexey Dejneka)
3078 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3079 fixed by Christophe Rhodes. (He actually submitted this patch
3080 months ago, and I delayed until after 0.7.0.)
3081 ** Bug 111 (internal compiler confusion about runtime checks on
3082 FUNCTION types) was fixed by Alexey Dejneka.
3083 * Some internal cleanups (getting rid of variables which aren't
3084 needed now that the byte interpreter is gone) caused the fasl
3085 file format number to change again.
3087 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3088 * major incompatible change: The default fasl file extension, i.e. the
3089 default extension for files produced by COMPILE-FILE, has changed
3090 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3092 ** There are many changes in the implementation of the compiler.
3093 SBCL is now essentially a compiler-only implementation of ANSI
3094 Common Lisp. EVAL still "interprets" a few special cases, but
3095 almost all the interesting cases are handled by creating
3096 a LAMBDA expression, calling COMPILE on it, then calling
3097 FUNCALL on the result.
3098 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3099 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3100 Since the code is newer, there might still be some new bugs
3101 (though not as many as before Martin Atzmueller's fixes:-). But
3102 the new code is substantially simpler and clearer, and hopefully
3103 any remaining bugs will be simpler, less fundamental, and more
3104 fixable then the bugs in the old code.
3105 ** The revised compiler is still a little unsteady on its feet.
3107 *** The debugging information it produces (particularly the names
3108 of FUNCTION objects) is sometimes much less useful than what
3109 the old compiler produced.
3110 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3111 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3112 within a MACROLET) has been temporarily weakened.
3113 ** There are new compiler optimizations for various functions:
3114 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3115 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3116 *** the math functions TRUNCATE, FLOOR, and CEILING
3117 *** the function-of-all-trades COERCE
3118 Mostly these should be transparent, but there's one
3119 potentially-annoying problem (bug 117): when the compiler
3120 inline-expands a function and does type analysis on the result,
3121 it can create control paths which have type mismatches, and
3122 when it can't prove that those control paths aren't taken,
3123 it will issue WARNINGs about the type mismatches. This is
3124 a particular problem in practice for the new sequence functions.
3125 It's not clear how this should be fixed, and for now, a
3126 workaround is given in the entry for 117 in the BUGS file.
3127 ** (Because of the interaction between the two previous items --
3128 occasional inlining problems and new inline expansions -- some
3129 of the new sequence function optimizations won't really kick in
3130 completely until debugging information, and then inlining, are
3131 straightened out in some future version.)
3132 * minor incompatible changes:
3133 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3134 DIRECTORY no longer implicitly promotes NIL slots of its
3135 pathname argument to :WILD. In particular, when you ask for the
3136 contents of a directory (which you used to be able to do without
3137 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3138 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3139 ** changes in behavior that ANSI explicitly defines to be
3140 implementation dependent:
3141 *** The new compiler-only implementation still conforms with ANSI,
3142 but acts a little different than before. Besides the obvious
3143 changes in performance tradeoffs (that the cost per form passed
3144 to EVAL has gone up, and the cost per form executed by EVAL
3145 has gone down), the behavior of the system changes a little
3146 because there are no longer any interpreted function objects.
3147 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3148 e.g. doing COMPILE on the output of interactive DEFUN is
3150 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3152 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3153 has changed from (:CL) to NIL.
3154 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3155 e.g. #\Newline and #\Tab, don't have names specified in the
3156 ANSI Common Lisp standard, is now based on their ASCII symbolic
3157 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3158 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3159 are no longer used for output.
3160 ** changes in internal implementation constants:
3161 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3162 4 million. (If your application spends a lot of time GCing and
3163 you have a lot of RAM, you might want to experiment with
3164 increasing it even more.)
3165 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3166 However, almost all old code should still continue to work without
3167 immediate update, as SB-C-CALL is now a (deprecated) nickname
3169 ** Old operator names in the style DEF-FOO are now deprecated in
3170 favor of new corresponding names DEFINE-FOO, for consistency with
3171 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3172 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3173 internal symbols, but a few supported extensions like
3174 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3175 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3176 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3177 etc. as you get deeper into recursive calls to the debugger
3178 command loop, instead of the old "5]", "5]]", "5]]]"
3179 sequence. (I was motivated to do this when squabbles between
3180 ILISP and SBCL left me very deeply nested in the debugger. In the
3181 short term, this change will probably provoke more ILISP/SBCL
3182 squabbles, but hopefully it will be an improvement in the long run.)
3183 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3184 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3185 function, since declaiming return types involving aliens is
3186 (1) annoyingly messy to do by hand and (2) vital to efficient
3187 compilation of code which calls such functions.
3188 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3189 longer reexported by the SB-EXT package. They're solely useful
3190 for alien code, so it seems more logical that you should get
3191 them from the SB-ALIEN package, not in SB-EXT.
3192 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3193 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3194 features. Instead, the code that they used to control is always
3195 built into the system.
3196 * many other bug fixes
3197 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3198 advantage of the new EVAL-WHEN stuff and to clean them up in
3199 general, and they are now more ANSI-compliant in a number of
3200 ways. Martin Atzmueller is responsible for a lot of this.
3201 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3203 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3204 *** correct ERROR type for various file operations
3205 *** some fixes for Lisp streams
3206 *** DEFMETHOD syntax checking
3207 *** changing old weird representation of debug information as
3208 strings (which, among their other deficiencies, don't transform
3209 correctly when you rename packages, and don't change their
3210 print representation when you change things like *PACKAGE*
3211 and *PRINT-LENGTH*) to symbols and lists of symbols
3212 He also made several improvements and fixed several bugs in DESCRIBE.
3213 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3215 *** misbehavior of WRITE-STRING/WRITE-LINE
3216 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3217 and several other LOOP problems as well
3218 *** DIRECTORY when similar filenames are present
3219 *** DEFGENERIC with :METHOD options
3220 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3221 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3222 *** argument ordering in FIND with :TEST option
3223 *** mishandled package designator argument in APROPOS-LIST
3224 *** various problems in the backquote readmacro
3225 *** a bug in APROPOS
3226 *** probably some others that I'm not describing very well here,
3227 since the CVS log documents them by reference to sbcl-devel
3228 messages, and the SourceForge archives aren't working well.:-(
3229 ** Dan Barlow improved the Alpha port (and is making progress on the
3230 PPC port, for those of you who think different).
3231 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3232 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3233 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3234 continues to work on the SPARC port (for those of you in a position
3235 to look down upon our little PC-compatible boxes from a great height).
3236 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3237 and uses that copy, rather than the current dynamic value, when
3238 it's trying to decide whether to truncate output. Thus e.g.
3239 (let ((*print-lines* 50))
3240 (pprint-logical-block (stream nil)
3242 (let ((*print-lines* 8))
3243 (print (aref possiblybigthings i) stream)))))
3244 should now truncate the logical block only at 50 lines, instead of
3245 often truncating it at 8 lines, as it did before.
3246 * The doc/cmucl/ directory, containing old CMU CL documentation
3247 from the time of the fork, is no longer part of the base system.
3248 SourceForge has shut down its anonymous FTP service, and with it
3249 my original plan for distributing the old CMU CL documentation
3250 there. For now, if you need these files you can download an old
3251 SBCL source release and extract them from it.
3252 * The fasl file version number changed again, for dozens of reasons,
3253 some of which are apparent above.
3255 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3256 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3257 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3258 patch, so that DISASSEMBLE output is much nicer.
3259 * The code in the SB-PROFILE package now seems reasonably stable.
3260 I still haven't decided what the final interface should look like
3261 (I'd like PROFILE to interact cleanly with TRACE, since both
3262 facilities use function encapsulation) but if you have a need
3263 for profiling now, you can probably use it successfully with
3264 the current CMU-CL-style interface.
3265 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3266 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3267 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3268 link now returns the dangling link itself, and for similar
3269 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3270 link itself. (In these cases the old code signalled an error and
3271 looped endlessly, respectively.) Thus, DIRECTORY now works even
3272 in the presence of dangling and cyclic symbolic links.
3273 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3274 is now a supported extension again, since the consensus on
3275 sbcl-devel was that it can be useful for ordinary development
3276 work, not just for debugging SBCL itself.
3277 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3278 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3279 that currently-defined functions might be redefined later with
3280 different return types.
3281 * Hash tables can be printed readably, as inspired by CMU CL code
3282 of Eric Marsden and SBCL code of Martin Atzmueller.
3283 * better error handling in CLOS method combination, thanks to
3284 Martin Atzmueller porting Pierre Mai's CMU CL patches
3285 * more overflow fixes for >16Mbyte I/O buffers
3286 * A bug in READ has been fixed, so that now a single Ctrl-D
3287 character suffices to cause end-of-file on character streams.
3288 In particular, now you only need one Ctrl-D at the command
3289 line (not two) to exit SBCL.
3290 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3292 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3293 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3294 does the right thing.
3295 * The implementation of some type tests, especially for CONDITION
3296 types, is now tidier and maybe faster, due to CMU CL code
3297 originally by Douglas Crosher, ported by Martin Atzmueller.
3298 * Some math functions have been fixed, and there are new
3299 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3300 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3301 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3302 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3303 likely to be built into sbcl-0.7.0.)
3304 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3305 CL, ported to SBCL by Martin Atzmueller)
3306 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3307 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3308 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3309 * There's a new slam.sh hack to shorten the edit/compile/debug
3310 cycle for low-level changes to SBCL itself, and a new
3311 :SB-AFTER-XC-CORE target feature to control the generation of
3312 the after-xc.core file needed by slam.sh.
3313 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3314 COMPILE-FILE is no longer supported, so that now every function
3315 gets an entry point, so that block compilation looks a little
3316 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3317 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3318 returns the number of bytes consed since the system started,
3319 rather than the number consed since the first time the function
3320 was called. (The new definition parallels ANSI functions like
3321 CL:GET-INTERNAL-RUN-TIME.)
3322 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3323 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3324 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3325 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3326 * Fasl file version numbers are now independent of the target CPU,
3327 since historically most system changes which required version
3328 number changes have affected all CPUs equally. Similarly,
3329 the byte fasl file version is now equal to the ordinary
3332 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3333 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3334 is no longer recognized. I apologize for this, because it was
3335 listed in SB-EXT as a supported extension, but I found that
3336 its existing behavior was poorly specified, as well as incorrectly
3337 specified, and it looked like too much of a mess to straighten it
3338 out. I have enough on my hands trying to get ANSI stuff to work..
3339 * many patches ported from CMU CL by Martin Atzmueller, with
3340 half a dozen bug fixes in pretty-printing and the debugger, and
3341 half a dozen others elsewhere
3342 * fixed bug 13: Floating point infinities are now supported again.
3343 They might still be a little bit flaky, but thanks to bug reports
3344 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3345 as flaky as they were.
3346 * The --noprogrammer command line option is now supported. (Its
3347 behavior is slightly different in detail from what the old man
3348 page claimed it would do, but it's still appropriate under the
3349 same circumstances that the man page talks about.)
3350 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3351 are now supported, and enabled by default. Thus, the compiler can
3352 handle many floating point and complex operations much less
3353 inefficiently. (Thus e.g. you can implement a complex FFT
3355 * The compiler now detects type mismatches between DECLAIM FTYPE
3356 and DEFUN better, and implements CHECK-TYPE more correctly, and
3357 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3358 thanks to patches from Martin Atzmueller.
3359 * various fixes to make the cross-compiler more portable to
3360 ANSI-conforming-but-different cross-compilation hosts (notably
3361 Lispworks for Windows, following bug reports from Arthur Lemmens)
3362 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3363 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3364 fixed thanks to Pierre Mai's CMU CL patches.
3365 * fixes to keep the system from overflowing internal counters when
3366 it tries to use i/o buffers larger than 16M bytes
3367 * fixed bug 45a: Various internal functions required to support
3368 complex special functions have been merged from CMU CL sources.
3369 (When I was first setting up SBCL, I misunderstood a compile-time
3370 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3371 * improved support for type intersection and union, fixing bug 12
3372 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3373 more obscure bugs as well
3374 * some steps toward byte-compiling non-performance-critical
3375 parts of the system, courtesy of patches from Martin Atzmueller
3376 * Christophe Rhodes has made some debian packages of sbcl at
3377 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3378 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3379 stable, but are nonetheless usable. When he's ready, I'd be happy
3380 to add them to the SourceForge "File Releases" section. (And if
3381 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3382 * new fasl file format version number (because of changes in
3383 internal representation of (OR ..) types to accommodate the new
3384 support for (AND ..) types, among other things)
3386 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3387 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3389 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3391 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3392 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3393 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3394 * DESCRIBE now works on structure objects again.
3395 * Most function call argument type mismatches are now handled as
3396 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3397 know whether the function will be redefined before the call is
3398 executed. (The compiler could flag local calls with full WARNINGs,
3399 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3400 it doesn't keep track of enough information to know whether calls
3401 are local in this sense.)
3402 * Compiler output is now more verbose, with messages truncated
3403 later than before. (There should be some supported way for users
3404 to override the default verbosity, but I haven't decided how to
3405 provide it yet, so this behavior is still controlled by the internal
3406 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3407 src/compiler/ir1util.lisp.)
3408 * Fasl file format version numbers have increased again, because
3409 support for the Gray streams extension changes the layout of the
3410 system's STREAM objects.
3411 * The Gray subclassable streams extension now works, thanks to a
3412 patch from Martin Atzmueller.
3413 * The full LOAD-FOREIGN extension (not just the primitive
3414 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3415 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3416 but like most other programs, it defaults to copying the Unix
3417 environment from the original process instead of starting the
3418 new process in an empty environment.
3419 * Extensions which manipulate the Unix environment now support
3420 an :ENVIRONMENT keyword option which doesn't smash case or
3421 do other bad things. The CMU-CL-style :ENV option is retained
3422 for porting convenience.
3423 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3424 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3426 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3428 * A patch from Martin Atzmueller seems to have solved the SIGINT
3429 problem, and as far as we know, signal-handling now works cleanly.
3430 (If you find any new bugs, please report them!)
3431 * The system no longer defaults Lisp source file names to types
3432 ".l", ".cl", or ".lsp", but only to ".lisp".
3433 * The compiler no longer uses special default file extensions for
3434 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3435 seems to expect a single default extension for all compiled code,
3436 and there's no compelling reason to try to stretch the standard
3437 to allow two different extensions.) Instead, byte-compiled files
3438 default to the same extension as native-compiled files.
3439 * Fasl file format version numbers have increased again, because
3440 a rearrangement of internal implementation packages made some
3441 dumped symbols in old fasl files unreadable in new cores.
3442 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3443 many fewer weird special cases.
3444 * Bug #17 (differing COMPILE-FILE behavior between logical and
3445 physical pathnames) has been fixed, and some related misbehavior too,
3446 thanks to a patch from Martin Atzmueller.
3447 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3448 by Tim Moore, ported to SBCL by Martin Atzmueller.
3449 * Martin Atzmueller fixed several filesystem-related problems,
3450 including bug #36, in part by porting CMU CL patches, which were
3451 written in part by Paul Werkowski.
3452 * More compiler warnings in src/runtime/ are gone, thanks to
3453 more patches from Martin Atzmueller.
3454 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3457 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3459 * DESCRIBE now works on CONDITION objects.
3460 * The debugger now handles errors which arise when trying to print
3461 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3463 * The build system now uses an additional file, customize-target-features.lisp,
3464 to allow local modifications to the target *FEATURES* list. (The point of
3465 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3466 debugging features enabled, without having to worry about propagating it
3467 into everyone's system when I do a "cvs update".) When no
3468 customize-target-features.lisp file exists, the target *FEATURES* list
3469 should be constructed the same way as before.
3470 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3471 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3472 warn about behavior which is undefined under the ANSI spec. Note: This
3473 is specified by ANSI, but it's not very popular with programmers.
3474 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3475 macro in the SBCL sources for an example of a workaround which you
3476 might use to make portable ANSI-standard code which does what you want.
3477 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3478 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3479 when it's used as a non-toplevel form. (This is required in order
3480 to implement the DEFCONSTANT-EQX macro.)
3481 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3482 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3483 funny things, due to bugs in the implementation of EVAL-WHEN
3484 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3485 non-toplevel DEFCONSTANTs are uncommon.)
3486 * The core file version number and fasl file version number have been
3487 incremented, because the old noncompliant DEFCONSTANT behavior involved
3488 calling functions which no longer exist, and because I also took the
3489 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3490 * fixed bug 1 (error handling before read-eval-print loop starts), and
3491 redid debugger restarts and related debugger commands somewhat while
3493 ** The QUIT debugger command is gone, since it did something
3494 rather different than the SB-EXT:QUIT command, and since it never
3495 worked properly outside the main toplevel read/eval/print loop.
3496 Invoking the new TOPLEVEL restart provides the same functionality.
3497 ** The GO debugger command is also gone, since you can just invoke
3498 the CONTINUE restart directly instead.
3499 ** The TOP debugger command is also gone, since it's redundant with the
3500 FRAME 0 command, and since it interfered with abbreviations for the
3502 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3504 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3505 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3506 some porting work by Martin Atzmueller)
3507 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3508 with a boatload of patches which he ported to SBCL. Now that those
3510 ** The system tries to make sure that its low-priority messages
3511 are prefixed by semicolons, to help people who like to use
3512 syntax highlighting in their ILISP buffer. (This patch
3513 was originally due to Raymond Toy.)
3514 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3515 patches originally written by Raymond Toy.
3516 ** The compiler understands coercion between single-value and
3517 multiple-VALUES type expressions better, getting rid of some very
3518 weird behavior, thanks to patches originally by Robert MacLachlan
3519 and Douglas Crosher.
3520 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3521 lambda lists, thanks to a patch originally by Pierre Mai.
3522 ** The system no longer bogusly warns about "abbreviated type
3524 ** The compiler gets less confused by inlining and RETURN-FROM,
3525 thanks to some patches originally by Tim Moore.
3526 ** The system no longer hangs when dumping circular lists to fasl
3527 files, thanks to a patch originally from Douglas Crosher.
3528 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3529 undefined function error.
3530 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3531 in structure slot initforms) for now, documented workaround instead:-|
3532 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3533 * fixed bug 5: added stubs for various Gray stream functions called
3534 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3535 installed, at least appropriate type errors are generated
3536 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3537 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3538 consistently in DEFMETHOD forms.
3539 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3540 it doesn't seem to affect SBCL after all
3541 * The C runtime system now builds with better optimization and many
3542 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3544 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3546 * The system is now under CVS at SourceForge (instead of the
3547 CVS repository on my home machine).
3548 * The new signal handling code has been tweaked to treat register
3549 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3550 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3551 downstream, e.g. in debug-int.lisp, has implicit dependencies
3552 on the unsignedness of integer representation of machine words,
3553 and that caused the system to bomb out with infinite regress
3554 when trying to recover from type errors involving signed values,
3555 e.g. (BUTLAST '(1 2 3) -1).)
3556 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3557 (This was one of the bugs Peter Van Eynde reported back in July.)
3558 * The system now uses code inspired by Colin Walters' O(N)
3559 implementation of MAP (from the cmucl-imp@cons.org mailing
3560 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3561 inline the MAP operation, and there is more than one
3562 sequence argument to the MAP call (so that it can't just
3563 do ETYPECASE once and for all based on the type of the
3564 single sequence argument). (The old non-inline implementation
3565 of the general M-argument sequence-of-length-N case required
3566 O(M*N*N) time when any of the sequence arguments were LISTs.)
3567 * The QUIT :UNIX-CODE keyword argument has been renamed to
3568 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3569 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3571 * Raymond Wiker's patches to port dynamic loading from Linux to
3572 FreeBSD have been added.
3573 * The BUGS file is now more nearly up to date, thanks in large part
3574 to Martin Atzmueller's review of it.
3575 * The debugger now flushes standard output streams before it begins
3576 its output ("debugger invoked" and so forth).
3577 * The core version number and fasl file version number have both
3578 been incremented, because of incompatible changes in the layout
3580 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3581 used to not be called for a saved Lisp image.)
3582 * Martin Atzmueller's version of a patch to fix a compiler crash,
3583 as posted on sbcl-devel 13 September 2000, has been installed.
3584 * Instead of installing Martin Atzmueller's patch for the
3585 compiler transform for SUBSEQ, I deleted the compiler transform,
3586 and transforms for some similar consing operations.
3587 * A bug in signal handling which kept TRACE from working on OpenBSD
3589 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3590 away by constant folding
3591 * The system now defines its address space constants in one place
3592 (in the Lisp sources), and propagates them automatically elsewhere
3593 (through GENESIS and the sbcl.h file). Therefore, patching the
3594 address map is less unnecessarily tedious and error-prone. The
3595 Lisp names of address space constants have also been systematized.
3596 * CVS tags like dollar-Header-dollar have been removed from
3597 the sources, because they have never saved me trouble and
3598 they've been source of trouble working with patches and other
3599 diff-related operations.
3600 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3601 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3603 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3605 * The system has been ported to OpenBSD.
3606 * The system now compiles with a simple "sh make.sh" on the systems
3607 that it's supported on. I.e., now you no longer need to tweak
3608 text in the target-features.lisp-expr and symlinks in src/runtime/
3609 by hand, the make.sh takes care of it for you.
3610 * The system is no longer so grossly inefficient when compiling code
3611 involving vectors implemented as general (not simple) vectors (VECTOR T),
3612 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3613 which dares to use the various sequence functions on non-simple
3614 vectors, takes less of a performance hit.
3615 * There is now a primitive type predicate VECTOR-T-P
3616 to test for the (VECTOR T) type, so that e.g.
3617 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3618 can now be compiled with some semblance of efficiency. (The old code
3619 turned the type declaration into a full call to %TYPEP at runtime!)
3620 * AREF on (VECTOR T) is still not fast, since it's still compiled
3621 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3622 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3623 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3624 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3625 without making full calls to SUBTYPEP at runtime.
3626 (Some analogous efficiency issues for non-simple vectors specialized to
3627 element types other than T, or for non-simple multidimensional arrays,
3628 have not been addressed. They could almost certainly be handled the
3629 same way if anyone is motivated to do so.)
3630 * The changes in array handling break binary compatibility, so
3631 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3632 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3633 instead of (VALUES T T).
3634 * By following the instructions that Dan Barlow posted to sbcl-devel
3635 on 2 July 2000, I was able to enable primitive dynamic object
3636 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3637 functionality is not implemented (since it calls ld to resolve
3638 library references automatically, requiring RUN-PROGRAM for its
3639 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3640 try to resolve library references) is now supported.
3641 * The system now flushes the standard output streams when it terminates,
3642 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3643 them at several other probably-convenient times, e.g. in each pass of
3644 the toplevel read-eval-print loop, and after evaluating a form given
3645 as an "--eval" command-line option. (These changes were motivated by a
3646 discussion of stream flushing issues on cmucl-imp in August 2000.)
3647 * The source transform for TYPEP of array types no longer assumes
3648 that an array whose element type is a not-yet-defined type
3649 is implemented as an array of T, but instead punts, so that the
3650 type will be interpreted at runtime.
3651 * There is now some support for cross-compiling in make.sh: each of
3652 the phases of make.sh has its own script. (This should be transparent
3653 to people doing ordinary, non-cross-compile builds.)
3654 * Since my laptop doesn't have hundreds of megabytes of memory like
3655 my desktop machine, I became more motivated to do some items on
3656 my to-do list in order to reduce the size of the system a little:
3657 ** Arrange for various needed-only-at-cold-init things to be
3658 uninterned after cold init. To support this, those things have
3659 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3660 symbols with such names are now uninterned after cold init).
3661 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3662 fasl files for cold load.
3663 ** Remove the old compiler structure pooling code (which used to
3664 be conditional on the target feature :SB-ALLOC) completely.
3665 ** Redo the representation of some data in cold init to be more compact.
3666 (I also looked into supporting byte compiled code at bootstrap time,
3667 which would probably reduce the size of the system a lot, but that
3668 looked too complicated, so I punted for now.)
3669 * The maximum signal nesting depth in the src/runtime/ support code has
3670 been reduced from 4096 to 256. (I don't know any reason for the very
3671 large old value. If the new smaller value turns out to break something,
3672 I'll probably just bump it back up.)
3673 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3675 * Many, many bugs reported by Peter Van Eynde have been added to
3676 the BUGS list; some have even been fixed.
3677 * While enabling dynamic object file loading, I tried to make the
3678 code easier to understand, renaming various functions and variables
3679 with less ambiguous names, and changing some function calling
3680 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3681 * While trying to figure out how to do the OpenBSD port, I tried to
3682 clean up some of the code in src/runtime/. In particular, I dropped
3683 support for non-POSIX signal handling, added various comments,
3684 tweaked the code to reduce the number of compilation warnings, and
3685 renamed some files to increase consistency.
3686 * To support the new automatic configuration functionality in make.sh,
3687 the source file target-features.lisp-expr has been replaced with the
3688 source file base-target-features.lisp-expr and the machine-generated
3689 file local-target-features.lisp-expr.
3690 * fixed a stupid quoting error in make.sh so that using CMU CL
3691 "lisp -batch" as cross-compilation host works again
3693 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3695 * DESCRIBE no longer tries to call itself recursively to describe
3696 bound/fbound values, so that it no longer fails on symbols which are
3697 bound to themselves (like keywords, T, and NIL).
3698 * DESCRIBE now works on generic functions.
3699 * The printer now prints less-screwed-up representations of closures
3700 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3701 * A private symbol is used instead of the :EMPTY keyword previously
3702 used to mark empty slots in hash tables. Thus
3703 (DEFVAR *HT* (MAKE-HASH-TABLE))
3704 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3705 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3706 now does what ANSI says that it should. (You can still get
3707 similar noncompliant behavior if bang on the hash table
3708 implementation with all the symbols you get back from
3709 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3710 This breaks binary compatibility, since tests for equality to
3711 :EMPTY are wired into things like the macroexpansion of
3712 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3714 * There's now a minimal placeholder implementation for CL:STEP,
3715 as required by ANSI.
3716 * An obscure bug in the interaction of the normal compiler, the byte
3717 compiler, inlining, and structure predicates has been patched
3718 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3719 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3721 * Missing ordinary arguments in a macro call are now detected even
3722 when the macro lambda list contains &KEY or &REST.
3723 * The debugger no longer complains about encountering the top of the
3724 stack when you type "FRAME 0" to explicitly instruct it to go to
3725 the top of the stack. And it now prints the frame you request even
3726 if it's the current frame (instead of saying "You are here.").
3727 * As specified by ANSI, the system now always prints keywords
3728 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3730 * The default initial SIZE of HASH-TABLEs is now smaller.
3731 * Type information from CLOS class dispatch is now propagated
3732 into DEFMETHOD bodies, so that e.g.
3733 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3735 is now basically equivalent to
3736 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3737 (DECLARE (TYPE SINGLE-FLOAT X))
3739 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3740 operation, without having to do run-time type dispatch.
3741 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3742 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3743 * Since I don't seem to be making big file reorganizations very often
3744 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3745 is overflowing my ability to conveniently back them up), I've finally
3746 checked the system into CVS. (The CVS repository is on my home system,
3747 not at SourceForge -- putting it on SourceForge might come later.)
3748 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3749 high-level GC-NOTIFY-FOO functions send their output. (There's
3750 still very little control of where low-level verbose GC functions
3751 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3752 less than it used to -- the GC-NOTIFY-FOO functions are now under
3753 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3754 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3755 in only one place in the source code, and propagates it automatically
3756 everywhere that it's needed. Thus e.g. when I bump the version from
3757 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3758 * The C source files now include boilerplate legalese and documentation
3759 at the head of each file (just as the Lisp source files already did).
3760 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3761 to his page will be replaced with a link to his new CLiki service.
3763 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3765 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3766 * The build process now looks for GNU make under the default name "gmake",
3767 instead of "make" as it used to. If GNU make is not available as "gmake"
3768 on your system, you can change this default behavior by setting the
3769 GNUMAKE environment variable.
3770 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3771 can build without error under CMU CL.
3773 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3775 * There is now a partial SBCL user manual (with some new text and some
3776 text cribbed from the CMU CL manual).
3777 * The beginnings of a profiler have been added (starting with the
3778 CMU CL profiler and simplifying and cleaning up). Eventually the
3779 main interface should be through the TRACE macro, but for now,
3780 it's still accessed through vaguely CMU-CL-style functions and macros
3781 exported from the package SB-PROFILE.
3782 * Some problems left over from porting CMU CL to the new
3783 cross-compilation bootstrap process have been cleaned up:
3784 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3785 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3786 part when getting rid of the compiler *BACKEND* stuff.)
3787 ** The value of *NULL-TYPE* was screwed up, because it was
3788 being initialized before the type system knew the final
3789 definition of the 'NULL type. This screwed up several key
3790 optimizations in the compiler, causing inefficiency in all sorts
3791 of places. (I found it because I wanted to understand why
3792 GET-INTERNAL-RUN-TIME was consing.)
3793 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3794 PROCLAIM FTYPE information. Unfortunately this broke binary
3795 compatibility again, since now the forms output by DEFGENERIC
3796 to refer to functions which didn't exist in 0.6.3.
3797 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3798 can use the new (as of 0.6.3) transform for SOME into MAP into
3800 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3801 values to (INTEGER 0 1000000), so that the time code will no longer
3802 occasionally get blown up by Linux returning 1000000 microseconds
3803 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3804 its output conform to the ANSI spec. (Alas, this makes its output
3805 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3806 * A full call to MAP NIL with a single sequence argument no longer conses.
3807 * fixes to problems pointed out by Martin Atzmueller:
3808 * The manual page no longer talks about multiprocessing as though
3809 it were currently supported.
3810 * The ILISP support patches have been removed from the distribution,
3811 because as of version 5.10.1, ILISP now supports SBCL without us
3812 having to maintain patches.
3813 * added a modified version of Raymond Toy's recent CMU CL patch for
3814 EQUALP comparison of HASH-TABLE
3816 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3818 * The system still can't cross-compile itself with
3819 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3820 entails), but at least it can compile more of itself that way
3821 than it used to be able to, and various buglets which were uncovered
3822 by trying to cross-compile itself that way have now been fixed.
3823 * This release breaks binary compatibility again. This time
3824 at least I've incremented the FASL file format version to 2, so that the
3825 problem can be detected reliably instead of just causing weird errors.
3826 * various new style warnings:
3827 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3828 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3829 ** using the lexical binding of a variable named in the *FOO* style
3830 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3831 as specified by ANSI.
3832 * *RANDOM-STATE* is no longer automatically initialized from
3833 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3834 default behavior of the system is to repeat its behavior every time
3835 it's run. If you'd like to change this behavior, you can always
3836 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3837 old convention there was no comparably easy way to get the system to
3838 repeat its behavior every time it was run.
3839 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3840 FTYPE declarations has been removed, in favor of their ANSI
3841 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3842 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3843 the types of their sequence arguments can be determined at compile time.
3844 This is done through a new open code expansion for MAP which eliminates
3845 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3846 argument types can be determined at compile time.
3847 * The optimizer now transforms COERCE into an identity operation when it
3848 can prove that the coerced object is already of the correct type. (This
3849 can be a win for machine generated code, including the output of other
3850 optimization transforms, such as the MAP transform above.)
3851 * Credit information has been moved from source file headers into CREDITS.
3852 * Source file headers have been made more standard.
3853 * The CASE macro now compiles without complaining even when it has
3856 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3858 * (Note that the way that the PCL macroexpansions were rewritten
3859 to accommodate the change in DEFGENERIC below breaks binary
3860 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3861 not run under sbcl-0.6.2. Once we get out of alpha releases,
3862 i.e. hit release 1.0.0, we'll probably try to maintain binary
3863 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3864 and sbcl-1.4.4. Until then, however, it might be fairly common
3865 for maintenance releases to break binary compatibility.)
3866 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3868 * The reporting of unhandled signals has been changed to print some
3869 explanatory text as well as the report form. (Previously only
3870 the report form was printed.)
3871 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3872 it defines, so that the compiler no longer issues undefined function
3873 warnings for compiled-but-not-yet-loaded generic functions.
3874 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3875 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3876 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3877 section of the ANSI spec are supported.
3878 * The "" nickname for the "KEYWORD" package has been removed.
3879 The reader still handles symbol tokens which begin with a package marker
3880 as keywords, but it doesn't expose its mechanism for doing so in the
3881 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3882 * The system now issues STYLE-WARNINGs for contradictory TYPE
3883 proclamations. (Warnings for contradictory FTYPE proclamations would
3884 be nice too, but those can't be done usefully unless the type system
3885 is made smarter about FUNCTION types.)
3886 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3887 systematized, so that "*target-*.lisp is supposed to exist only on the
3888 target and imply that there's a related file which exists on the
3889 host, and *host-*.lisp is supposed to exist only on the host and imply
3890 that there's a related file which exists on the target. This involves a
3891 lot of renaming. Hopefully the acute confusion caused by the renaming
3892 will be justified by the reduction in chronic confusion..
3893 ** runtime-type.lisp -> early-target-type.lisp
3894 ** target-type.lisp -> late-target-type.lisp
3895 ** early-host-format.lisp -> early-format.lisp
3896 ** late-host-format.lisp -> late-format.lisp
3897 ** host-error.lisp -> misc-error.lisp
3898 ** early-error.lisp -> early-target-error.lisp
3899 ** late-error.lisp -> late-target-error.lisp
3900 ** host-defboot.lisp -> early-defboot.lisp
3901 ** code/misc.lisp -> code/target-misc.lisp
3902 ** code/host-misc.lisp -> code/misc.lisp
3903 ** code/numbers.lisp -> code/target-numbers.lisp
3904 ** code/early-numbers.lisp -> numbers.lisp
3905 ** early-host-type.lisp -> early-type.lisp
3906 ** late-host-type.lisp -> late-type.lisp
3907 ** host-typep.lisp -> typep.lisp
3908 ** load.lisp -> target-load.lisp
3909 ** host-load.lisp -> load.lisp
3910 ** host-disassem.lisp -> disassem.lisp
3911 ** host-insts.lisp -> insts.lisp
3912 ** byte-comp.lisp -> target-byte-comp.lisp
3913 ** host-byte-comp.lisp -> byte-comp.lisp
3914 ** host-signal.lisp -> signal.lisp
3915 ** host-defstruct.lisp -> defstruct.lisp
3916 ** late-target-type.lisp -> deftypes-for-target.lisp
3917 Furthermore, several other previously target-only files foo.lisp (e.g.
3918 hash-table.lisp and random.lisp) have been split into a target-and-host
3919 foo.lisp file and a target-only target-foo.lisp file, with their key type
3920 definitions in the target-and-host part, so that the cross-compiler will
3921 know more about target types.
3922 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3923 gone away. In their place are various *BACKEND-FOO* variables
3924 corresponding to the slots of the old structure.
3925 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3926 into the target SBCL has been fixed.
3927 * The chill.lisp system for loading cold code into a running SBCL
3929 * Support for the CMU CL "scavenger hook" extension has been removed.
3930 (It was undocumented and unused in the CMU CL sources that SBCL was
3931 derived from, and stale in sbcl-0.6.1.)
3932 * Various errors in the cross-compiler type system were detected
3933 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3934 (enabling various consistency checks). Many of them were fixed,
3935 but some hard problems remain, so the compiler is back to
3936 running without *TYPE-SYSTEM-INITIALIZED* for now.
3937 * As part of the cross-compiler type system cleanup, I implemented
3938 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3939 * I have started adding UNCROSS calls throughout the type system
3940 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3941 on cross-compiler input files will be able to go away, and various
3943 * CONSTANTP now returns true for quoted forms (as explicitly required
3946 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3948 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3949 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3950 (MAKE-LIST -1). In the longer term, it ought to become true
3951 that declarations are assertions even at SAFETY 1. For now, it's not
3952 quite true even at SAFETY 3, but it's at least more nearly true..
3953 (Note that this change seems to increases the size of the system by
3954 O(5%) and to decrease the speed of the compiler by 20% or more.)
3955 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3956 for the problem of printing dozens of lines of distracting information
3957 about low-level system machinery as part of the top stack frame
3958 on entry to the debugger when an undefined function was called.
3959 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3960 is not reset to COMMON-LISP-USER.
3961 * Compilation of stuff related to dyncount.lisp has been made conditional
3962 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3963 smaller. The various dyncount-related symbols have been moved into
3964 a new "SB-DYNCOUNT" package.
3965 * tty-inspect.lisp has been renamed to inspect.lisp.
3966 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3967 feature has gone away. (When we eventually port to other flavors of
3968 libc and/or Unix, we'll try to make the differences between flavors
3969 invisible at the user level.)
3970 * Various other *FEATURES* tags, and/or their associated conditionals,
3971 have been removed if obsolescent, or given better documentation, or
3972 sometimes given more-mnemonic names.
3974 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3976 * tidied up "make.sh" script
3977 * tidied up system directory structure
3978 * better "clean.sh" behavior
3979 * added doc/FOR-CMUCL-DEVELOPERS
3980 * many many small tweaks to output format, e.g. removing possibly-confusing
3981 trailing #\. character in DESCRIBE-INSTANCE
3982 * (EQUALP #\A 'A) no longer signals an error.
3983 * new hashing code, including EQUALP hashing
3984 * tidied up Lisp initialization and toplevel
3985 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3986 * command line argument processing
3987 * added POSIX-GETENV function to deal with Unix-ish environment variables
3988 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3989 terminating SBCL on EOF
3990 * non-verbose GC by default
3991 * There is no more "sbcl" shell script; the sbcl file is now the C
3992 runtime executable (just like CMU CL).
3993 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3995 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3996 symbol and package information as Lisp native symbol and package objects
3997 instead of strings naming symbols and strings naming packages. This way,
3998 whenever packages are renamed (as in warm init), debug information is
3999 transformed along with everything else.
4000 * tweaked the optimization policy declarations which control the building
4001 of SBCL itself. Now, among other things, the system no longer saves
4002 source location debugging information. (This helps two problems at once
4003 by reducing SBCL size and by keeping SBCL from trying to look for its
4004 sources -- which may not exist -- when reporting errors.)
4005 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4006 debugging and testing purposes
4007 * cleaned up printing, making the printer call PRINT-OBJECT for
4008 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4009 methods, giving nearly-ANSI behavior
4010 * converted almost all special variables to use *FOO* naming convention
4011 * deleted PARSE-TIME functionality, since it can be done portably
4012 * moved some files out of cold init into warm init
4013 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4015 * regularized formatting of source files
4016 * added an install.sh script
4017 * fixed ridiculous memory usage of cross-compiler by making
4018 compiler/alloc.lisp not try to do pooling unless it can hook
4019 itself into the GC of the cross-compilation host. Now the system
4020 builds nicely on my old laptop.
4021 * added :SB-ALLOC in target-features.lisp-expr
4022 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4023 was not implemented)
4024 * re-did condition handling and note reporting in the compiler. Notes
4025 are no longer handled by signalling conditions. Style warnings
4026 and warnings are handled more correctly and reported in such a way
4027 that it's easy to find one or the other in your output (so that you
4028 can e.g. figure out which of many problems caused COMPILE-FILE to
4030 * changed the severity of several compiler warnings from full WARNING
4031 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4032 compiler note reporting so that it doesn't use the condition system
4033 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4034 COMPILE-FILE command)
4035 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4036 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4037 no longer supported, since their implementation was deeply intertwingled
4038 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4039 * removed bogus "support" for compiler macros named (SETF FOO), and
4040 removed the compiler macro for SETF INFO (but only after making a fool
4041 of myself on the cmucl-imp mailing list by posting a bogus patch for
4042 DEFINE-COMPILER-MACRO..)
4043 * Compiled files containing forms which have side effects on the Lisp
4044 reader (such as DEFPACKAGE forms) are now handled more correctly.
4045 (Compiler queuing of top level lambdas has been suppressed by setting
4046 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4047 * deleted various currently-unused source files, e.g. gengc.lisp. They
4048 may be added back at some point e.g. when porting to other architectures,
4049 but until they are it's distracting to distribute them and to try to
4051 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4052 there were so many of them they're just distractions, and UNCROSS is
4053 known to be able to handle the current sources
4054 * moved PROFILE functionality into TRACE, so that it will be clear
4055 how the wrapping and unwrapping of functions when you profile them
4056 interacts with the wrapping and unwrapping of functions when you
4057 trace them. (Actually, the functionality isn't there yet, but at least
4058 the interface specification is there. Hopefully, the functionality will
4059 arrive with some future maintenance release.)
4060 * removed host-oops.lisp
4061 * changed signature of QUIT function to allow UNIX-CODE argument
4062 * fixed READ-SEQUENCE bug
4063 * tweaked verbose GC output so that it looks more like the progress
4064 output that ANSI specifies for functions like LOAD
4065 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4066 * added <http://sbcl.sourceforge.com> to the banner information printed by