1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.1 relative to sbcl-1.0:
3 * new feature: the compiler stores cross-referencing information
4 abount function calls (who-calls), macroexpansion (who-macroexpands)
5 and special variables (who-binds, who-sets, who-references) for code
6 compiled with (< SPACE 3). This information is available through the
8 * new feature: users may subclass SEQUENCE, and have instances of
9 these classes interoperate with standard Common Lisp functions if
10 a number of methods are defined. (This feature is experimental
11 and the interface subject to change based on feedback from SBCL
12 users and the general community)
13 * improvement: sb-sprof traces call stacks to an arbitrary depth on
14 x86/x86-64, rather than the previous fixed depth of 8
15 * bug fix: non-ascii command-line arguments are processed correctly
16 (thanks to Yaroslav Kavenchuk)
17 * bug fix: non-required arguments were not passed correctly when a method
18 defined using DEFMETHOD was called from a mop-generated method using
19 CALL-NEXT-METHOD (reported by Pascal Costanza)
20 * bug fix: recursion is now permitted in accessors through
21 SLOT-UNBOUND. (reported by Pascal Costanza)
22 * bug fix: an error was signaled at startup if the HOME environment
23 variable was defined, but had an empty value (reported by Peter Van Eynde)
24 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
25 signal a PROGRAM-ERROR, not a BUG.
26 * bug fix: SB-POSIX:READDIR works when built with large file support.
27 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
29 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
30 for better automatic indentation support. (thanks to Matt Pillsbury)
31 * optimization: loading generic functions no longer takes O(n^2) time,
32 proportional to the amount of methods in the generic function
33 (reported by Todd Sabin and Jeremy Brown)
34 * optimization: the FIND and POSITION family of sequence functions
35 are significantly faster on arrays whose element types have been
37 * improvements to the Windows port:
38 ** Intermittent heap corruption problems have been fixed. (thanks
39 to Alastair Bridgewater)
40 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
42 ** Lisp is able to unwind foreing exception frames from alien
43 callbacks. (thanks to Alastair Bridgewater)
45 changes in sbcl-1.0 relative to sbcl-0.9.18:
46 * improvement: experimental support for threading on FreeBSD/x86.
47 (thanks to NIIMI Satoshi)
48 * improvement: runtime option --dynamic-space-size can be used
49 to set the size of the dynamic space reserved on startup.
50 * improvement: floating point modes in effect are now saved in
51 core, and restored on startup.
52 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
53 startup, not time since first call to GET-INTERNAL-REAL-TIME.
54 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
55 threads are not running after *SAVE-HOOKS* have run.
56 * improvement: writes to CLOS instance slots are type-checked in code
57 compiled with (SAFETY 3)
58 * improvement: floating-point exception handling on FreeBSD (thanks to
60 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
61 (thanks to Zach Beane)
62 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
64 * improvement: added support for the Shift-JIS external format.
65 (contributed by NIIMI Satoshi)
66 * improvement: callbacks are supported on Linux/PPC. (thanks to
68 * bug fix: compiler bug triggered by a (non-standard) VALUES
69 declaration in a LET* was fixed. (reported by Kaersten Poeck)
70 * bug fix: file compiler no longer confuses validated and already
71 dumped structurres. (reported by Kaersten Poeck)
72 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
73 fill-pointer signals a type-error as required. (thanks to
75 * bug fix: disassemly of funcallable instances works.
76 * bug fix: single stepping on PPC.
77 * bug fix: fix thread-safety problems in the type system (generally
78 manifesting as nonsensical errors like "STRING is a bad type specifier
79 for sequences" or "The value 1 is not of type FIXNUM")
80 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
81 (reported by Josip Gracin)
82 * bug fix: an error is signaled for attempts to displace arrays with
83 incompatible element types (thanks to Mario Mommer)
84 * bug fix: more correct handling of wide characters in debug info
85 (bug reported by Attila Lendvai and fixed by Juho Snellman)
86 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
87 and don't cause extra consing
88 * optimization: MAP and MAP-INTO are significantly faster on vectors
89 whose elements types have been declared.
90 * Improvements to SB-SPROF:
91 ** Support for allocation profiling
92 ** Reduced profiling overhead, especially for long profiling runs
93 * Improvements to the Windows port:
94 ** floating point exceptions are now reported correctly.
95 ** stack exhaustion detection works partially.
96 ** more accurate GET-INTERNAL-REAL-TIME.
97 ** better RUN-PROGRAM behaviour in case of errors in the child
99 ** PROBE-FILE now simplifies pathnames correctly.
100 ** DIRECTORY now works correctly with :WILD-INFERIORS.
101 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
102 1st 1970 (although time timezone data will be unreliable).
103 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
105 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
106 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
107 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
109 * bug fix: two potential GC deadlocks affecting threaded builds.
110 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
111 returns T (reported by Anton Kazennikov)
112 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
113 with non-variable places
114 * bug fix: remove a race condition in the setting of
115 funcallable-instance functions, this should make threaded CLOS
116 code more stable against memory faults.
117 * bug fix: corruption of specials when unbinding is interrupted by an
118 asynchronous unwind (reported by Hannu Koivisto)
119 * improvement: the debugger will now also display local variables that
120 are only used once, for code compiled with a DEBUG optimization quality
123 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
124 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
125 * incompatible change: External-format support for FFI calls. The
126 SB-ALIEN:C-STRING no longer implies an ASCII
127 external-format. Instead the string is subject to external-format
128 conversion. Additionally return values of type SB-ALIEN:C-STRING
129 are by default of type (SIMPLE-ARRAY CHARACTER), not
130 SIMPLE-BASE-STRING. For an alien type that behaves like the old
131 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
132 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
133 * incompatible change: SB-EXT package no longer contains the
134 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
135 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
137 * incompatible change: the single-stepper is no longer available
138 on Alpha, Mips and Sparc platforms.
139 * minor incompatible change: the direct superclasses of
140 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
141 not (STANDARD-OBJECT FUNCTION). This makes the
142 class-precedence-lists of GENERIC-FUNCTION and
143 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
145 * minor incompatible change: the default stream external format on
146 non-unicode SBCL is no longer detected from the locale, but is always
148 * new feature: Add a version of evaluator that uses an interpreter instead
149 of the compiler. EVAL still uses the compiler by default, to switch it
150 to use the interpreter, set the value of the variable
151 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
152 * minor incompatible change: the single-stepper REPL has been merged
153 with the normal debugger (see the "Stepping" heading of the debugger help
154 for more details). The debugger command STEP will no longer switch
155 to the single-stepper REPL.
156 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
157 for a type now works.
158 * bug fix: loading of default sysinit file works. (thanks to Leonid
160 * bug fix: better detection of circularities in the file-compiler.
161 (reported by Marco Monteiro)
162 * bug fix: the CL pathname functions now work with files that have
163 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
164 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
165 systems with Unix98 pty semantics.
166 * bug fix: ASDF-INSTALL will now work with bsd tar.
167 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
169 * bug fix: timers expiring in dead threads no longer cause a
170 type-error (reported by Paul "Nonny Mouse").
171 * bug fix: thanks to more lightweight single-stepper instrumentation,
172 code compiled with (DEBUG 3) will compile and execute significantly faster,
173 and will have more accurate type-inferencing than before
174 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
175 whose bindings are modified
176 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
177 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
178 ** replace dummy implementations of CL:MACHINE-INSTANCE and
179 CL:SOFTWARE-VERSION with proper version
181 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
182 * feature: implemented the READER-METHOD-CLASS and
183 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
184 as specified by AMOP.
185 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
187 * optimization: faster LOGCOUNT implementation on x86 and x86-64
188 (thanks to Lutz Euler)
189 * optimization: hashing of general arrays and vectors has been
190 improved. (reported by Any Fingerhut)
191 * enhancement: SB-INTROSPECT is now able to find definitions of
192 profiled functions. (thanks to Troels Henriksen)
193 * enhancement: compiler-macro expansion applies now to FUNCALL forms
194 as well, allowing compiler-macros for SETF-functions to expand.
195 * enhancement: step-instrumentation no longer wraps known
196 single-value functions with multiple-value context, allowing
197 better type inference.
198 * fixed bug #337: use of MAKE-METHOD in method combination now works
199 even in the presence of user-defined method classes. (reported by
200 Bruno Haible and Pascal Costanza)
201 * fixed bug #339(c): if there are applicable methods not part of any
202 long-form method-combination group, call INVALID-METHOD-ERROR.
203 (reported by Bruno Haible)
204 * fixed bug #361: the :FUNCTION initarg in the protocol for
205 initialization of methods can now be used to override
206 internally-produced optimized functions. (reported by Bruno
208 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
209 system-provided lambda expression no longer cause warnings about
210 unbound #:|pv-table| symbols.
211 * bug fix: improved the handling of type declarations and the
212 detection of violations for keyword arguments with non-constant
214 * bug fix: potentially erronous calls to PATHNAME and
215 MERGE-PATHNAMES were being flushed in some cases.
216 (reported by Richard Kreuter)
217 * bug fix: compiled calls to TYPEP were mishandling obsolete
218 instances. (reported by James Bielman and Attila Lendvai)
219 * bug fix: format strings with ~> without matching ~< no longer
220 trigger an AVER, but signal an understandable error instead.
221 (reported by Antonio Martinez)
222 * bug fix: specifying an output-file without a pathname-type for
223 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
224 of input-file instead of "fasl". (reported by Robert Dodier)
225 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
226 (reported by James Y Knight).
227 * bug fix: compiler-macros-function did not consider the environment
228 argument for shadowing by local functions.
229 * bug fix: compiler-macros expansion was inhibited by local INLINE
231 * bug fix: inline expansions of known functions were subject to
232 step-instrumentation in high debug policies, leading to problems
234 * bug fix: compiler failed to differentiate between different CONS
236 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
237 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
238 * bug fix: #368: incorrect use of expressed vs. upgraded array
240 * bug fix: #306a: more precise unions of array types.
241 * thread-safety improvements:
242 ** CONDITION-WAIT could return early on Linux, if the thread was
243 interrupted and subsequently continued with SIGCONT.
244 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
246 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
247 * added support for the ucs-2 external format. (contributed by Ivan
249 * minor incompatible change: pretty printing of objects of type
250 (cons symbol) is, in the default pprint-dispatch-table, now
251 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
253 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
254 called later in a class's lifetime, possibly as late as when the
255 first instance of the class is created. Previously,
256 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
257 class became finalizeable.
258 * fixed bug: FILE-POSITION sometimes returned inconsistent results
259 for multibyte external-format streams. (thanks to "vbzoli")
260 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
261 with :ALLOCATION :CLASS inherited from superclasses of the
263 * fixed bug: anonymous classes can now be created using the :NAME
264 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
265 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
266 * fixed bug: core-files saved with :EXECUTABLE T can again be
267 executed when SBCL_HOME isn't set. (reported by James Knight)
268 * fixed bug: toplevel LOCALLY forms with declarations could
269 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
270 * fixed bug: printing from several different threads using different
271 values of *print-case* could cause invalid output, due to
272 some internal special variables of the printer not being bound
273 thread-locally (reported by Max Mikhanosha)
274 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
275 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
276 (reported by Pascal Costanza)
277 * fixed bug: FINALIZE-INHERITANCE is called from
278 REINITIALIZE-INSTANCE on classes when the class has previously
279 been finalized, as required by AMOP.
280 * minor code generation optimizations:
281 ** better register allocation in CLOS dispatching functions
282 ** overflow detection when coercing signed bytes to fixnums on x86-64
283 ** is now implemented with one IMUL instruction instead of three shifts
284 ** more efficient bit-vector access on x86 and x86-64
285 ** more efficient access to raw structure slots on x86 and x86-64
286 * fixed some bugs revealed by Paul Dietz' test suite:
287 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
290 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
291 * feature: thread support on Solaris/x86, and experimental thread support
293 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
295 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
296 default method for SLOT-UNBOUND.
297 * minor incompatible change: prevent the user from specializing the
298 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
299 somewhat counter to the intent of the protocol, I (CSR) think, and
300 additionally it just doesn't work in SBCL as currently
301 implemented, thanks to optimizations (that are always valid for
302 the other three SLOT-VALUEish functions, but not for the setter).
303 * minor incompatible change: the :SB-LDB feature is now enabled by
304 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
305 the low-level debugger.
306 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
307 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
308 down to an SB-SYS:FD-STREAM, allowing direct communication in
309 more cases. Temporary files and pipes are still used as a fallback
311 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
312 on the non-reentrant C functions localtime(3) and gmtime(3).
313 * bug fix: global optimization policy was not visible in LOCALLY and
315 * bug fix: class objects can be used as specializers in methods.
316 (reported by Pascal Costanza)
317 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
318 with a type error. (reported by blitz_ on #lisp)
319 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
320 failed with a type error. (reported by Pascal Bourguignon)
321 * bug fix: merging pathnames against defaults with :DIRECTORY
322 starting with '(:RELATIVE :BACK) should preserve the :BACK.
323 (reported by James Y Knight)
324 * bug fix: saving large (>2GB) cores on x86-64 now works
325 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
326 constant index and a (simple-array (signed-byte 32)) array
327 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
328 error (patch by Robert J. Macomber)
329 * feature: TIME now displays the amount of run-time spent in GC
330 * bug fix: The debugger now does a better job of respecting
331 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
332 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
333 workaround for bug 403.)
334 * bug fix: floating point exception handling now works on FreeBSD
335 (thanks to NIIMI Satoshi)
336 * fixed some bugs revealed by Paul Dietz' test suite:
337 ** MISC.641: LET-conversion were not supposed to work in late
339 * improvements to the Win32/x86 port:
340 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
341 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
343 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
344 * new feature: source path information is generated for macro-expansion
345 errors for use in IDE's like Slime (thanks to Helmut Eller)
346 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
347 * bug fix: compiler does not loop forever on an invalid type in
349 * improvement: compilation of most CLOS applications is significantly
351 * optimization: added a limited bytecode compiler for simple toplevel
352 forms, speeding up compilation and FASL loading
353 * bug fix: the statistical profiler now properly distinguishes anonymous
356 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
357 * minor incompatible change: in sbcl-0.9.11 (but not earlier
358 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
359 into subclasses of the finalized class. Now user calls to
360 FINALIZE-INHERITANCE finalize just the one class, and calls by the
361 system return before any subclasses are finalized.
362 * minor incompatible change: The reader no longer ignores errors
363 regarding non-existent packages in #+ and #- feature tests.
364 * new feature: command line options --no-sysinit, --no-userinit to
365 inhibit loading the corresponding init files
366 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
367 for sending data through UDP sockets (thanks to François-René Rideau)
368 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
369 error is signalled instead (thanks to François-René Rideau)
370 * minor incompatible change: Error signalling behaviour of lexical
371 operations violating package locks has changed slightly. Refer to
372 documentation on package locks for details.
373 * enhancement: EVAL can process IF-expressions without resorting to the
375 * bug fix: Some operations involving SIN, COS, or TAN weren't being
376 constant-folded properly due to missing out-of-line functions.
377 (reported by Mika Pihlajamäki)
378 * bug fix: LISTEN sometimes returned T even in cases where no data was
379 immediately available from the stream
380 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
381 were reversed. (reported by Levente Mészáros)
382 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
383 when given a vector displaced to an adjustable array. (reported
385 * bug fix: derivation of float boundaries from numbers outside the
386 appropriate float range (reported by John Wiseman)
387 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
389 * bug fix: precomputing cache entries for generic functions with
390 some subclasses of specializers as yet invalid does not attempt to
391 fill a cache line with a negative offset. (reported by Levente
393 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
394 allow condition class objects as arguments to DOCUMENTATION and
395 (SETF DOCUMENTATION); only find and set documentation for
396 structure names for the STRUCTURE doc-type. (suggested by Gary
398 * improvements to the Win32/x86 port:
399 ** (user-homedir-pathname) and default initialization file
400 locations now know about the user's "Documents and Settings"
401 directory (thanks to Yaroslav Kavenchuk)
402 ** run-program is implemented (thanks to Mike Thomas)
403 ** sockets support (thanks to Timothy Ritchey)
404 ** better backtrace support (thanks to Alastair Bridgewater)
405 ** sb-grovel supported
406 ** asdf-install and sb-posix work somewhat
407 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
408 * improvements to the Solaris/x86 port:
409 ** works on Solaris 11/Solaris Express
410 ** floating-point exception handling support
411 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
412 * fixed some bugs revealed by Paul Dietz' test suite:
413 ** REMOVE-METHOD returns its generic function argument even when
414 the method is not one of the generic functions' methods.
415 ** objects declared as MEMBER types can be admissible arguments to
417 ** printing characters should simply be printed by the FORMAT ~:C
419 ** compiler failure when compiling functions with hairy constant
420 defaults for optional parameters.
421 ** compiler produces wrong code when MAYBE-INLINE-expanding a
422 function, which is already optimized.
424 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
425 * new platform: experimental support for SBCL x86/Darwin, including
426 MacOS X 10.4.5 on Intel.
427 * new feature: Unicode character names are now known to the system
428 (through CHAR-NAME and NAME-CHAR).
429 * new feature: the filesystem location of SBCL's core file is
430 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
431 * minor incompatible change: the contrib modules SB-POSIX and
432 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
433 this change is to make it easier to distribute
434 location-independent binaries.
435 * bug fix: as implied by AMOP, standardized classes no longer have
436 slots named by external symbols of public packages. (reported by
438 * optimization: calling functions via constant symbols -- as in
439 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
440 via the function object as in (FUNCALL #'FOO).
441 * optimization: on x86 and x86-64, the calling convention for the
442 case of dealing with an unknown number of values has been altered
443 to be friendlier to the prediction heuristics implemented,
444 particularly with reference to CALL and RET pairing. (thanks to
445 Alastair Bridgewater)
446 * optimization: on x86, the code for access to thread-local symbol
447 values and binding/unbinding of thread-local symbols is smaller.
448 (thanks to Alastair Bridgewater)
449 * enhancement: CONSTANTP is now able to determine constantness of
450 more complex forms, including calls to constant-foldable standardized
451 functions and some special forms beyond QUOTE.
452 * fixed bug: occasional GC crashes on Solaris/x86
453 * optimization: x86-64 supports stack allocation of results of simple
454 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
455 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
456 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
457 (thanks to James Knight)
458 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
459 by the standard (thanks to Stephen Compall)
461 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
462 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
463 be used for bundling the runtime and the core file into one
464 executable binary. This feature is not currently supported on all SBCL
465 platforms. (thanks to James Bielman and NIIMI Satoshi)
466 * new feature: a generational or ephemeral garbage collector is now
467 the default on the PowerPC platform (both Linux and Darwin). The
468 old Cheney (stop and copy) collector is a build-time option.
469 (thanks to Cyrus Harmon, after Raymond Toy)
470 * minor incompatible change: the method by which SBCL finds its
471 contributed modules has changed; it no longer relies on symbolic
472 links from an $SBCL_HOME/systems directory, but searches directly
473 in the subdirectories of $SBCL_HOME.
474 * enhancement: the dynamic heap size on the Linux/PPC platform is
475 markedly larger, even using the older Cheney garbage collector.
476 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
477 certain complicated string types. (reported by Gary King)
478 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
479 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
480 * fixed bug: DOCUMENTATION on structures no longer signals an
481 error if no structure type of the right name exists. (reported by
483 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
484 applicable non-standard primary or :AROUND method on
485 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
486 no longer cause unbound variable errors. (reported by Kevin Reid)
487 * fixed bug: metacircle resolution in cases where methods have slots
488 added before the slots from STANDARD-METHOD. (reported by Jean
490 * fixed bug: the Power PC assembly code for calling into lisp
491 sometimes computed its return address wrongly (depending on the
492 15th bit of the address, fixed at link time). This used to
493 manifest itself in a segmentation violation while building PCL.
494 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
495 many others over the years)
496 * fixed bug: ROOM no longer reports silly numbers for stack usage if
497 the stack pointer (treated as a signed integer) is negative.
498 (thanks to Peter van Eynde)
500 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
501 * new platform: experimental support for the Windows operating
502 system has been added. (thanks to Alastair Bridgewater)
503 * fixed several bugs in and robustified the PPC FFI (including
504 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
505 * bug fix: allow non-simple string symbol names (reported by Paul
507 * bug fix: interrupt handling on NetBSD (thanks to Richard M
509 * bug fix: saving a core corrupted callbacks on x86/x86-64
510 * bug fix: closed a loophole in metacircularity detection and
511 grounding in the PCL implementation of CLOS.
512 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
513 (reported by tomppa on #lisp)
514 * bug fix: clear the direction flag on Lisp -> C transitions, as
515 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
516 (reported by Andrej Grozin and Hendrik Maryns)
517 * optimization: major improvements to GC efficiency on GENCGC platforms
518 * optimization: faster implementation of EQUAL
519 * optimization: emit more efficient opcodes for some common
520 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
521 * optimization: several other minor code-generation improvements on x86-64
522 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
524 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
525 * minor incompatible change: (SETF CLASS-NAME) and (SETF
526 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
527 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
528 by ANSI as a generic function, and for consistency (SETF
529 GENERIC-FUNCTION-NAME) is treated likewise.
530 * fixed bug #233.b: make constraint propagation notice when a variable
531 value is changed after it is referenced but before it is used
532 * fixed bug #296: no more arbitrary behaviour from filenames with
533 odd characters as --load arguments. (reported by Adam Warner)
534 * fixed bug #390: :CHARACTER-SET pathname components now work as
535 expected. (reported by Tim Daly Jr)
536 * fixed bug #391: complicated :TYPE intersections in slot
537 definitions no longer cause an error in PCL internals.
538 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
539 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
540 returning the number of octets which would be written to the
541 file-stream. (thanks to Robert J. Macomber)
542 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
543 arguments to the internal stream functions. (thanks to David
545 * bug fix: stack exhaustion is now handled on the x86/SunOS
546 platform. (thanks to Daisuke Homma)
547 * bug fix: more accurate ROOM results on GENCGC platforms
548 * optimization: improved type inference for arithmetic-for
549 index variables in LOOP
550 * optimization: faster floating-point SQRT on x86-64
551 * fixed some bugs revealed by Paul Dietz' test suite:
552 ** DOCUMENTATION returns NIL instead of "" for method combinations
553 that don't have a docstring
555 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
556 * minor incompatible change: (SETF CLASS-NAME) and (SETF
557 GENERIC-FUNCTION-NAME) are no longer generic functions, and
558 therefore are not customizeable by user code (as seems to be at
559 least permitted and maybe required by AMOP). As a consolation,
560 however, the SBCL implementation of these functions now calls
561 REINITIALIZE-INSTANCE as specified by AMOP.
562 * bug fix: slot-definition documentation is propagated as per ANSI
563 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
564 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
565 argument, as do eslotd initialization methods. (from Pascal
566 Costanza's "Closer" project)
567 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
568 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
570 * bug fix: it is now possible to have more than one subclass of
571 STANDARD-GENERIC-FUNCTION without causing stack overflow.
572 (reported by Bruno Haible, Pascal Costanza and others)
573 * bug fix: the dependent update protocol now works for generic
574 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
576 * bug fix: condition-class instances corresponding to
577 DEFINE-CONDITION forms are now created eagerly. (reported by
578 Kalle Olavi Niemitalo on comp.lang.lisp)
579 * bug fix: floating point printing is more accurate in some
580 circumstances. (thanks to Simon Alexander)
581 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
582 merged with *DEFAULT-PATHNAME-DEFAULTS*.
583 * bug fix: callbacks on OS X now preserve stack-alignment.
584 (thanks to Cyrus Harmon)
585 * enhancement: the x86-64 disassembler is much better at
586 disassembling SSE instructions. (thanks to Lutz Euler)
587 * enhancement: CHECK-TYPE on a variable now allows type-inference
588 (thanks to Vincent Arkesteijn)
589 * optimization: improved performance of EUC-JP external format.
590 (thanks to NIIMI Satoshi)
591 * optimization: performance improvements to IO on file streams of
592 :ELEMENT-TYPE CHARACTER
593 * optimization: much faster memory allocation on x86-64
594 * optimization: faster garbage collections (latency and throughput)
596 * optimization: faster bignum subtraction and division on x86 and x86-64
597 (thanks to Lutz Euler)
598 * optimization: more accurate type-inference for loops that use a
599 floating point index variable or a negative step.
601 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
602 * bug fix: add a workaround to SBCL looping infinitely at startup on
603 Linux kernels with apparently buggy implementations of personality().
604 (thanks to Svein Ove Aas)
605 * bug fix: Unicode symbols are correctly printed in LDB backtraces
606 (thanks to David Lichteblau)
607 * bug fix: local bindings shadow global symbol macros for
608 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
609 * optimization: non-open coded uses of numeric comparison operators
610 (e.g. >) no longer cons when called with more than one parameter
611 on platforms supporting dynamic-extent allocation.
612 * enhancement: saving cores with foreign code loaded is now supported
613 on MIPS/Linux in addition to the previously supported platforms.
614 * bug fix: FIND-RESTART now tests for activity, not applicability when given
615 a restart object as identifier. (reported by Helmut Eller for CMUCL)
616 * bug fix: division by zero in sb-sprof when no samples were collected
617 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
618 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
619 (thanks to Alexander Kjeldaas)
620 * new feature: ignoring the return values of destructive functions
621 like NREVERSE signals a compile-time style-warning.
622 (thanks to Kevin Reid)
623 * new feature: the alignment of alien structure fields can be
624 explicitly specified. (thanks to Cyrus Harmon)
625 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
626 and dump core on SIGQUIT
628 ** incompatible change: threads do not inherit values of specials
629 from their parents (see manual)
630 ** bug fix: threads stacks belonging to dead threads are freed by the
631 next exiting thread, no need to gc to collect thread stacks anymore
632 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
633 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
634 ** bug fix: dynamic variable and thread start related gc lossage
635 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
637 * fixed some bugs revealed by Paul Dietz' test suite:
638 ** SUBTYPEP is slightly more accurate on heinously complicated
639 CONS types where some of the members have uncertain (in the
640 NIL, NIL sense) type relationships to each other.
642 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
643 no longer purified unless :PURIFY T is explicitly specified.
644 ** Non-purified cores are significantly smaller than before
646 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
647 * new feature: timers based on Zach Beane's excellent timer package
648 * added support for the following external formats: koi8-u,
649 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
650 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
651 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
652 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
653 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
654 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
655 * incompatible change: a threaded SBCL will no longer revert to
656 non-threaded mode on non-NPTL systems, but refuse to start entirely.
657 * bug fix: interrupts are disabled until startup is complete; no
658 more sigsegvs when receiving a signal to soon
659 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
661 * bug fix: add a workaround for the memory randomization features in
662 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
663 workaround will only be in effect on systems with the proc filesystem
665 * bug fix: printing objects of type HASH-TABLE signals a
666 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
668 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
669 have been running for over 50 days. (reported by Gilbert Baumann)
670 * bug fix: the logic for getting names of functions gets less
671 confused when confronded with alternate-metaclass
672 funcallable-instances. (reported by Cyrus Harmon)
673 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
674 consistent, even on internal alternate-metaclass objects.
675 * bug fix: flush closure information collected by physical
676 environment analysis prepass before the main pass. (bug reported
678 * bug fix: compiler pack phase does not modify a hashtable, which is
679 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
680 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
681 may be expanded. (reported by Alexander <malishev> on c.l.l)
682 * bug fix: no more style-warnings when compiling code using LAST
683 under high-SPEED low-SPACE settings. (reported by David Wallin
685 * fixed some incompatibilities between SBCL's MOP and the MOP
687 ** the METAOBJECT class is now implemented;
688 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
689 STANDARD-OBJECT, as required;
690 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
691 now compatible (as required by VALIDATE-SUPERCLASS); there
692 remains a consistency requirement over the presence or absence
693 of the FUNCTION class in the superclass list of finalized
694 classes; see the manual for more details;
695 ** the :AROUND method for COMPUTE-SLOTS on
696 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
697 requested slot ordering.
699 ** bug fix: parent thread now can be gc'ed even with a live
701 ** bug fix: binding a special with PROGV to no value is not
702 the same as that symbol not having been bound (thanks to
704 ** bug fix: binding specials is thread safe (thanks to
706 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
707 and SB-SPROF do not die with 'no handler for signal XX in
708 interrupt_handle_now(..)' anymore
709 ** bug fix: WITH-TIMEOUT works with multiple running threads
710 * fixed some bugs revealed by Paul Dietz' test suite:
711 ** ENSURE-GENERIC-FUNCTION should take a method class object for
712 the :method-class keyword argument.
714 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
715 * new port: the Solaris operating system on x86 processors is now
716 mostly supported, though some rough edges in the environment
717 remain. (thanks to Daisuke Homma)
718 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
719 space for its heap. (thanks to Thiemo Seufer)
720 * optimization: SBCL on MIPS platforms now supports dynamic-extent
721 closures. (thanks to Thiemo Seufer)
722 * minor incompatible change: eof selects abort in the debugger.
723 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
724 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
726 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
727 deprecated, and will go away in a future revision of SBCL.
728 * minor incompatible change: GC-ON and GC-OFF are no longer
729 implemented with a counter, it does not matter how many times gc
730 is switched on or off
731 * bug fix: discriminating functions for generic function classes
732 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
733 make invalid assumptions about method precedence order. (reported
735 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
736 platforms. (thanks to Thiemo Seufer)
737 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
738 * bug fix: correct pathname printing with printer escaping is on.
739 (thanks to Kevin Reid)
740 * bug fix: complex VOP definitions in "user-space" no longer trigger
741 package locks. (reported by Zach Beane)
742 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
743 for funcallable-instance functions, and is no different from
745 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
746 characters in the current readtable. (reported by Nicholas Neuss)
747 * bug fix: SUBTYPEP on various CONS types returns more a more
748 accurate acknowledgment of its certainty.
749 * optimizations: REMOVE-DUPLICATES now runs in linear time on
750 lists in some cases. This partially fixes bug 384.
751 * flush all standard streams before prompting in the REPL and the
753 * bug fix: signal handling and triggering gc do not conflict
754 directly anymore, in particular a high frequency sb-sprof does
755 not prevent gc from running
756 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
757 approximation for timezone and DST information between the
758 universal time epoch and the smallest negative 32-bit time_t.
759 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
760 year 1899 so long as the encoded time is non-negative.
761 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
762 the correct answer for non-constant rotate amounts when performing
763 an inline 32-bit rotation.
765 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
766 there is only one thread in the session
767 ** bug fix: memory leak for streams created in one thread and
768 written to in another
769 ** bug fix: lockup when compiled with gcc4
770 ** bug fix: race that allows the gc to be triggered when gc is
772 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
773 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
775 ** bug fix: finished threads are gc'ed properly
776 * fixed some bugs revealed by Paul Dietz' test suite:
777 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
778 has arguments with a different set of applicable methods from
779 the orignal arguments.
780 ** The type error thrown by MAP now has a correct expected-type
782 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
783 name a compiled function.
784 ** Three MISC tests where a large bignum was improperly coerced to
785 a float that couldn't represent that bignum during type
786 derivation were fixed.
787 ** SUBTYPEP can now handle types involving the negation of a
788 list-form FUNCTION type.
789 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
790 as (COMPLEX (AND RATIO (NOT FIXNUM))).
791 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
793 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
794 * New feature: Experimental support for bivalent streams: streams
795 opened with :element-type :default now allow character and binary
796 (unsigned-byte 8) I/O
797 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
798 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
800 * bug fix: no more highly sporadic "couldn't check whether ~S is
801 readable" when reading a stream and an interrupt hits in the middle
802 of a select system call
803 * compiler better recognizes complex arrays (reported by Eduardo
805 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
806 for variables on SBCL built with linkage-tables. (reported by Luis
808 * various error reporting improvements.
809 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
810 (Thanks to Thiemo Seufer)
811 * bug fix: nested reader invokations work correctly
812 * bug fix: it is possible to have simultaneous references to foreign
813 code and foreign data with the same name.
815 ** added x86-64 support
816 ** incompatible change: the threading api now works with thread
817 objects instead of thread ids
818 ** bug fix: threads are protected from signals and interruption when
819 starting up or going down
820 ** bug fix: a race where an exiting thread could lose its stack to gc
821 ** bug fix: don't halt on infinite error in threads if possible
822 ** fixed numerous gc deadlocks introduced in the pthread merge
823 ** bug fix: fixed thread safety issues in read and print
824 ** bug fix: debugger doesn't hang on session lock if interrupted at
825 an inappropriate moment
826 ** bug fix: run-program is now thread safe(r)
827 ** bug fix: inner with-recursive-lock no longer releases the mutex
828 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
829 * fixed some bugs revealed by Paul Dietz' test suite:
830 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
831 EXPECTED-TYPE slots filled.
832 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
834 ** MAKE-CONDITION accepts classes as type-designators.
835 ** COMPILE may never return NIL.
836 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
837 range before calling Unix time functions
839 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
840 * numerous signal handling fixes to increase stability
841 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
842 * minor incompatible change: we now correctly canonize default
843 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
844 as per the MOP, rather than the historical (INITARG INITFUNCTION
845 INITFORM). (reported by Bruno Haible)
846 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
847 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
848 for more information.
849 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
850 pathname is a directory pathname.
851 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
852 * dynamic space size on PPC has been increased to 768Mb. (thanks to
854 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
855 :METACLASS argument in addition to a class name. (reported by
856 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
857 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
858 (reported by Bruno Haible; patch from Gerd Moellmann)
859 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
861 * bug fix: more cleanups to the floating point exception handling on
862 x86-64 (thanks to James Knight)
863 * bug fix: the compiler does not try to constant fold calls of
864 COERCE and BIT-* functions when they return freshly constructed
865 objects. (reported by Nikodemus Siivola and Paul Dietz)
866 * optimization: improved the MIPS versions of generic (in the
867 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
868 * optimization: direct conversion of (unsigned-byte 32) to floats on
869 the PowerPC platform.
870 * optimization: structure instances with raw slots now use less
871 memory, and probably show better memory locality. (thanks to
873 * optimization: DYNAMIC-EXTENT declarations for lists and closures
874 are treated as requests for stack allocation on the x86-64,
875 Alpha-32, PPC and SPARC platforms.
876 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
877 the wrong file descriptor; implementation of SOCKET-OPEN-P.
878 (thanks to Tony Martinez)
880 ** gcing a dead thread can no longer lead to lockups
881 ** threads block signals until they are set up properly
882 ** errno is no longer shared by threads
883 ** interrupt-thread restores the eflags register on x86
884 ** fixed some lockups due to gc/thread interaction
885 * fixed some bugs revealed by Paul Dietz' test suite:
886 ** invalid dotted lists no longer raise a read error when
888 ** signal an error if a symbol that names a declaration is used
889 as the name of a type, or vice versa
890 ** allow using the (declare (typespec var*)) abbreviation for
891 (declare (type typespec var*)) with all type specifiers
892 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
893 ** always return NIL from PPRINT-POP when OBJECT is NIL
894 ** don't signal errors when pretty-printing malformed LABELS,
895 FLET or MACROLET forms
896 ** declarations in a DOLIST body are also in scope for the
898 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
900 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
903 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
904 * fixed cross-compiler leakages that prevented building a 32-bit
905 target with a 64-bit host compiler.
906 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
907 opened with :IF-EXISTS :OVERWRITE.
908 * fixed bug 281: error for an invalid qualifier in a short-form method
909 combination method is not signalled until the faulty method is called.
910 * bug fix: iteration variable type inferrer failed to deal with open
911 intervals. (reported by Alan Shields)
912 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
913 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
914 or more runtime options were provided to the sbcl binary.
915 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
916 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
917 * optimization: there's now a fast-path for fixnum arguments in the
918 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
919 * optimization: the code generated on x86-64 is more compact thanks
920 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
921 * bug fix: floating-point exception handling now partly works on
922 x86-64. (Thanks to James Knight)
923 * improvement to the MIPS backend from Thiemo Seufer: C-style
924 64-bit long long arguments and return values to alien functions
926 * fixed some bugs revealed by Paul Dietz' test suite:
927 ** the type-error signalled from WARN has a filled-in DATUM slot.
928 ** the type-error required when a stream is not associated with
929 a file has the stream as its datum.
930 ** type-errors on single-floats on x86-64 no longer have
931 :INVALID-OBJECT as the datum
932 ** the type-errors signalled for invalid function names now have
933 a correct expected type
934 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
935 for structures defined with a :TYPE. Documentation strings for
936 typed structures are no longer immediately discarded
937 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
938 broadcast streams and synonym streams. FILE-LENGTH now also works
939 on broadcast streams.
941 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
942 * incompatible change: the --noprogrammer option, deprecated since
943 version 0.7.5, has been removed. Please use the equivalent
944 --disable-debugger option instead.
945 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
947 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
948 inoperational for a while now) has been completely removed.
949 * null lexical environments are now printed as #<NULL-LEXENV>,
950 significantly reducing the amount of clutter in typical backtraces.
951 * documentation on weak pointers, finalization, and after GC hooks
952 has been added to the manual.
953 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
954 as other specialized array types, is much faster. SUBSEQ and
955 COPY-SEQ on such arrays have also been sped up.
956 * optimization: EQL is now more efficient when at least other argument
957 is known to be of type (OR FIXNUM (NOT NUMBER)).
958 * fixed bug: the runtime is now less vulnerable to changes in the
959 size of the SBCL object on OS X, and virtual memory is reserved for
960 all spaces that need to be at a fixed address.
961 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
962 * fixed bug: finalizers and after GC hooks that cause consing are now
964 * fixed bug: compiler error messages and summaries are now printed to
965 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
966 * fixed inference of the upper bound of an iteration variable.
967 (reported by Rajat Datta).
968 * fixed bug 211e: calling local functions with duplicated constant
969 keyword argument no longer causes a bogus style warning about an
971 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
972 ftype declaration to be disregarded. (reported by Dave Roberts)
973 * fixed bug 373: caused by erronous compilation of references to alien
974 variables in the runtime on ppc/darwin.
975 * fixed bug 376: CONJUGATE type deriver.
976 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
977 (reported by Baughn on #lisp)
978 * a cross-compiler bug on non-x86 platforms has been identified and
979 fixed. (thanks to Bruno Haible)
980 * improvements to the MIPS runtime code for increased stability.
981 (thanks to Thiemo Seufer)
982 * increased the maximimum compact environment size to allow
983 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
984 * partial workaround for bug 135: don't name the function we're
985 calling for hairy cases of EVAL, so as not to accumulate one environment
986 entry per call to EVAL. (reported by Kevin Reid)
987 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
988 * optimization: type testing for non-vector arrays should be faster.
989 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
991 * bug fix: setting 31st element of a bit vector to zero did not work
993 * bug fix: redefining a class definition which failed due to a
994 previous accessor / function clash now works (but see BUGS entry
995 #380 for more problems in this area). (thanks to Zach Beane)
996 * the long form of DEFINE-METHOD-COMBINATION disables method group
997 checking when given a single method group with pattern *. (thanks
999 * on x86 compiler supports stack allocation of results of simple
1000 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1001 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1002 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1003 * fixed some bugs related to Unicode integration:
1004 ** the restarts for recovering from input and output encoding
1005 errors only appear when there is in fact such an error to
1007 * fixed some bugs revealed by Paul Dietz' test suite:
1008 ** MISC.549 and similar: late transformation of unsafe type
1009 assertions into derived types caused unexpected code
1011 ** SCALE-FLOAT type deriver is less wrong.
1012 ** type derivers for EXP, LOG and similar functions compute result
1013 types for complex arguments better.
1014 ** (MISC.563) CONJUGATE type deriver works for very restricted
1016 ** out-of-line type testers for character strings are available.
1017 ** EQUAL compiler transform understands specialness of objects
1019 ** accessing double-floats stored on the stack now works on x86-64.
1020 ** debugger internals could sometimes create invalid lispobjs,
1021 resulting in GC crashes.
1022 ** MISC.548: type check weakening can convert required type into
1024 ** initialization forms for bindings are not in scope of free special
1027 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1028 * incompatible change: thread support for non-NPTL systems has
1029 been removed - locking is buggy and unreliable. A threaded
1030 SBCL build will now warn at startup and refuse to create
1031 new threads, unless futex support is detected in the kernel
1032 * incompatible change: the top level REPL now has only an ABORT
1033 restart associated with it, not TOPLEVEL and ABORT as it used to.
1034 TOP and TOPLEVEL are now available as debugger commands for
1035 returning to the top level.
1036 * incompatible change: forms evaluated in the REPL now use the
1037 global optimization policy.
1038 * incompatible change: user- and system-initialization files are
1039 no longer processed with LOAD, but by READ and EVAL; hence the
1040 global optimization policy, startup package, readtable, etc,
1042 * The .fasl file format number has been incremented because of
1043 various incompatible changes.
1044 * internal entry point details and argument counts no longer appear
1045 in backtraces unless explicitly requested by setting
1046 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1047 * built-in and standard functions no longer have names like "top
1048 level local call to FOO".
1049 * fixed bug 32: functions defined in non-null lexical environments
1050 now have more legible printed representation
1051 * fixed bug 33: functions defined in non-null lexical environemnts
1052 are now more amenable to inspection by INSPECT.
1053 * workaround for bug 354: XEPs no longer appear in backtraces unless
1054 explicitly requested.
1055 * fixed bug: receiving the signal which results from attempting to
1056 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1057 SIGBUS on other *BSDs) on architectures where the C stack is also
1058 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1059 to debugger confusion or wild instability. (reported by Cheuksan
1061 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1062 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1063 (reported by Lutz Euler)
1064 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1065 compiled code. (reported by Paul Dietz)
1066 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1067 specialized parameter to a method to something that is not TYPEP
1068 the specializer is now possible.
1069 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1070 face of package deletion.
1071 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1072 pathname. (reported by David Tolpin on c.l.l)
1073 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1074 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1075 STRING-STREAM and FILE-STREAM, respectively.
1076 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1077 than previously; in addition, its test suite is now run on build.
1078 (thanks to Luke Gorrie)
1079 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1080 * optimization: added a immediate representation for single-floats
1082 * fixed some bugs revealed by Paul Dietz' test suite:
1083 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1084 ** attempting to create a package with a colliding nickname causes
1085 correctable errors to be signalled.
1086 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1087 ** several x86-64 backend bugs related to sign-extension of immediate
1090 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1091 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1092 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1094 * fixed loading of multiply forward-referenced layouts.
1095 (thanks to Cheuksan Wang)
1096 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1097 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1098 (see COMPILE-FILE documentation for details of :PRINT option)
1099 * fixed bugs 19 and 317: fixed-format floating point printing is
1100 more accurate. This also fixes a bug reported by Adam Warner
1101 related to the ~@F format directive.
1102 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1104 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1105 dispatch table if the from-char is a dispatch macro character.
1106 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1107 by the wordsize no longer ignore the last word. (reported by Lutz
1109 * fixed bug in type checking of dynamic-extent variables. (reported
1111 * optimization: sequence traversal functions use their freedom to
1112 coerce function designators to functions.
1113 * optimization: code with many calls to CLOS methods specialized on
1114 CLOS classes has had redundant type checks removed.
1115 * optimization: added declarations to speed up operations that access
1116 the internal character database (for example STRING-UPCASE)
1117 * optimization: comparison operations between floats and sufficiently small
1118 fixnums no longer create extra rationals
1119 * fixed some bugs related to Unicode integration:
1120 ** portions of multibyte characters at the end of buffers for
1121 character-based file input are correctly transferred to the
1122 start of the buffer at the next read.
1123 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1124 passing it through to OPEN.
1125 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1126 argument given to internal calls to OPEN.
1127 ** the built-in comment readers (introduced by character sequences
1128 ";" and "#|") are more forgiving to encoding errors; they will
1129 STYLE-WARN and then attempt to resync the stream at a character
1130 boundary. (thanks to Teemu Kalvas)
1131 * fixed some bugs in the x86-64 port:
1132 ** Negative short int return values from c-calls are sign-extended
1134 ** The stack is aligned to 16-bytes for c-calls, as required by
1135 the ABI. (thanks to Cheuksan Wang)
1136 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1137 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1138 for enums. (thanks to Vincent Arkesteijn)
1139 ** Multiple small optimizations and bugfixes for floating point
1141 * fixed some bugs revealed by Paul Dietz' test suite:
1142 ** Space, Tab, Linefeed, Return and Page have the invalid
1143 secondary constituent character trait.
1144 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1146 ** WITH-INPUT-FROM-STRING should only update the index place on
1148 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1149 works more reliably.
1150 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1151 with one-dimensional arrays with fill pointers.
1152 ** TYPE-OF failed on a complex with an integer realpart and a
1154 ** compiler failure during type inference for the code of form
1155 (IF (EQL X (THE ...)) ...) (MISC.535).
1157 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1158 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1159 platform. The port supports 61-bit fixnums, large memory spaces
1160 and reloading shared object files.
1161 * enhancement: saving cores with foreign code loaded is now
1162 supported on ppc/Darwin in addition to the previously supported
1164 * enhancement: the statistical profiler now walks deeper into the
1165 call stack for more meaningful call-graphs and accrued time
1166 reports (x86/x86-64 only). It also now reports time spent in
1168 * enhancement: it is now possible to trace most individual methods
1169 of a generic function in addition to tracing the generic function
1171 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1173 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1174 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1175 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1176 produces an error. (thanks to Vincent Arkesteijn)
1177 * bug fix: NAMESTRING on pathnames with :WILD components in their
1178 directories works correctly. (thanks to Artem V. Andreev)
1179 * fixed bug 125: compiler preserves identity of closures. (reported
1181 * bug fixed: functions with &REST arguments sometimes failed with
1182 "Undefined function" when compiled with (DEBUG 3). (reported by
1184 * bug fix: overflow during compiling of setting element of a bit
1185 vector with constant index and value. (reported by Timmy Douglas)
1186 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1187 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1188 * on x86 compiler partially supports stack allocation of dynamic-extent
1190 * GO and RETURN-FROM do not check the extent of their exit points
1191 when compiled with SAFETY 0.
1192 * fixed some bugs related to Unicode integration:
1193 ** encoding and decoding errors are now much more robustly
1194 handled; it should now be possible to recover even from invalid
1195 input or output to the terminal. (thanks to Teemu Kalvas)
1196 ** provided a first cut at implementing STRING-TO-OCTETS and
1197 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1198 ** altered the SB-MD5 contributed module slightly, changing the
1199 interface just enough for it to be supportable for builds where
1200 lisp characters are not eight bits.
1201 * fixed some bugs revealed by Paul Dietz' test suite:
1202 ** the FORMATTER-generated functions for ~V[ conditionals require
1203 the correct number of arguments.
1204 ** READ-FROM-STRING returns the mandated second value when applied
1205 to displaced strings.
1206 ** the #\Rubout and #\Backspace characters are treated as invalid
1207 constituent characters by the tokenizer.
1209 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1210 * new feature: reloading changed shared object files with
1211 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1212 * new feature: references to foreign variables and functions
1213 can now be compiled and loaded before the corresponding shared
1214 object file is loaded, as long as the foreign definitions are
1215 available at runtime.
1216 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1217 supported. (thanks to Dan Debertin)
1218 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1219 just keywords. (thanks to Vincent Arkesteijn)
1220 * fixed bug #331: structure-class instances corresponding to
1221 DEFSTRUCT forms are now created eagerly.
1222 * fixed bug #345: backtraces from calls to undefined functions work
1223 on x86 as well. Related bug #61 is now also partially fixed on x86
1224 (backtraces from throws to unknown catch tags.)
1225 * bug fix: lambda-list parsing is now stricter vrt. order and number
1226 of lambda-list keywords.
1227 * bug fix: as specified by AMOP, an error is signalled if a
1228 class-option appears multiple times in a DEFCLASS form. (reported
1230 * bug fix: RANDOM can be compiled when the compiler derives the type
1231 of its numeric argument as a disjoint set of small integers.
1232 (reported by Paul Dietz)
1233 * bug fix: starting a core saved with shared objects loaded when
1234 those objects are not available no longer causes threaded SBCL to
1235 hang. (reported by Sean Ross)
1236 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1238 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1239 argument, which must always be NIL. (reported by Kalle Niemitalo)
1240 * bug fix: printing 1.0d+23 no longer results in an error.
1241 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1242 * bug fix: structure slot setters preserve evaluation order. (thanks
1244 * fixed some bugs related to Unicode integration:
1245 ** RUN-PROGRAM can allow its child to take input from a Lisp
1246 stream. (reported by Stefan Scholl)
1247 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1248 locales. (reported by Ken Causey)
1249 * fixed some bugs revealed by Paul Dietz' test suite:
1250 ** INCF, DECF and REMF evaluate their place form as specified in
1252 ** FORMATTER expands ~{ ~} iteration directives with V or #
1253 parameters correctly.
1254 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1255 consequent uses no arguments correctly.
1256 ** the system has a partial understanding of the (COMPLEX RATIO)
1258 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1261 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1262 * new feature: a build-time option (controlled by the :SB-UNICODE
1263 keyword feature, enabled by default) for building the system with
1264 support for the entire 21-bit character space defined by the
1266 * new feature: the system now has rudimentary external-format
1267 support; the primary user-visible change at this time is that
1268 characters with the high bit set (such as lower-case-e-acute) will
1269 print correctly to a terminal in a UTF-8 environment.
1270 * minor incompatible change: BASE-CHAR no longer names a class;
1271 however, CHARACTER continues to do so, as required by ANSI.
1272 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1273 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1274 has been moved to the SB-EXT package (temporarily re-exported from
1276 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1277 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1279 * fixed bug #351: better error-handlind and reporting for malformed
1281 * fixed bug #350: bignum-printing is now more memory-efficient,
1282 allowing printing of very large bignums, eg. (expt 2 10000000).
1283 (reported by Bruno Haible)
1284 * fixed bug #302: better primitive-type selection for intersection
1286 * fixed bug #308: non-graphic characters now all have names, as
1287 required. (reported by Bruno Haible)
1288 * bug fix: redefining a class with different superclasses now correctly
1289 removes it from the direct-subclasses of its previous superclasses.
1290 (reported by David Morse)
1291 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1292 new value now works. (reported by Bruno Haible)
1293 * bug fix: correct canonicalization of multiple non-standard slot
1294 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1295 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1296 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1297 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1298 now exists, an signals an error.
1299 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1300 during macroexpansion for non-iteration variables as well. (reported
1301 by Bruno Haible for CMUCL)
1302 * bug fix: Cyclic structures and unprintable objects in compiler
1303 messages no longer cause errors. (reported by Bruno Haible)
1304 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1305 and READ-FROM-STRING all now return a primary value of NIL if
1306 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1307 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1308 (reported by Bruno Haible for CMUCL)
1309 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1310 do not designate a condition. (reported by Bruno Haible for
1312 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1313 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1314 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1315 permitting resolution in favour of any of the conflicting symbols.
1316 (reported by Bruno Haible for CMUCL)
1317 * bug fix: EQUAL compiler optimizations is less aggressive on
1318 strings which can potentially compare true despite having distinct
1319 specialized array element types.
1320 * bug fix: unit enumerations can be defined without dividing by
1321 zero. (thanks to Vincent Arkesteijn)
1322 * FORMAT compile-time argument count checking has been enhanced.
1323 (report from Bruno Haible for CMUCL)
1324 * a partial workaround for the bug 262: the compiler does not try to
1325 inline-expand a local function doing RETURN-FROM from a deleted
1326 BLOCK. (thanks to Peter Denno for the bug report and to David
1327 Wragg for the simple test case)
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** RENAME-PACKAGE allows all package designators as new package
1331 ** constraint propagation and lambda variable substitution are
1332 more cautious in dealing with partially deleted code.
1333 ** compiler sometimes lost reoptimization passes.
1334 ** CERROR, given a condition as condition designator, passes its
1335 remaining arguments to the continue format control without
1337 ** Case-altering FORMAT directives work correctly on non-ASCII
1339 ** The REAL type specifier handles bounds outside the floating
1340 point ranges without signalling FLOATING-POINT-OVERFLOW.
1341 ** Functions with IR1-transformations can create intercomponent
1342 references to global functions.
1343 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1345 ** FORMAT ~R treats a nil value for its first parameter correctly.
1347 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1348 * enhancement: saving cores with foreign code loaded is now
1349 supported on x86/NetBSD and sparc/Linux in addition to the previously
1350 supported platforms.
1351 * bug fix: on some platforms repeated installations caused multiple
1352 copies of HTML documentation to be installed -- should not happen
1353 any more. (reported by Stefan Scholl)
1354 * bug fix: parsing self-recursive alien record types multiple times
1355 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1356 original patch by Helmut Eller for CMUCL)
1357 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1358 (thanks to Richard Kreuter)
1359 * bug fix: defining classes whose accessors are methods on existing
1360 generic functions in other (locked) packages no longer signals
1361 bogus package lock violations. (reported by François-René Rideau)
1362 * bug fix: special variables as DEFMETHOD parameters no longer have
1363 associated bogus type declarations. (reported by David Wragg and
1365 * bug fix: read-write consistency on streams of element-type
1366 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1367 * bug fix: redefiniton of the only method of a generic function with
1368 no DEFGENERIC no longer emits a full WARNING. In addition,
1369 redefinition of generic functions with no DEFGENERIC to an
1370 incompatible lambda list now signals an error. (thanks to Zach
1372 * bug fix: DEFGENERIC now works even when there's a function of the
1373 same name in an enclosing lexical environment. (thanks to Zach
1375 * fixed compiler failure, caused by instrumenting code during
1376 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1377 * optimization: added loop analysis and improved register allocation
1378 * fixed some bugs revealed by Paul Dietz' test suite:
1379 ** POSITION on displaced vectors with non-zero displacement
1380 returns the right answer.
1381 ** (SIMPLE-STRING) is a valid type specifier for sequence
1383 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1385 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1386 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1388 ** The justification version of the FORMAT ~< directive treats
1389 non-zero minpad parameter correctly.
1391 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1392 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1393 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1394 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1395 the supported interface.
1396 * new feature: Single-stepping of code compiled with DEBUG 2 or
1397 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1398 * new feature: saving cores with foreign code loaded is now
1399 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1400 Timothy Moore's work for CMUCL)
1401 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1402 parameters to * instead of NIL if no initform is supplied.
1403 (reported by Johan Bockgaard)
1404 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1405 correctly when FUNCALL appears as the car of the form. Note:
1406 despite this FUNCALL forms are not currently subject to
1407 compiler-macro expansion. (port of Raymond Toy's fix for the
1408 same from CMUCL, reported by Johan Bockgaard)
1409 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1410 (thanks for Teemu Kalvas)
1411 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1412 case instead of terminating the iteration (thanks for Julian
1413 Squires, Sean Champ and Raymond Toy)
1414 * bug fix: incorrect expansion of defgeneric that caused a style
1415 warning. (thanks for Zach Beane)
1416 * bug fix: slot accessor effective method computation works properly
1417 for classes with multiple non-standard applicable methods on
1418 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1419 * on x86 compiler supports stack allocation of results of LIST and
1420 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1421 CMUCL implementation by Gerd Moellmann)
1422 * fixed some bugs revealed by Paul Dietz' test suite:
1423 ** FORMAT strings with both the ~<~:;~> form of the justification
1424 directive and pretty-printing directives cause an error.
1426 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1427 * incompatible change: the internal functions
1428 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1429 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1430 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1431 instead of the old functions.
1432 * new feature: on platforms where "dladdr" is available foreign
1433 function names now appear in backtraces. (based on Helmut Eller's
1435 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1436 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1438 * documentation: ASDF manual is now distributed with SBCL.
1439 * bug fix: SBCL can now load its contributed modules with REQUIRE
1440 even if the system-provided entries have been removed from
1441 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1443 * bug fix: backtraces involving undefined functions or assembly
1444 routines are more informative. (thanks to Brian Downing)
1445 * bug fix: mutually referent alien structures now work correctly.
1446 (reported by Rick Taube)
1447 * bug fix: structures defined by WITH-ALIEN can be referred to
1448 within other definitions in the same WITH-ALIEN.
1449 * bug fix: division operators (MOD, TRUNCATE and the like) with
1450 constant zero divisors and integer dividends no longer generate
1452 * bug fix: provide default methods for INPUT-STREAM-P and
1453 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1454 * bug fix: improve the detection and resolution of MOP metacycles
1455 (where effective method computation depends on the generic function
1456 having its effective method computed). (reported by Bruno Haible)
1457 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1458 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1459 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1460 * bug fix: pathnames with relative directory components can be
1461 represented relative to default pathnames.
1462 * optimization: in taking the GCD of bignums, reduce the two bignums
1463 to approximately the same size (using Euclid's algorithm) before
1464 applying the more sophisticated binary GCD. (thanks to Juho
1466 * optimization: COUNT on bitvectors now operates word-at-a-time.
1467 * optimization: ASH with a positive, but not necessarily constant,
1468 (leftwards) shift, when in a modular context, is compiled to a
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1472 if the corresponding argument is NIL.
1473 ** FORMAT directives accepting a minpad parameter treat negative
1475 ** literal commas as character parameters to FORMAT directives are
1477 ** literal spaces directly after ~<Newline> directives within a
1478 format-logical-block (~:< ~@:>) do not induce :FILL-style
1479 conditional newlines.
1480 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1481 as specified: it no longer includes conditional newlines).
1482 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1484 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1485 :PER-LINE-PREFIX argument does not evaluate to a string.
1487 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1488 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1489 the manual for details; for now, package locks can be disabled by
1490 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1491 if no major problems are found then it is likely that they will be
1492 compiled in unconditionally.
1493 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1494 unsupported operators on all platforms. To load a shared library
1495 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1496 object file, link it into a shared library outside of SBCL and
1497 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1499 * bug workaround: ROOM T and the GC were not working together
1500 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1501 reporting facility aren't preserved. That reporting has been
1502 disabled, so now until and unless someone figures out how to make it
1503 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1504 * minor incompatible change: as threatened around sbcl-0.8.0, the
1505 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1506 The symbols are also exported from SB-PCL for backwards
1507 compatibility, but more so than before SB-PCL should be treated as
1508 an implementation-internal package.
1509 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1511 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1512 It is known as of this release not to work on the Alpha, however.
1513 * fixed bug #167: errors signalled due to illegal syntax in method
1514 bodies are now more legible.
1515 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1516 designators and can hence be used with TYPEP.
1517 * fixed bug #333: CHECK-TYPE now ensures that the type error
1518 signalled, if any, has the right object to be accessed by
1519 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1520 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1521 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1523 * fixed bug #334: programmatic addition of slots using specialized
1524 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1525 and :ALLOCATION :CLASS slots.
1526 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1527 float exponents. (rereported by Peter Seibel)
1528 * fixed a bug: #\Space (and other whitespace characters) are no
1529 longer considered to be macro characters in standard syntax by
1530 GET-MACRO-CHARACTER.
1531 * fixed bug: initialization of condition class metaobjects no longer
1532 causes an instance of the condition to be created. (reported by
1534 * fixed bug: it is now possible to have slots such that
1535 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1536 description is neither :INSTANCE nor :CLASS.
1537 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1538 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1539 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1540 non-standard SLOT-DEFINITION-ALLOCATION values.
1541 * fixed bugs: various race conditions handling exiting threads.
1542 CL-PPCRE's thread torture test now passes.
1543 * fixed arguably-a-bug: GC time proportional to number of threads in
1544 system even when most of them are idle
1545 * optimization: improved performance of BIT and SBIT on bit-vectors.
1546 * .fasl file incompatibility: The fasl file version number has
1547 been incremented because of changes associated with package locks.
1549 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1550 * minor incompatible change: the system no longer provides
1551 optimization hints (conditions which are TYPEP
1552 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1553 mode; these hints are still emitted when higher SPEED optimization
1555 * new contrib module: a sampling profiler (profiling by statistical
1556 sampling, rather than by instrumenting functions) is available as
1557 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1558 * the behaviour of the standard function ED is now customizeable by
1559 third parties through a hook variable: see ED's documentation
1560 string for information on the protocol.
1561 * the compiler no longer emits efficiency notes for (FUNCALL X)
1562 when the type of X is uncertain under default optimization
1564 * fixed bug 276: mutating a binding of a specialized parameter to a
1565 method to something that is not TYPEP the specializer is now
1567 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1568 exhibit the required behaviour on the x86 platform. (thanks to
1569 Peter van Eynde, Eric Marsden and Bruno Haible)
1570 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1571 even for difficult arguments. (reported by Peter Graves)
1572 * fixed bug 141a: the backquote printer now descends quoted
1574 * fixed another bug in backquote printing: no more destructive
1575 modification of the form's list structure. (reported by Brian
1577 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1578 move between its address being taken and the call to
1579 interrupt_thread, fixing a crashing race condition.
1580 * the SB-POSIX contrib implementation has been adjusted so that it
1581 no longer exhibits ridiculously poor performance when constructing
1582 instances corresponding to C structs.
1584 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1585 * minor incompatible change: the sb-grovel contrib now treats C
1586 structures as alien (in the sense of SB-ALIEN) objects rather than
1587 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1588 has implications for memory management of client code
1589 (sb-grovel-returned objects must now be manually managed) and for
1590 type safety (alien objects now have full types).
1591 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1592 used to control emission of compiler diagnostics, rather than the
1593 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1594 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1595 quality should be considered deprecated.
1596 * install.sh now installs the user manual as well
1597 * (not quite a new documentable feature, but worth considering in
1598 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1599 beginnings of a semantically meaningful condition hierarchy is
1600 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1602 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1603 designator as the defaults argument.
1604 * fixed bug: Displaced arrays whose displaced-to array has become
1605 too small now cause ARRAY-DIMENSION to signal an error, providing
1606 for safer bounds-checking. (reported by Bruno Haible)
1607 * fixed bug: DEFCLASS slot definitions with identical :READER and
1608 :WRITER names now signal a reasonable error. (reported by Thomas
1610 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1612 * fixed bug: Invalid slot specification errors now print correctly.
1613 (thanks to Zach Beane)
1614 * fixed bug 320: Shared to local slot value transfers in class
1615 redefinitions now happen corrently from superclasses as
1616 well. (reported by Bruno Haible)
1617 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1619 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1620 improper lists correctly. (reported by Bruno Haible)
1621 * fixed bug 313: source-transform for <fun-name> was erroneously
1622 applied to a call of a value of a variable with name <fun-name>.
1623 (reported by Antonio Menezes Leitao)
1624 * fixed bug 307: The obsolete instance protocol ensures that
1625 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1626 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1627 remove dead unknown-values globs from the middle of the stack.
1628 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1629 easier to resume long computations after using *BREAK-ON-SIGNALS*
1630 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1631 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1632 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1633 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1634 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1635 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1637 * fixed a (fairly theoretical) bug in string printing: if
1638 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1639 does not have array-element-type equal to the most general string
1641 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1642 function name. (thanks to Nikodemus Siivola)
1643 * fixed bug: compile-time format string checker failed on
1644 non-closed ~{. (reported by Thomas F Burdick)
1645 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1646 don't warn on legal loop code involving more than one aggregate
1647 boolean. (thanks to Nikodemus Siivola)
1648 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1649 in keyword position to MAKE-PACKAGE was not being reported as
1651 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1652 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1654 * fixed bugs 280 and 312: the checking for multiple definitions in a
1655 file is less likely to become confused by uses of inline
1657 * fixed bug: the #S reader macro performs the keyword coercion
1658 specified for slot names. (reported by Kalle Niemitalo)
1659 * fixed bug: lambda lists may contain symbols whose names start with
1660 & but are not lambda-list-keywords; their occurrence triggers a
1662 * fixed bug 321: define-method-combination argument lambda lists do
1663 not cause invalid code to be generated when &optional or &aux
1664 variables have default values. (reported by Bruno Haible)
1665 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1666 have CLOS classes; this appears also to have fixed reported
1667 instability in stack exhaustion detection.
1668 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1669 MUFFLE-WARNING when no associated restart is present is now
1671 * optimization: rearranged the expansion of various defining macros
1672 so that each expands into only one top-level form in a
1673 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1675 * optimization: used a previously little-used slot in symbols to
1676 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1678 * fixed some bugs revealed by Paul Dietz' test suite:
1679 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1681 ** Fixed an optimization bug related to inheritance of initforms
1682 from local to shared slots.
1683 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1684 ** CAST optimizer forgot to flush argument derived type.
1685 ** print/read consistency on floats is now orders of magnitude
1686 more likely. (thanks also to Bruno Haible for a similar report
1688 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1689 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1690 for objects of type REAL. Make it so.
1691 ** SXHASH is specified (infelicitously) to respect similarity,
1692 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1693 it so. (thanks to Markus Ziegler)
1694 ** on the Alpha, the compiler succeeds in compiling functions
1695 returning a known number of arguments greater than 63.
1696 ** fixed handling of invalid NIL arguments in keyword position
1698 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1700 ** provided a readably-printable representation for RANDOM-STATE
1702 ** ensured that pathnames, where they have a namestring, always
1703 print using #P"..." syntax.
1705 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1706 * Support for the forthcoming 2.0 version of the NetBSD kernel
1707 running on x86 hardware has been added. (thanks to Perry
1708 E. Metzger most immediately, and others for their past work)
1709 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1710 current release version; SBCL's OpenBSD support had been broken
1711 since about the time of OpenBSD's switch to ELF binary format.)
1712 * [placeholder for DX summary]
1713 ** user code with &REST lists declared dynamic-extent, under high
1714 speed or space and low safety and debug optimization policy.
1715 * The manual has been converted to Texinfo format and the debugger
1716 chapter from the cmucl manual has been added.
1717 * A facility has been added to extract documentation strings from
1718 sbcl and store them as Texinfo-formatted snippets for inclusion in
1719 the manual (via Texinfo's @include directive)
1720 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1721 the test case to Dave Roberts)
1722 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1723 pointers. (reported by Sean Ross)
1724 * bug fix: PROFILE output is printed nicely even for large numerical
1725 values. (thanks to Zach Beane)
1726 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1727 greater than 32 handle EOF correctly.
1728 * bug fix: on X86 an immediate argument of the IMUL instruction is
1729 correctly printed in disassembly. (thanks to Lutz Euler)
1730 * bug fix: class slots in redefined classes preserve their old
1731 values. (thanks to Bruno Haible and Nikodemus Siivola)
1732 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1733 succeeds. (reported by Marco Baringer)
1734 * fixed some bugs revealed by Paul Dietz' test suite:
1735 ** READ-SEQUENCE now works on ECHO-STREAMs.
1736 ** RATIONALIZE works more according to its specification. (thanks
1739 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1740 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1741 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1742 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1743 should matter to you only if you rebind the printer control
1744 variables and then find you want different bindings in the
1745 debugger than in the ordinary execution of your program.)
1746 * The runtime build system has been tweaked to support building
1747 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1748 assembler and linker. (thanks to Nikodemus Siivola)
1749 * Unbound, undefined, undeclared variables now trigger full
1750 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1751 more often programmer error than deliberate exploitation of undefined
1753 * optimization: the hash algorithm for strings has changed to one
1754 that is less vulnerable to spurious collisions. (thanks to Juho
1756 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1757 needless bounds checking. (thanks to Juho Snellman)
1758 * optimization: implemented multiplication as a modular
1759 (UNSIGNED-BYTE 32) operation on the PPC backend.
1760 * fixed some bugs revealed by Paul Dietz' test suite:
1761 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1763 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1764 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1765 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1767 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1769 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1771 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1772 their output stream on EOF from read.
1773 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1774 have been read to end-of-file.
1775 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1777 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1778 description of determination of which consecutive characters
1780 ** Printing the "Space" character with escaping on now yields "#\\ ",
1781 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1782 ** Reading floating-point numbers with *READ-BASE* set to a number
1783 less than 10 works correctly.
1784 ** Reading floating-point numbers with *READ-BASE* set to a number
1785 more than 10 works correctly.
1786 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1787 the readtable currently in effect.
1789 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1790 * minor incompatible change: parsing of namestrings on a physical
1791 (Unix) host has changed; numbers after the final #\. in a
1792 namestring are no longer interpreted as a version field. This is
1793 intented to be largely invisible to the user, except that the
1794 meaning of the namestring "*.*.*" has changed: it now refers to a
1795 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1796 should usually be replaced by
1797 (make-pathname :name :wild :type :wild :version :wild)
1798 with the added benefit that this is more likely to be portable.
1799 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1800 option to OPEN now signals an error if the file being opened
1801 exists; this may have an impact on existing code.
1802 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1803 signals received, on the PowerPC platforms (both Linux and
1804 Darwin). (thanks to Pierre Mai for pointing out the location of
1806 * several fixes on OS X: The system now builds and runs cleanly on
1807 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1808 (thanks to Brian Mastenbrook)
1809 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1810 second-resolution: integer multiples of 1/3600 between -24 and 24.
1811 (thanks to Vincent Arkesteijn)
1812 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1813 type when called with 1 argument; PEEK-CHAR checked type of
1814 PEEK-TYPE only after having read first character from a
1815 stream. (reported by Peter Graves)
1816 * bug fix: the garbage collector now has much better locality
1817 behaviour, and in particular no longer treats all memory as being
1818 exhausted when many small objects point to each other in a deeply
1820 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1821 recognized as being TYPEP their class.
1822 * bug fix: the PUSHNEW documentation string has been corrected.
1823 (thanks to Vincent Arkesteijn)
1824 * bug fix: defaulting of the value for the last of an atypically
1825 large number of multiple values being bound was not being
1826 performed correctly on the Alpha or PPC platforms
1827 * optimization: implemented multiplication as a modular
1828 (UNSIGNED-BYTE 32) operation on the x86 backend.
1829 * optimization: SEARCH on simple-base-strings can now be open-coded.
1830 (see also contrib/compiler-extras.lisp for inspiration for
1831 teaching the compiler about the Boyer-Moore algorithm).
1832 * value, returned by MAX (and MIN) called with several EQUALP, but
1833 not EQL, arguments now does not depend on compiler settings.
1834 * fixed some bugs revealed by Paul Dietz' test suite:
1835 ** in stack analysis liveness information is propagated from
1836 non-local entry points.
1837 ** pathwise CAST removing failed when the CAST node did not start
1839 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1840 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1842 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1843 host is already defined.
1844 ** RENAME-FILE works on streams instead of signalling an internal
1846 ** PEEK-CHAR uses the current readtable when determining whether
1847 or not a character is whitespace.
1848 ** MERGE-PATHNAMES handles the case when the pathname does not
1849 specify a name while the default-pathname specifies a version.
1850 ** Pathnames now stand a better chance of respecting print/read
1852 ** Attempting to use standardized file system operators with a
1853 pathname with invalid :DIRECTORY components signals a
1855 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1857 ** DIRECTORY on logical pathnames is more correct.
1858 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1859 signal a TYPE-ERROR if their argument is not a stream.
1860 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1861 designator argument does not designate a stream.
1862 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1863 examining the synonym.
1864 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1866 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1867 element-type larger than ([UN]SIGNED-BYTE 32).
1869 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1870 * When built with the :SB-FUTEX feature, threaded builds now take
1871 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1872 for faster/more reliable mutex and condition variable support.
1873 * Incompatible change (but one you probably shouldn't have been using
1874 anyway): the interface and code for arbitrating between multiple
1875 threads in the same user session has been redesigned.
1876 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1877 error when called without an explicit environment argument.
1878 (thanks to Nikodemus Siivola)
1879 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1880 confused about their position. (thanks to Adam Warner and Gerd
1882 * bug fix: Pretty printing backquoted forms with unquotations in the
1883 argument list position of various code constructs such as LAMBDA
1884 now works correctly. (reported by Paul Dietz)
1885 * bug fix: Pretty printing unquotations no longer loses all
1886 stream position information.
1887 * optimization: performance of string output streams is now less
1888 poor for multiple small sequence writes.
1889 * optimization: performance of CSUBTYPEP in the presence of complex
1890 expressions involving CONS and NOT many times has been improved.
1891 (reported by Paul Dietz)
1892 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1894 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1895 streams for the REPL, for the debugger, and so on. (thanks to
1897 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1898 * fixed some bugs revealed by Paul Dietz' test suite:
1899 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1900 is ignored if it is too small, rather than propagating through
1901 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1902 ** extremely complex negations of CONS types were not being
1903 sufficiently canonicalized, leading to inconsistencies in
1905 ** VALUES tranformer lost derived type.
1907 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1908 * fixed a bootstrapping bug: the build process no longer assumes
1909 that the various BOOLE-related constants have the same value in
1910 host and target lisps. (noted by Paul Dietz' test suite on an
1911 SBCL binary built from CLISP)
1912 * The system can now be dynamically linked on the MIPS platform,
1913 which enables dynamic loading of foreign code from Lisp. (thanks
1914 to Ralf Baechle for discussions on the MIPS ABI)
1915 * The system now records debugging information for its own source
1916 files in a filesystem-position-independent manner, relative to
1917 the "SYS" logical host.
1918 * fixed a compiler bug: MV-LET convertion did not check references
1919 to the "max args" entry point. (reported by Brian Downing)
1920 * tweaked disassembly notes to be less confident about proclaiming
1921 some instruction as an LRA. (thanks to Brian Downing)
1922 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1923 now each have their own history, command character, and other
1924 characteristics. (thanks to David Lichteblau)
1925 * fixed some bugs revealed by Paul Dietz' test suite:
1926 ** compiler failure in compiling LOGAND expressions including a
1928 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1929 shift greater than 32.
1930 ** FLUSH-DEST did not mark blocks for type check regeneration.
1931 ** HANDLER-CASE failed to accept declarations in handler clauses
1932 in some circumstances.
1934 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1935 * New code in contrib/sb-introspect (still probably not entirely
1936 stable yet) provides some support for smart Lisp development
1937 environments like SLIME.
1938 * The conditions signalled for errors occurring when loading .fasl
1939 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1940 in a way which should help ASDF recover gracefully.
1941 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1942 stuff has been cleaned up. If you code contrib/ stuff, this might
1943 affect you, and you can look at contrib/README, contrib/STANDARDS,
1944 and/or the 0.8.4.27 diff to check.
1945 * In full calls the compiler now does not generate checks for declared
1946 argument types for all arguments.
1947 * various threading fixes
1948 ** and some experimental patches which didn't make it into
1949 the main tree for this release, but which are shipped in
1950 contrib/experimental-thread.patch as a possible fix for some
1951 failures (deadlock, spinning...) in GC-intensive multithreaded
1953 * fixed PPC build problem (source code incompatibility of different
1954 library versions): added offsetof() hackery which attempts to divine
1955 where glibc maintainers put uc_mcontext today
1956 * fixed bug 282: compiler does not trust type assertions while passing
1957 arguments to a full call.
1958 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1959 &OPTIONAL VALUES type parameter.
1960 * fix bug 214: algorithm for noting rejected templates is now more
1961 similar to that of template seletion. (also reported by rydis on
1963 * fixed bug 141b: printing backquoted information readably and prettily
1964 inserts a space where necessary.
1965 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1966 (CHARACTER 10) are now reported as errors, rather than propagated
1967 as unknown types. (reported by piso on #lisp)
1968 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1969 with values NIL and :ERROR. (thanks to Milan Zamazal)
1970 * fixed bug 191c: CLOS now does proper keyword argument checking as
1971 described in CLHS 7.6.5 and 7.6.5.1.
1972 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1973 longer raise an error; further, using a list as a for-as-arithmetic
1974 counter now raises a meaningful error.
1975 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1976 now understood by sequence creation functions such as MAKE-SEQUENCE
1978 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1979 a STRING-INPUT-STREAM.
1980 * compiler enhancement: SIGNUM is now better able to derive the type
1982 * type declarations inside WITH-SLOTS are checked. (reported by
1984 * fixed some bugs revealed by Paul Dietz' test suite:
1985 ** incorrect optimization of TRUNCATE for a positive first
1986 argument and negative second.
1987 ** compiler failure in let-convertion during flushing dead code.
1988 ** compiler failure while deriving type of TRUNCATE on an
1989 interval, containing 0.
1990 ** ASH of a negative bignum by a negative bignum count now returns
1992 ** intersection of CONS types now canonicalizes properly, fixing
1993 inconsistencies in SUBTYPEP.
1995 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1996 * incompatible change: The --disable-debugger command line
1997 option now clobbers the debugger at a more fundamental
1998 level, by redefining #'INVOKE-DEBUGGER instead of by
1999 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2000 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2001 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2002 enter the debugger REPL and then suffer recursive errors
2003 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2004 while under the new system, BREAK in --disable-debugger mode
2005 terminates the system just as an unhandled error would.
2006 * fixed compiler performance when processing loops with a step >1;
2007 * bug fix: DOCUMENTATION now retrieves generic function
2008 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2009 support has been systematized, and now supports the methods
2010 specified by ANSI, along with a default method and a method for
2011 slot documentation. (reported by Nathan Froyd)
2012 * bug fix: effective methods associated with a generic function are
2013 no longer cached over a change of that generic function's method
2014 combination. (reported by Andreas Fuchs)
2015 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2016 names a non-existent directory. (thanks to Andreas Fuchs)
2017 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2018 the PPC platform, lead to stack corruption; this has been fixed.
2019 (reported by Rainer Joswig)
2020 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2021 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2022 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2023 arithmetic in general, is now much more reliable on the PPC
2025 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2026 platform now returns the right answer.
2027 * optimization: restored some effective method precomputation in
2028 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2029 precomputation is now tunable.
2030 * optimization: compiler-internal data structure use has been
2031 reviewed, and changes have been made that should improve the
2032 performance of the compiler by about 20%.
2033 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2034 simple-base-strings and simple-bit-vectors is improved.
2035 * optimization: the optimization of 32-bit logical and arithmetic
2036 functions introduced in version 0.8.3 on the x86 has been
2037 implemented on the mips, ppc and sparc platforms; an
2038 implementation of the same facility, but for 64-bit arithmetic,
2039 has been added for the alpha.
2040 * microoptimization: the compiler is better able to make use of the
2041 x86 LEA instruction for multiplication by constants.
2042 * bug fix: in some situations compiler did not report usage of
2043 generic arithmetic in (SPEED 3) policy.
2044 * bug 145b fix: compiler used wrong type specifier while converting
2045 MEMBER-types to numeric.
2046 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2048 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2049 index argument now works properly on the Alpha platform.
2050 * bug fix: floating point exception treatment on the Alpha platform
2052 * bug fix: FILE-POSITION works much better on string input and
2053 output streams. (thanks to Nikodemus Siivola)
2054 * bug fix: many threading/garbage collection symptoms sorted.
2055 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2056 might be pseudo-atomic.
2057 * internal change: Stopping for GC is now done with signals not
2058 ptrace. GC is now done in whichever thread wanted it, instead of
2060 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2062 * bug fix: result form in DO is not contained in the implicit
2064 * incompatible change: ICR structure is changed; the value part of
2065 CONTINUATION is now called LVAR; corresponding functions are
2066 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2067 * added type deriver for ISQRT (thanks to Robert E. Brown).
2068 * bug fix: better support for loading from the command line when an
2069 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2071 * fixed some bugs revealed by Paul Dietz' test suite:
2072 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2073 ** ROUND and FROUND now give the right answer when given very
2074 small float arguments.
2075 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2077 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2078 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2079 ** LCM should return a non-negative integer.
2080 ** PARSE-INTEGER returned the index of a terminator instead of the
2081 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2082 ** PARSE-INTEGER returned an incorrect index being applied to a
2084 ** LCM with two arguments of 0 returns 0 rather than signalling
2086 ** unsigned addition of a 32-bit constant with the high bit set no
2087 longer causes an internal compiler error on the x86.
2088 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2089 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2090 with negative last argument.
2091 ** byte specifiers with zero size and position no longer cause
2092 an error during type derivation.
2093 ** bignum multiplication on the Alpha platform now returns the
2095 * porting: The system now builds on SuSE AMD64, although it still
2096 generates a 32-bit binary.
2097 * .fasl file incompatibility: The fasl file version number has
2098 been incremented (because of the changes to internal compiler
2099 data structures referred to above).
2101 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2102 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2103 more accurately, on the Darwin kernel running on PowerPC hardware.
2104 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2105 * Compiler code deletion notes now signal a condition of type
2106 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2107 an associated MUFFLE-WARNING restart.
2108 * The compiler now performs limited argument count validation of
2109 constant format strings in FORMAT, and where appropriate in ERROR,
2110 CERROR and WARN. (thanks to Gerd Moellmann)
2111 * New ASDF-INSTALL contrib can be used for automatic download and
2112 installation of third-party Lisp code from CCLAN or other sites
2114 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2115 which forces another thread to execute a function supplied by the
2117 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2118 now accept and act upon their :ELEMENT-TYPE keyword argument.
2119 (reported by Martin Atzmueller, Edi Weitz)
2120 * bug fix: FILE-POSITION now accepts position designators up to
2121 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2122 is the greater. (thanks to Patrik Nordebo)
2123 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2124 Kalle Olavi Niemitalo)
2125 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2126 printable. (reported by Eric Marsden)
2127 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2128 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2129 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2130 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2131 (thanks to Stig Sandoe)
2132 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2133 UNREAD-CHAR now correctly includes the unread character in the
2134 target sequence. (thanks to Gerd Moellmann)
2135 * bug fix in threaded builds: the system can now be suspended and
2136 resumed by shell job control with minimal disruption.
2137 * bug fixes in times and timezones >2038AD
2138 * better handling of "where is GNU make?" problem in build scripts
2139 (thanks to Nikodemus Siivola)
2140 * new optimization: inside a named function any reference to a
2141 function with the same name is considered to be a self-reference;
2142 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2143 optimization quality.
2144 * new optimization on x86: logical functions and + now have
2145 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2146 used when the result is truncated to 32 bits.
2147 * VALUES declaration is partially enabled.
2148 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2149 * fixed some bugs revealed by Paul Dietz' test suite:
2150 ** The system now obeys the constraint imposed by
2151 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2152 types form a lattice under type intersection.
2153 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2154 ** ASSOC now ignores NIL elements in an alist.
2155 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2156 and (1+ MOST-POSITIVE-FIXNUM) answers.
2157 ** The addition of a method with invalid qualifiers to a generic
2158 function does not cause an error to be signalled immediately;
2159 a warning is signalled, and the error is generated only on
2160 calling the generic function.
2161 * changed .fasl file version number, in part to add type codes for
2162 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2163 obscure ANSI requirements
2165 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2166 * fixed bug 148: failure to inline-expand a local function left
2167 garbage, confusing the compiler.
2168 * fixed bugs 3cd: structure slot readers perform type check if the
2169 slot can have an invalid value (i.e. it is either not initialized
2170 or can be written with a less specific slot writer).
2171 * bug fix: the compiler now traps array references to elements off
2172 the end of an array; previously, the bounds checking in some
2173 circumstances could go off-by-one.
2174 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2176 * type declarations for array element types now obey the description
2177 on the CLHS page "Declaration TYPE", as per discussions on
2178 sbcl-help around 2003-05-08. This means that a declaration
2179 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2180 declaration, all references to BAR will be asserted or assumed
2181 (with THE, so dependent on compiler policy) to involve objects of
2182 type FOO. Note that no such declaration is implied in
2183 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2184 * declared types of functions from the "Conditions"
2185 chapter. (reported by Paul Dietz)
2186 * bug fix: CERROR accepts a function as its first argument.
2187 * bug fix: NTH an NTHCDR accept a bignum as index
2188 arguments. (reported by Adam Warner)
2189 * optimization: character compare routines now optimize comparing
2190 against a constant character. (reported by Gilbert Baumann)
2191 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2192 argument now works properly on the MIPS platform.
2193 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2194 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2195 declared to have a functional type, does not cause an error
2197 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2198 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2199 * fixed bug: if last continuation of a deleted block has a
2200 destination, this destination should be deleted too. (reported by
2202 * fixed a bug in the bootstrap process: the host compiler's values
2203 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2204 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2205 test case from Patrik Nordebo)
2206 * improved the ability of the disassembler on the PPC platform to
2207 provide helpful disassembly notes.
2208 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2209 the class in more cases than previously.
2210 * bug fix: FILE-POSITION now understands :START and :END for
2211 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2212 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2213 8) data. (thanks to David Lichteblau)
2214 * bug fix: it is possible to add a method to a generic function
2215 without lambda list.
2216 * bug fix: reader failed to signal END-OF-FILE inside an
2217 object representation. (reported by Nikodemus Siivola)
2218 * fixed some bugs revealed by Paul Dietz' test suite:
2219 ** LAST and [N]BUTLAST should accept a bignum.
2220 ** condition slot accessors are methods.
2221 ** (VECTOR NIL) is a subtype of STRING.
2223 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2224 * minor incompatible change: some nonsensical specialized lambda
2225 lists (used in DEFMETHOD) which were previously ignored now signal
2227 * minor incompatible change: the system is now aware of the types of
2228 variables in the COMMON-LISP package, and will signal errors for
2229 most violations of these type constraints (where previously they
2230 were silently accepted).
2231 * minor incompatible change: COMPILE-FILE now uses the freedom
2232 afforded (ANSI 3.2.2.3) to use derived function types for
2233 functions defined in the same file. This also permits the system
2234 to warn on static type mismatches and function
2235 redefinition. (Currently it does not work with high DEBUG level.)
2236 * minor incompatible change: VALUES declaration is disabled.
2237 * When issuing notes, the compiler now signals a condition of type
2238 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2239 restart for use in user handlers. It is expected that the
2240 COMPILER-NOTE condition will eventually become a condition
2241 supertype to a hierarchy of note types, which will then be
2242 handleable in a similar fashion. However, at the moment, no such
2243 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2244 but it's an implementation detail, not a classification for the
2246 * Changes in type checking closed the following bugs:
2247 ** type checking of unused values (192b, 194d, 203);
2248 ** template selection based on unsafe type assertions (192c, 236);
2249 ** type checking in branches (194bc).
2250 * A short form of VALUES type specifier has ANSI meaning (it has
2251 increased the number of situations when SBCL cannot perform type
2253 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2254 keywords or constants is permissible.
2255 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2256 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2257 classes. (thanks to Antonio Martinez)
2258 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2259 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2260 argument to a pretty-print stream. Instead, it leaves any such
2261 operations to DESCRIBE-OBJECT methods.
2262 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2264 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2265 lambda lists. (thanks to David Lichteblau)
2266 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2267 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2268 (thanks to Gerd Moellmann)
2269 * fixed bug 246: increased compilation speed of long
2270 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2272 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2274 * DEFCONSTANT now throws a condition of type
2275 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2276 constant to a non-EQL value; CONTINUE and ABORT restarts
2277 respectively change and preserve the value.
2278 * fixed bug 63: The code walker, part of the implementation of CLOS,
2279 is now better at handling symbol macros.
2280 * bug fix: There is no longer an internal implementation type named
2281 CL:LENGTH. (reported by Raymond Toy)
2282 * bug fix: In macro-like defining macros/special operators the
2283 implicit block does not enclose the lambda list.
2284 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2285 enclosed in parentheses) are not suitable as type specifiers, and
2286 their use properly signals an error now.
2287 * bug fix: An argument count mismatch for a type specifier in code
2288 being compiled no longer causes an unhandled error at compile
2289 time, but signals a compile-time warning.
2290 * fixed simple vector readable printing
2291 * bug fix: DESCRIBE takes more care over whether the class
2292 precedence list slot of a class is bound before accessing it.
2293 (reported by Markus Krummenacker)
2294 * bug fix: FORMATTER can successfully compile pretty-printer format
2295 strings which use variants of the ~* directive inside.
2296 * bug fix: SEARCH now applies its TEST predicate to the elements of
2297 the arguments in the correct order. (thanks to Wolfhard Buss)
2298 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2299 in type inference. (thanks to Robert E. Brown)
2300 * bug fix: Reading in symbols with an explicit package name of ""
2301 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2302 * fixed some bugs revealed by Paul Dietz' test suite:
2303 ** NIL is now allowed as a structure slot name.
2304 ** Arbitrary numbers, not just REALs, are allowed in certain
2305 circumstances in LOOP for-as-arithmetic clauses.
2306 ** Multiple class redefinitions before slot access no longer
2307 causes a type error.
2308 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2309 association between the name and a class.
2310 ** Generic functions with non-standard method-combination and over
2311 five methods all of which return constants no longer return NIL
2312 after the first few invocations. (thanks to Gerd Moellmann)
2313 ** CALL-NEXT-METHOD with no arguments now passes the original
2314 values of the arguments, even in the presence of assignment.
2315 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2317 ** DISASSEMBLE works with closures and funcallable instances.
2318 ** ADD-METHOD now returns the generic function, not the new method.
2319 ** FIND-METHOD signals an error if the lengths of the specializers
2320 is incompatible with the generic function, even if the ERRORP
2322 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2323 which its argument is a member.
2324 ** DEFCLASS only redefines the class named by its class-name
2325 argument if that name is the proper name of the class;
2326 otherwise, it creates a new class.
2327 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2328 of the UNBOUND-SLOT condition to the name of the slot.
2329 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2330 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2331 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2332 SLOT-MAKUNBOUND in the specified fashion.
2334 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2335 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2336 cross-compilation host. As a consequence, we can now bootstrap our
2337 way up to SBCL starting with a bare gcc toolchain and human-readable
2338 source code (first the source to CLISP, then the source to SBCL).
2339 * A contributed module containing a partial implementation of the
2340 simple-streams interface has been included. (thanks to Rudi
2342 * A contributed module implementing the RFC1321 Message Digest
2343 Algorithm, known as MD5, has been included.
2344 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2345 no longer has any effect, as the code controlled by this feature
2346 has been deleted. (As far as we know, no-one has ever built using
2347 this feature, and its semantics were confused in any case).
2348 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2349 work on conditions (as required by the ANSI specification),
2350 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2351 expected behaviour on conditions. Users should note, however,
2352 that such behaviour is not required by the ANSI specification,
2353 and so use of this behaviour may render their code unportable.
2354 * fixed some bugs revealed by Paul Dietz' test suite:
2355 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2357 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2358 ** SXHASH on condition objects no longer returns NIL.
2359 ** :ALLOCATION :CLASS slots are better treated; their values are
2360 updated on class redefinition, and initforms inherited from
2361 superclasses are applied.
2362 ** REMOVE-METHOD returns its generic function argument even when
2363 no method was removed.
2364 ** SHARED-INITIALIZE now initializes the values of the requested
2365 slots, including those with :ALLOCATION :CLASS.
2366 ** ALLOCATE-INSTANCE now works on structure classes defined via
2367 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2369 ** SLOT-EXISTS-P now works on conditions, as well as structures
2371 ** MAKE-LOAD-FORM now has the required methods on
2372 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2373 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2374 keyword, and now implements the SLOT-NAMES argument.
2375 ** methods with &OPTIONAL arguments no longer allow too many
2376 arguments to be passed in the call without error.
2377 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2378 option is consistent with the required arguments of the generic
2379 function lambda list.
2380 * bug fix: REQUIRE accepts a string designator. (Thanks to
2382 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2383 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2384 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2385 * bug fix: The compiler now checks for duplicated variables in macro
2387 * bug fix: SETQ on globals returns the correct value.
2388 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2389 not a COMPILER-ERROR (followed by some other strange error on
2390 choosing the CONTINUE restart).
2391 * bug fix: make.sh and friends are now more consistent in the way that
2392 they look for GNU "make".
2394 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2395 * experimental native threads support (on x86 Linux >=2.4 only).
2396 This is not compiled in by default: you need to add :SB-THREAD to
2397 the target features. See the "Beyond ANSI" chapter of the manual
2399 * fix for longstanding nonANSIism: The old distinction between
2400 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2401 The return value from CL:FIND-CLASS is now a CLOS class, and
2402 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2403 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2404 * An interface to the MetaObject Protocol, as described in Kiczales,
2405 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2406 MIT Press, 1991, is available from the SB-MOP package.
2407 * incompatible change: the SB-PCL package should now be considered
2408 a private implementation detail, and no longer a semi-private MOP
2410 * minor incompatible change: due to rearrangement for threads, the
2411 control stack and binding stack are now allocated at arbitrary
2412 addresses instead of being hardcoded per-port. Users affected by
2413 this probably have to be doing advanced things with shared
2414 libraries, and will know who they are.
2415 * minor incompatible change: Previously, all --eval forms used were
2416 processed with READ before any of them were processed with EVAL.
2417 Now each --eval form is processed with both READ and EVAL before
2418 the next --eval form is processed. (Thus package operations like
2419 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2420 work as the user might reasonably expect.)
2421 * minor incompatible change: *STANDARD-INPUT* is now only an
2422 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2424 * minor incompatible change: Y-OR-N-P is now character-oriented,
2425 not line oriented. Also, YES-OR-NO-P now works without errors.
2426 (thanks to Antonio Martinez)
2427 * sb-aclrepl module improvements: an integrated inspector, added
2428 repl features, and a bug fix to :trace command.
2429 * Known functions, which cannot be open coded by the backend, are
2430 considered to be able to check types of their arguments. (fixing
2431 a bug report by Nathan J. Froyd)
2432 * fixed a bug in computing method discriminating functions: It is
2433 now possible to define methods specialized on classes which have
2434 forward-referenced superclasses. (thanks to Gerd Moellmann)
2435 * fixed evaluation order in optional entries (reported by Gilbert
2437 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2438 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2439 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2440 required-by-AMOP NAME argument, as well as CLASS and
2441 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2442 * fixed bug 20: DEFMETHOD can define methods using names that are
2443 not the proper names of classes to designate class specializers.
2444 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2445 Unix file descriptors, instead of blowing up. (thanks to Antonio
2447 * Garbage collection refactoring: user-visible change is that a
2448 call to the GC function during WITHOUT-GCING will not do garbage
2449 collection until the end of the WITHOUT-GCING. If you were doing
2450 this you were probably losing anyway.
2451 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2452 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2453 TYPEP the latter but not the former.
2454 * The compiler issues a full WARNING for calls to undefined functions
2455 with names from the CL package.
2456 * MAP-INTO for a vector destination is open coded. (reported by
2457 Brian Downing on c.l.l)
2458 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2459 documentation string.
2460 * fixed some bugs revealed by Paul Dietz' test suite:
2461 ** COPY-ALIST now signals an error if its argument is a dotted
2463 ** Condition slots are now accessed more correctly in the presence
2464 of multiple initargs for a given slot.
2465 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2466 exclude restarts of the same name associated with a different
2468 ** DEFCLASS of forward-referenced classes with another
2469 forward-referenced class in the superclasses list no longer
2471 ** Condition slots are now initialized once each, not multiple
2472 times. (thanks to Gerd Moellmann)
2473 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2475 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2476 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2477 arguments contain duplicated elements.
2478 ** RESTART-CASE understands local macros.
2479 ** RESTART-CASE associates exactly its own restarts with a condition.
2480 ** ENDP in safe mode checks its argument to be of type LIST.
2481 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2482 in question is unbound.
2483 ** Optimized MAKE-INSTANCE functions no longer cause internal
2484 assertion failures in the presence of duplicate initargs.
2485 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2486 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2487 propagated even in the ctor optimized implementation of
2489 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2491 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2492 * a better implementation of SXHASH on (simple) bit vectors,
2493 measured both in execution speed and in distribution of results
2494 over the positive fixnums, has been installed. Likewise, a better
2495 implementation of EQUAL for simple bit vectors is now available.
2496 * fixed CEILING optimization for a divisor of form 2^k.
2497 * fixed bug 240 (emitting extra style warnings "using the lexical
2498 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2499 by Antonio Martinez)
2500 * fixed SXHASH, giving different results for NIL depending on type
2501 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2502 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2503 variable. (found by Rolf Wester)
2504 * SBCL does not ignore type declarations for special
2505 variables. (reported by rif on c.l.l 2003-03-05)
2506 * some bug fixes in contrib/sb-aclrepl/
2507 * fixed some bugs revealed by Paul Dietz' test suite:
2508 ** a bug in the CONS type specifier, whereby the CAR and CDR
2509 types got intertwined, has been fixed;
2510 ** the type system is now able to reason about the interaction
2511 between INTEGER and RATIO types more completely;
2512 ** APPEND, [N]REVERSE and NRECONC check that those their
2513 arguments, which must be proper lists, are really so;
2514 ** An array specialized to be unable to hold elements has been
2515 implemented, as required -- yes, really -- by ANSI;
2516 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2517 on malformed property lists;
2519 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2520 * incompatible packaging change: in line with Unix convention,
2521 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2522 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2523 * REQUIRE and PROVIDE are now optionally capable of doing something
2524 useful. See the documentation string for REQUIRE.
2525 * infrastructure for a managed SBCL contrib system: contributed
2526 modules in this release include:
2527 ** the ASDF system definition facility;
2528 ** an interface to the BSD Sockets API;
2529 ** an ACL-like convenience interface to the repl;
2530 (thanks to Kevin Rosenberg)
2531 ** an implementation of ROTATE-BYTE, with efficient implementation
2533 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2534 gives a closer approximation to the right answer than previously.
2535 (thanks to Raymond Toy)
2536 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2537 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2538 optional environment arguments, as required by ANSI.
2539 * fixed bugs in other functions taking environment objects, allowing
2540 calls with an explicit NIL environment argument to be compiled
2542 * fixed bug 228: primary return values from
2543 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2544 COMPILE or FUNCTION.
2545 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2546 :TYPE will no longer signal errors on innocuous objects.
2547 * fixed bug 231b: SETQ is better at respecting type declarations in
2548 the lexical environment.
2549 * fixed a bug in DEFCLASS: classes named by symbols with no or
2550 unprintable packages can now be defined.
2551 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2552 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2553 * fixed compiler failure related to checking types of functions.
2554 (reported by Robert E. Brown)
2555 * the compiler is now much more consistent in its error-checking
2556 treatment of bounding index arguments to sequence functions: in
2557 (SAFETY 3) code, errors will be signalled in almost all cases if
2558 invalid sequence bounding indices are passed to functions defined
2559 by ANSI to operate on sequences.
2560 * fixed a bug in the build procedure: documentation of SBCL-specific
2561 packages is now preserved and available in the final Lisp image.
2562 * lifted FDEFINITION lookup out of loops in the implementation of
2563 many list operations. (thanks to Robert E. Brown)
2564 * fixed a bug in the reader: the #n# reader macro now works for
2565 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2566 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2567 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2568 E. Brown and Tony Martinez respectively)
2569 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2570 SLOT-BOUNDP: in particular, it is now able to optimize them much
2571 better, and is now not vulnerable to having packages renamed.
2572 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2573 always called when a slot is not present in an instance. (thanks
2575 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2576 longer signal an error on unreading a character following EOF on
2577 the previous constituent stream. (thanks to Tony Martinez)
2578 * fixed some bugs revealed by Paul Dietz' test suite:
2579 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2580 not just nonnegative fixnums;
2581 ** the logical bit-array operators such as BIT-AND now accept an
2582 explicit NIL for their "opt-arg" argument (to indicate a
2583 freshly-consed result bit-array);
2584 ** ELT now signals an error on an invalid sequence index in safe
2586 ** the type system is now cleverer about negations of numeric
2587 types, and consequently understands the BIGNUM and RATIO types
2589 ** the type system is now cleverer about the interaction between
2590 INTEGER and RATIO types: while bugs still remain, many more
2591 cases are accurately computed;
2592 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2593 if it is in the last clause;
2594 ** CONSTANTLY now correctly returns a side-effect-free function in
2596 ** DECLARE is no longer treated as a special-operator; in
2597 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2598 * incremented fasl file version number due to the change in the
2599 DEFSTRUCT-SLOT-DESCRIPTION structure.
2601 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2602 * minor incompatible change: code processed by the "interpreter" or
2603 EVAL now has a compilation optimization policy of (DEBUG 2)
2604 (changed from (DEBUG 1)) to improve debuggability of interactive
2605 development, and to allow the use of the debug RETURN command in
2607 * an experimental implementation of the RETURN command for the
2608 debugger has been included. (thanks to Frederik Kuivinen)
2609 * fixed bug 62: constraints were not propagated into a loop.
2610 * fixed bug in embedded calls of SORT (reported and investigated by
2612 * fixed some bugs revealed by Paul Dietz' test suite:
2613 ** printing and reading of arrays with some dimensions having
2614 length 0 (thanks to Gerd Moellmann);
2615 ** BOA constructor with &AUX argument without a default value does
2616 not cause a type error;
2617 ** CONSTANTP now returns true for all self-evaluating objects.
2619 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2620 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2621 accessors that are related by inheritance, as specified in the
2622 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2623 Valtteri Vuorikoski)
2624 * The compiler is now able to inline functions that were defined in
2625 a complex lexical environment (e.g. inside a MACROLET).
2626 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2628 * Support for the upcoming FreeBSD-5.0 release has been included.
2629 (thanks to Dag-Erling Smorgrav)
2630 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2631 effect when it is not in a toplevel context.
2632 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2633 stand a better chance of being correct. (thanks to Gerd
2635 * fixed bug in COERCE, which now signals an error on coercing a
2636 rational to a bounded real type which excludes the expected
2638 * The compiler is now able to derive types more accurately from the
2639 COERCE and COMPILE functions.
2640 * fixed bug 223: functional binding is considered to be constant
2641 only for symbols in the CL package.
2642 * fixed bug 231: SETQ did not check the type of a variable being set
2643 (reported by Robert E. Brown)
2644 * A new optimization for MAKE-INSTANCE has been included, fixing
2645 various bugs (including relating to :ALLOCATION :CLASS slots and
2646 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2648 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2649 ** As required by ANSI, LOOP now disallows anonymous collection
2650 clauses such as COLLECT I in conjunction with aggregate boolean
2651 clauses such as THEREIS (= I 1);
2652 ** LOOP now signals an error when any variable is reused in the
2653 same loop (including the potentially useful construct analogous
2654 to WITH A = 1 WITH A = (1+ A);
2655 ** IT is only a special loop symbol within the first clause of a
2656 conditional loop clause;
2657 ** LOOP with a typed iteration variable over a hashtable now
2658 signals a type error iff it should.
2659 * fixed some other bugs revealed by Paul Dietz' test suite:
2660 ** FILE-STREAM now names the class previously known as FD-STREAM;
2661 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2662 argument) no longer signals an error;
2663 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2664 of the slot symbol, rather than using the current package
2665 ((:CONC-NAME "") continues to intern the slot's name in the
2667 * incremented fasl file version number, because of the incompatible
2668 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2669 of the new implementation of DEFINE-COMPILER-MACRO.
2671 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2672 * Support for building SBCL for MIPS platforms running in
2673 little-endian mode has now been checked in, and basic
2674 functionality on said platforms verified.
2675 * minor incompatible change: PCL now records the pathname of a file
2676 in which methods and the like are defined, rather than its
2678 * minor incompatible change: TRUENAME now considers the truename of
2679 a file naming a directory to be the pathname with :DIRECTORY
2680 component indicating that directory.
2681 * minor incompatible change: a NAMED clause in the extended form of
2682 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2683 reason for the previous behaviour is unclear.
2684 * more systematization and improvement of CLOS and MOP conformance
2685 in PCL (thanks to Gerd Moellman and Pierre Mai):
2686 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2688 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2689 primary methods with no specializers;
2690 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2692 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2693 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2694 FORWARD-REFERENCED-CLASSes; error reporting on
2695 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2697 ** SXHASH on CLOS instances now uses a slot internal to the
2698 instance to return different numbers on distinct instances,
2699 while preserving the same return value through invocations of
2701 ** DEFMETHOD signals errors when methods with longer incongruent
2702 lambda lists are added to generic functions;
2703 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2704 CLASS, as specified in AMOP;
2705 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2706 on the order returned by the primary method for classes of
2707 class STANDARD-CLASS;
2708 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2709 * fixed some bugs shown by Paul Dietz' test suite:
2710 ** DOLIST puts its body in TAGBODY;
2711 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2713 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2714 value producing form;
2715 ** if more variables are given to PROGV than values, extra
2716 variables are bound and made to have no value;
2717 ** NSUBSTITUTE on list arguments gets the right answer with
2719 ** ELT signals an error of type TYPE-ERROR when the index argument
2720 is not a valid sequence index;
2721 ** LOOP signals (at macroexpansion time) an error of type
2722 PROGRAM-ERROR when duplicate variable names are found;
2723 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2724 ** FUNCALL of special-operators now cause an error of type
2726 ** PSETQ now works as required in the presence of side-effecting
2727 symbol-macro places;
2728 ** NCONC accepts any object as its last argument;
2729 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2731 ** loop-for-as-package does not require a package to be explicitely
2733 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2735 * fixed bug 166: compiler preserves "there is a way to go"
2736 invariant when deleting code.
2737 * fixed bug 172: macro lambda lists with required arguments after
2738 &REST arguments now cause an error to be signalled. (thanks to
2740 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2741 bug. (thanks to Matthew Danish)
2742 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2744 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2745 arguments are lexically rebound. (thanks to Gerd Moellmann and
2747 * fixed bug 194: error messages are now more informative when there
2748 is no primary method applicable in a call to a generic
2749 function. (thanks to Gerd Moellmann)
2750 * fixed bug in command line argument checking (thanks to Julian
2752 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2753 (thanks to Matthew Danish)
2754 * incremented fasl file version number, because of the
2755 SXHASH-related changes in the layout of CLOS data structures
2757 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2758 * minor incompatible change: The runtime (the Unix executable named
2759 "sbcl") is now much pickier about the .core files it will load.
2760 Essentially it now requires .core files to descend from the same
2761 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2762 as the runtime does. (The intent is to prevent the crashes which
2763 can occur, and which can even be reported as mysterious failures,
2764 when people patch the sources or change the build parameters
2765 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2766 sbcl and .core files.)
2767 * fixed bug: VALUES-LIST is no longer optimized away.
2768 * fixed bug 142: The FFI conversion of C string values to Lisp
2769 string values no longer conses excessively. (thanks to Nathan
2770 Froyd porting Raymond Toy's fix to CMU CL)
2771 * began to systematize and improve MOP conformance in PCL (thanks to
2772 Nathan Froyd, Gerd Moellman and Pierre Mai):
2773 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2775 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2776 ** FINALIZE-INHERITANCE is now called on class finalization;
2777 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2778 argument precedence order.
2779 * fixed bug 202: The compiler no longer fails on functions whose
2780 derived types contradict their declared type.
2781 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2782 so it can be non-toplevel.
2783 * The fasl file version number has changed (because of the new
2784 implementation of DEFMACRO).
2785 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2786 safe code, that any length requirement by their type-specifier
2787 argument is valid. The exceptions to this are described in bug
2789 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2790 functions now check the types of their inputs as required by ANSI.
2791 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2792 introduced symbol is DECLAREd to be SPECIAL.
2793 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2794 * fixed bug: :COUNT argument to sequence functions may be negative
2795 * fixed bug: body of DO-SYMBOLS may contain declarations
2796 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2797 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2798 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2799 symbol macro only once
2800 * fixed printing of call frame when argument list is unavailable
2801 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2802 * compiler no longer signals WARNING on unknown keyword
2805 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2806 * A beta-quality port to the mips architecture running Linux,
2807 based on the old CMUCL backend, has been made. It has been tested
2808 on a big-endian kernel, and works sufficiently well to be able to
2809 rebuild itself; it has not been tested in little-endian mode.
2810 * fixed an inconsistency between gencgc.c and purify.c which made
2811 dumping/loading .core files unreliable
2812 * fixed bug 120a: The compiler now deals correctly with IFs where
2813 the consequent is the same as the alternative, instead of
2814 misderiving the return type. (thanks to Alexey Dejneka)
2815 * fixed bug 113: Logical pathnames are now dumpable (the logical
2816 host is resolved at load-time, throwing an error if it is not
2818 * fixed bug 174: FORMAT's error message is slightly clearer when a
2819 non-printing character is used in a format directive.
2820 * fixed several bugs in compiler checking of type declarations, i.e.
2821 violations of the Python "declarations are assertions" principle
2822 (thanks to Alexey Dejneka)
2823 * fixed several bugs in PCL's error checking (thanks to Gerd
2825 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2827 * fixed bug in compilation of functions as first class values
2828 (thanks to Antonio Martinez-Shotton)
2829 * The compiler's handling TYPE-ERRORs which it can prove will
2830 inevitably happen at runtime has been cleaned up and corrected
2831 in several ways. (thanks to Alexey Dejneka)
2832 * improved argument type checking for various basic arithmetic
2833 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2834 lavished on them in the past that they can be compiled in many
2835 ways in different special cases
2836 * fixed bug 181: compiler checks validity of user supplied type
2838 * cleaned up code flushing in optimization: Function calls which
2839 should signal errors for safety purposes (e.g. which ANSI says
2840 should signal errors when their arguments are of incorrect type)
2841 are no longer optimized away.
2842 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2843 * incremented fasl file version number, because changes in the
2844 implementation of sequence functions like COERCE caused
2845 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2846 in old inline expansions) to become undefined. (Actually these
2847 changes were later undone, so we might very well be binary
2848 compatible with 0.7.7 after all, but leaving the version number
2849 incremented seemed like the simplest and most conservative
2852 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2853 * An alpha-quality port to the parisc architecture running Linux,
2854 based on the old CMUCL backend, has been made. This, even more so
2855 than the other backends, should be considered still a work in
2856 progress; known problems include that the Linux kernel in 64-bit
2857 mode does not propagate the correct sigcontext structure to
2858 userspace, and consequently SBCL on a parisc64 kernel will not
2860 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2861 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2862 suppressed.) Also now that the compiler is looking at declarations
2863 in the environment, it checks optimization declarations as well,
2864 and suppresses inlining when (> DEBUG SPEED).
2865 * More fixes have been made to treatment of floating point exception
2866 treatment and other Unix signals. In particular, floating point
2867 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2868 * The detection and handling of control stack exhaustion (infinite
2869 or very deeply nested recursion) has changed. Stack exhaustion
2870 detection is now done by write-protecting pages at the OS level
2871 and applies at all optimization settings; when found, a
2872 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2873 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2874 be caught using IGNORE-ERRORS.
2875 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2876 correctly when transforming recursive LABELS functions to LETs.
2877 (thanks to Alexey Dejneka)
2878 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2879 that are names of constants or global variables.
2880 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2881 alien routines with docstrings.
2882 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2883 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2885 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2886 (thanks to Eric Marsden)
2887 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2888 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2889 * Bug fix: The SPARC backend can now compile functions involving
2890 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2891 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2892 to a functions expecting a PATHNAME-DESIGNATOR.
2893 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2894 lambda lists. (thanks to Alexey Dejneka)
2895 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2896 functions so that EXPORTing the name of the class doesn't cause
2897 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2898 to Antonio Martinez for reporting this)
2899 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2900 constant-folded without failing an assertion. (thanks to Einar
2901 Floystad Dorum for reporting this)
2902 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2903 types are now dealt with more correctly by the compiler.
2904 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2905 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2906 behaviour analogous to RENAME-FILE. This puts its behaviour more
2907 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2909 * The fasl file version number has changed again. (because of the
2910 bug fix involving the names of PCL MAKE-INSTANCE functions)
2912 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2913 * bug fix: Floating point exceptions are treated much more
2914 consistently on the x86/Linux and PPC/Linux platforms.
2915 * Array initialization with :INITIAL-ELEMENT is now much faster for
2916 cases when the compiler cannot open code the array creation, but
2917 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2918 array accesses have also seen a speed increase.
2919 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2920 specified. (This was at the root of some bad interactions between
2921 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2923 * bug fix: Internal error arguments for undefined functions are now
2924 computed correctly on the PPC/Linux platform.
2925 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2926 Raymond Toy's patch for CMU CL)
2927 * Support for the Solaris 9 operating environment has been included
2928 (thanks to Daniel Merritt)
2929 * A very ugly but hopefully complete draft of the missing FFI chapter
2930 of the manual has been created by reformatting the corresponding
2931 CMU CL manual chapter into (currently very ugly and incoherent)
2932 DocBook and bringing it up to date for SBCL behavior. Thus, the
2933 manual is now essentially complete, at least by my extreme
2934 once-and-only-once standards, whereby it's acceptable to refer to
2935 the doc strings of SB-EXT functions as the primary documentation.
2936 * The fasl file version number has changed again, due to cleanup of
2937 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2938 is no longer a static symbol.)
2940 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2941 * SBCL now builds with OpenMCL (version 0.12) as the
2942 cross-compilation host; also, more progress has been made toward
2943 bootstrapping under CLISP.
2944 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2946 * bug 158 fixed: The compiler can now deal with integer loop
2947 increments different from 1; fixing this turned out also to fix
2949 * bug 169 fixed: no more bogus warnings about using lexical bindings
2950 despite the presence of perfectly good SPECIAL declarations (thanks
2951 to David Lichteblau)
2952 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2953 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2954 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2956 * bug 180 fixed: Method combination specifications no longer ignore
2957 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2958 * bug fix: Structure type predicate functions now check their argument
2959 count as they should.
2960 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2961 correctly. (thanks to Pierre Mai)
2962 * minor incompatible change: The --noprogrammer option is deprecated
2963 in favor of the new --disable-debugger option, which is very similar.
2964 (The major difference is that it takes effect at a slightly different
2965 time at startup, causing handling of errors in --sysinit and
2966 --userinit files will be affected differently.) The
2967 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2968 been added to allow this functionality to be controlled from ordinary
2969 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2970 maintainers, who might want to run non-interactive scripts to
2971 build SBCL cores which will later be used interactively.)
2972 * minor incompatible change: The LOAD function no longer, when given
2973 a wild pathname to load, loads all files matching that pathname.
2974 Instead, an error of type FILE-ERROR is signalled.
2976 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2977 * bug 147 fixed: The compiler preserves its block link/count
2978 invariants more correctly now so that it doesn't crash. (thanks
2980 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2982 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2983 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2984 no-value-for-this-character as NIL (as specified by ANSI).
2985 * HOST-NAMESTRING on physical pathnames now returns a string that is
2986 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2987 (thanks to Christophe Rhodes)
2988 * The Alpha port handles icache flushing more correctly. (thanks to
2990 * More progress has been made toward bootstrapping under CLISP. (thanks
2991 to Christophe Rhodes)
2992 * The fasl file format has changed again, because dynamic loading
2993 on OpenBSD (which has non-ELF object files) motivated some cleanups
2994 in the way that foreign symbols are transformed and passed around.
2995 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2996 is no longer treated as whitespace by the reader, but instead as
2997 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2998 |A
\7fB|, instead of A as it used to.
3000 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3001 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3002 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3003 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3004 this as of 0.7.1.45, but was left out of the previous news section
3005 (thanks to Dan Barlow)
3006 * SBCL now runs on the Solaris operating system on SPARC architectures
3007 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3008 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3009 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3010 * SPARC backend cleanups, allowing builds of cores optimized for V8
3011 and V9 SPARCS, and also emission of code targeted to a particular
3012 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3014 * SBCL is closer to bootstrapping under CLISP, thanks to various
3015 fixes by Christophe Rhodes.
3016 * The fasl file format has changed again, to allow the compiler's
3017 INFO database to support symbol macros.
3018 * The user manual (in doc/) is formatted into HTML more nicely.
3019 (thanks to coreythomas)
3020 * The system is smarter about SUBTYPEP relationships, especially
3021 those involving NOT types (including types such as ATOM which are
3022 represented internally using NOT types). Thus SUBTYPEP is less
3023 likely to return (VALUES NIL NIL) in general, and in particular
3024 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3026 * The fasl file format has changed again, because the internal
3027 representation of types now includes a new slot to support the new
3028 SUBTYPEP-of-NOT-types logic.
3029 * (not a change in the main branch of SBCL, but a related prototype
3030 which can hopefully be merged into the main branch of SBCL in the
3031 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3032 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3033 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3034 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3035 dispatch macro character combinations. (thanks to Alexey Dejneka)
3036 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3037 that superficially look like logical namestrings correctly.
3038 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3040 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3041 default supplied-p) for &optional and &key arguments. (thanks to
3044 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3045 * incompatible change: The compiler is now less aggressive about
3046 tail call optimization, doing it only when (> SPACE DEBUG) or
3047 (> SPEED DEBUG). (This is an incompatible change because there are
3048 programs which relied on the old CMU-CL-style behavior to optimize
3049 away their unbounded recursion which will now die of stack overflow.)
3050 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3051 for non-GENCGC systems has been increased to 20M (since that
3052 seems much closer to the likely performance optimum for modern
3053 systems than the old 4M value was)
3054 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3055 and *DEBUG-PRINT-LEVEL*
3056 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3057 of CMU CL's support for SPARC, and various endianness and other
3058 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3059 * new syntactic sugar for the Unix command line: --load foo.bar is now
3060 an alternate notation for --eval '(load "foo.bar")'.
3062 ** The system now detects stack overflow and handles it gracefully,
3063 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3064 optimization settings. (This is a good thing in general, and
3065 its introduction in this version should be particularly timely
3066 for anyone whose code fails because of suppression of tail
3068 ** The system now hunts for the C variable "environ" in a more
3069 devious way, to avoid segfaults when the C library version differs
3070 between compile time and run time. (thanks to Christophe Rhodes)
3071 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3072 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3073 ** The compiler no longer issues bogus style warnings for undefined
3074 classes in the same source file as the DEFCLASSes which defined
3075 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3077 ** fixes in CONDITION class precedence list for undefined function
3078 errors (thanks to Alexei Dejneka)
3079 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3080 correctly. (thanks to Dan Barlow)
3081 ** portability fixes aiming at bootstrapping under CLISP (thanks
3082 to Dave McDonald and Christophe Rhodes)
3083 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3084 ** fixes in type translation and and type inference (thanks to
3086 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3087 ** various fixes in the new ports (thanks to Dan Barlow)
3088 * several changes related to debugging:
3089 ** suppression of tail recursion, as noted above
3090 ** stack overflow detection, as noted above
3091 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3092 is now the default. (For some time encapsulation has been more
3093 reliable than the breakpoint-based :ENCAPSULATE NIL
3094 implementation, at least on X86 systems; and I just noticed that
3095 encapsulation also seems closer to the spirit of the ANSI
3098 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3100 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3101 up properly again. (There was a packaging bug in 0.7.0 which
3102 left their definitions in SB-SYS::LOAD-FOREIGN and
3103 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3104 vital for most things which interface to C-level interfaces,
3105 like extensions working with sockets or databases or
3106 Perl-compatible regexes or whatever, and the need to fix
3107 this bug is the main reason that 0.7.1 was released so
3109 ** DEFGENERIC is now choosier about the methods it redefines, so that
3110 reLOADing a previously-LOADed file containing DEFGENERICs does
3111 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3112 cycle now works as it should. (thanks to Alexey Dejneka)
3113 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3114 fixed by Christophe Rhodes. (He actually submitted this patch
3115 months ago, and I delayed until after 0.7.0.)
3116 ** Bug 111 (internal compiler confusion about runtime checks on
3117 FUNCTION types) was fixed by Alexey Dejneka.
3118 * Some internal cleanups (getting rid of variables which aren't
3119 needed now that the byte interpreter is gone) caused the fasl
3120 file format number to change again.
3122 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3123 * major incompatible change: The default fasl file extension, i.e. the
3124 default extension for files produced by COMPILE-FILE, has changed
3125 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3127 ** There are many changes in the implementation of the compiler.
3128 SBCL is now essentially a compiler-only implementation of ANSI
3129 Common Lisp. EVAL still "interprets" a few special cases, but
3130 almost all the interesting cases are handled by creating
3131 a LAMBDA expression, calling COMPILE on it, then calling
3132 FUNCALL on the result.
3133 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3134 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3135 Since the code is newer, there might still be some new bugs
3136 (though not as many as before Martin Atzmueller's fixes:-). But
3137 the new code is substantially simpler and clearer, and hopefully
3138 any remaining bugs will be simpler, less fundamental, and more
3139 fixable then the bugs in the old code.
3140 ** The revised compiler is still a little unsteady on its feet.
3142 *** The debugging information it produces (particularly the names
3143 of FUNCTION objects) is sometimes much less useful than what
3144 the old compiler produced.
3145 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3146 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3147 within a MACROLET) has been temporarily weakened.
3148 ** There are new compiler optimizations for various functions:
3149 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3150 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3151 *** the math functions TRUNCATE, FLOOR, and CEILING
3152 *** the function-of-all-trades COERCE
3153 Mostly these should be transparent, but there's one
3154 potentially-annoying problem (bug 117): when the compiler
3155 inline-expands a function and does type analysis on the result,
3156 it can create control paths which have type mismatches, and
3157 when it can't prove that those control paths aren't taken,
3158 it will issue WARNINGs about the type mismatches. This is
3159 a particular problem in practice for the new sequence functions.
3160 It's not clear how this should be fixed, and for now, a
3161 workaround is given in the entry for 117 in the BUGS file.
3162 ** (Because of the interaction between the two previous items --
3163 occasional inlining problems and new inline expansions -- some
3164 of the new sequence function optimizations won't really kick in
3165 completely until debugging information, and then inlining, are
3166 straightened out in some future version.)
3167 * minor incompatible changes:
3168 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3169 DIRECTORY no longer implicitly promotes NIL slots of its
3170 pathname argument to :WILD. In particular, when you ask for the
3171 contents of a directory (which you used to be able to do without
3172 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3173 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3174 ** changes in behavior that ANSI explicitly defines to be
3175 implementation dependent:
3176 *** The new compiler-only implementation still conforms with ANSI,
3177 but acts a little different than before. Besides the obvious
3178 changes in performance tradeoffs (that the cost per form passed
3179 to EVAL has gone up, and the cost per form executed by EVAL
3180 has gone down), the behavior of the system changes a little
3181 because there are no longer any interpreted function objects.
3182 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3183 e.g. doing COMPILE on the output of interactive DEFUN is
3185 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3187 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3188 has changed from (:CL) to NIL.
3189 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3190 e.g. #\Newline and #\Tab, don't have names specified in the
3191 ANSI Common Lisp standard, is now based on their ASCII symbolic
3192 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3193 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3194 are no longer used for output.
3195 ** changes in internal implementation constants:
3196 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3197 4 million. (If your application spends a lot of time GCing and
3198 you have a lot of RAM, you might want to experiment with
3199 increasing it even more.)
3200 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3201 However, almost all old code should still continue to work without
3202 immediate update, as SB-C-CALL is now a (deprecated) nickname
3204 ** Old operator names in the style DEF-FOO are now deprecated in
3205 favor of new corresponding names DEFINE-FOO, for consistency with
3206 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3207 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3208 internal symbols, but a few supported extensions like
3209 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3210 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3211 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3212 etc. as you get deeper into recursive calls to the debugger
3213 command loop, instead of the old "5]", "5]]", "5]]]"
3214 sequence. (I was motivated to do this when squabbles between
3215 ILISP and SBCL left me very deeply nested in the debugger. In the
3216 short term, this change will probably provoke more ILISP/SBCL
3217 squabbles, but hopefully it will be an improvement in the long run.)
3218 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3219 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3220 function, since declaiming return types involving aliens is
3221 (1) annoyingly messy to do by hand and (2) vital to efficient
3222 compilation of code which calls such functions.
3223 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3224 longer reexported by the SB-EXT package. They're solely useful
3225 for alien code, so it seems more logical that you should get
3226 them from the SB-ALIEN package, not in SB-EXT.
3227 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3228 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3229 features. Instead, the code that they used to control is always
3230 built into the system.
3231 * many other bug fixes
3232 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3233 advantage of the new EVAL-WHEN stuff and to clean them up in
3234 general, and they are now more ANSI-compliant in a number of
3235 ways. Martin Atzmueller is responsible for a lot of this.
3236 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3238 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3239 *** correct ERROR type for various file operations
3240 *** some fixes for Lisp streams
3241 *** DEFMETHOD syntax checking
3242 *** changing old weird representation of debug information as
3243 strings (which, among their other deficiencies, don't transform
3244 correctly when you rename packages, and don't change their
3245 print representation when you change things like *PACKAGE*
3246 and *PRINT-LENGTH*) to symbols and lists of symbols
3247 He also made several improvements and fixed several bugs in DESCRIBE.
3248 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3250 *** misbehavior of WRITE-STRING/WRITE-LINE
3251 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3252 and several other LOOP problems as well
3253 *** DIRECTORY when similar filenames are present
3254 *** DEFGENERIC with :METHOD options
3255 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3256 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3257 *** argument ordering in FIND with :TEST option
3258 *** mishandled package designator argument in APROPOS-LIST
3259 *** various problems in the backquote readmacro
3260 *** a bug in APROPOS
3261 *** probably some others that I'm not describing very well here,
3262 since the CVS log documents them by reference to sbcl-devel
3263 messages, and the SourceForge archives aren't working well.:-(
3264 ** Dan Barlow improved the Alpha port (and is making progress on the
3265 PPC port, for those of you who think different).
3266 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3267 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3268 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3269 continues to work on the SPARC port (for those of you in a position
3270 to look down upon our little PC-compatible boxes from a great height).
3271 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3272 and uses that copy, rather than the current dynamic value, when
3273 it's trying to decide whether to truncate output. Thus e.g.
3274 (let ((*print-lines* 50))
3275 (pprint-logical-block (stream nil)
3277 (let ((*print-lines* 8))
3278 (print (aref possiblybigthings i) stream)))))
3279 should now truncate the logical block only at 50 lines, instead of
3280 often truncating it at 8 lines, as it did before.
3281 * The doc/cmucl/ directory, containing old CMU CL documentation
3282 from the time of the fork, is no longer part of the base system.
3283 SourceForge has shut down its anonymous FTP service, and with it
3284 my original plan for distributing the old CMU CL documentation
3285 there. For now, if you need these files you can download an old
3286 SBCL source release and extract them from it.
3287 * The fasl file version number changed again, for dozens of reasons,
3288 some of which are apparent above.
3290 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3291 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3292 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3293 patch, so that DISASSEMBLE output is much nicer.
3294 * The code in the SB-PROFILE package now seems reasonably stable.
3295 I still haven't decided what the final interface should look like
3296 (I'd like PROFILE to interact cleanly with TRACE, since both
3297 facilities use function encapsulation) but if you have a need
3298 for profiling now, you can probably use it successfully with
3299 the current CMU-CL-style interface.
3300 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3301 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3302 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3303 link now returns the dangling link itself, and for similar
3304 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3305 link itself. (In these cases the old code signalled an error and
3306 looped endlessly, respectively.) Thus, DIRECTORY now works even
3307 in the presence of dangling and cyclic symbolic links.
3308 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3309 is now a supported extension again, since the consensus on
3310 sbcl-devel was that it can be useful for ordinary development
3311 work, not just for debugging SBCL itself.
3312 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3313 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3314 that currently-defined functions might be redefined later with
3315 different return types.
3316 * Hash tables can be printed readably, as inspired by CMU CL code
3317 of Eric Marsden and SBCL code of Martin Atzmueller.
3318 * better error handling in CLOS method combination, thanks to
3319 Martin Atzmueller porting Pierre Mai's CMU CL patches
3320 * more overflow fixes for >16Mbyte I/O buffers
3321 * A bug in READ has been fixed, so that now a single Ctrl-D
3322 character suffices to cause end-of-file on character streams.
3323 In particular, now you only need one Ctrl-D at the command
3324 line (not two) to exit SBCL.
3325 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3327 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3328 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3329 does the right thing.
3330 * The implementation of some type tests, especially for CONDITION
3331 types, is now tidier and maybe faster, due to CMU CL code
3332 originally by Douglas Crosher, ported by Martin Atzmueller.
3333 * Some math functions have been fixed, and there are new
3334 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3335 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3336 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3337 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3338 likely to be built into sbcl-0.7.0.)
3339 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3340 CL, ported to SBCL by Martin Atzmueller)
3341 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3342 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3343 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3344 * There's a new slam.sh hack to shorten the edit/compile/debug
3345 cycle for low-level changes to SBCL itself, and a new
3346 :SB-AFTER-XC-CORE target feature to control the generation of
3347 the after-xc.core file needed by slam.sh.
3348 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3349 COMPILE-FILE is no longer supported, so that now every function
3350 gets an entry point, so that block compilation looks a little
3351 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3352 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3353 returns the number of bytes consed since the system started,
3354 rather than the number consed since the first time the function
3355 was called. (The new definition parallels ANSI functions like
3356 CL:GET-INTERNAL-RUN-TIME.)
3357 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3358 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3359 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3360 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3361 * Fasl file version numbers are now independent of the target CPU,
3362 since historically most system changes which required version
3363 number changes have affected all CPUs equally. Similarly,
3364 the byte fasl file version is now equal to the ordinary
3367 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3368 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3369 is no longer recognized. I apologize for this, because it was
3370 listed in SB-EXT as a supported extension, but I found that
3371 its existing behavior was poorly specified, as well as incorrectly
3372 specified, and it looked like too much of a mess to straighten it
3373 out. I have enough on my hands trying to get ANSI stuff to work..
3374 * many patches ported from CMU CL by Martin Atzmueller, with
3375 half a dozen bug fixes in pretty-printing and the debugger, and
3376 half a dozen others elsewhere
3377 * fixed bug 13: Floating point infinities are now supported again.
3378 They might still be a little bit flaky, but thanks to bug reports
3379 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3380 as flaky as they were.
3381 * The --noprogrammer command line option is now supported. (Its
3382 behavior is slightly different in detail from what the old man
3383 page claimed it would do, but it's still appropriate under the
3384 same circumstances that the man page talks about.)
3385 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3386 are now supported, and enabled by default. Thus, the compiler can
3387 handle many floating point and complex operations much less
3388 inefficiently. (Thus e.g. you can implement a complex FFT
3390 * The compiler now detects type mismatches between DECLAIM FTYPE
3391 and DEFUN better, and implements CHECK-TYPE more correctly, and
3392 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3393 thanks to patches from Martin Atzmueller.
3394 * various fixes to make the cross-compiler more portable to
3395 ANSI-conforming-but-different cross-compilation hosts (notably
3396 Lispworks for Windows, following bug reports from Arthur Lemmens)
3397 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3398 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3399 fixed thanks to Pierre Mai's CMU CL patches.
3400 * fixes to keep the system from overflowing internal counters when
3401 it tries to use i/o buffers larger than 16M bytes
3402 * fixed bug 45a: Various internal functions required to support
3403 complex special functions have been merged from CMU CL sources.
3404 (When I was first setting up SBCL, I misunderstood a compile-time
3405 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3406 * improved support for type intersection and union, fixing bug 12
3407 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3408 more obscure bugs as well
3409 * some steps toward byte-compiling non-performance-critical
3410 parts of the system, courtesy of patches from Martin Atzmueller
3411 * Christophe Rhodes has made some debian packages of sbcl at
3412 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3413 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3414 stable, but are nonetheless usable. When he's ready, I'd be happy
3415 to add them to the SourceForge "File Releases" section. (And if
3416 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3417 * new fasl file format version number (because of changes in
3418 internal representation of (OR ..) types to accommodate the new
3419 support for (AND ..) types, among other things)
3421 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3422 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3424 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3426 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3427 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3428 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3429 * DESCRIBE now works on structure objects again.
3430 * Most function call argument type mismatches are now handled as
3431 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3432 know whether the function will be redefined before the call is
3433 executed. (The compiler could flag local calls with full WARNINGs,
3434 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3435 it doesn't keep track of enough information to know whether calls
3436 are local in this sense.)
3437 * Compiler output is now more verbose, with messages truncated
3438 later than before. (There should be some supported way for users
3439 to override the default verbosity, but I haven't decided how to
3440 provide it yet, so this behavior is still controlled by the internal
3441 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3442 src/compiler/ir1util.lisp.)
3443 * Fasl file format version numbers have increased again, because
3444 support for the Gray streams extension changes the layout of the
3445 system's STREAM objects.
3446 * The Gray subclassable streams extension now works, thanks to a
3447 patch from Martin Atzmueller.
3448 * The full LOAD-FOREIGN extension (not just the primitive
3449 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3450 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3451 but like most other programs, it defaults to copying the Unix
3452 environment from the original process instead of starting the
3453 new process in an empty environment.
3454 * Extensions which manipulate the Unix environment now support
3455 an :ENVIRONMENT keyword option which doesn't smash case or
3456 do other bad things. The CMU-CL-style :ENV option is retained
3457 for porting convenience.
3458 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3459 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3461 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3463 * A patch from Martin Atzmueller seems to have solved the SIGINT
3464 problem, and as far as we know, signal-handling now works cleanly.
3465 (If you find any new bugs, please report them!)
3466 * The system no longer defaults Lisp source file names to types
3467 ".l", ".cl", or ".lsp", but only to ".lisp".
3468 * The compiler no longer uses special default file extensions for
3469 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3470 seems to expect a single default extension for all compiled code,
3471 and there's no compelling reason to try to stretch the standard
3472 to allow two different extensions.) Instead, byte-compiled files
3473 default to the same extension as native-compiled files.
3474 * Fasl file format version numbers have increased again, because
3475 a rearrangement of internal implementation packages made some
3476 dumped symbols in old fasl files unreadable in new cores.
3477 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3478 many fewer weird special cases.
3479 * Bug #17 (differing COMPILE-FILE behavior between logical and
3480 physical pathnames) has been fixed, and some related misbehavior too,
3481 thanks to a patch from Martin Atzmueller.
3482 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3483 by Tim Moore, ported to SBCL by Martin Atzmueller.
3484 * Martin Atzmueller fixed several filesystem-related problems,
3485 including bug #36, in part by porting CMU CL patches, which were
3486 written in part by Paul Werkowski.
3487 * More compiler warnings in src/runtime/ are gone, thanks to
3488 more patches from Martin Atzmueller.
3489 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3492 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3494 * DESCRIBE now works on CONDITION objects.
3495 * The debugger now handles errors which arise when trying to print
3496 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3498 * The build system now uses an additional file, customize-target-features.lisp,
3499 to allow local modifications to the target *FEATURES* list. (The point of
3500 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3501 debugging features enabled, without having to worry about propagating it
3502 into everyone's system when I do a "cvs update".) When no
3503 customize-target-features.lisp file exists, the target *FEATURES* list
3504 should be constructed the same way as before.
3505 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3506 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3507 warn about behavior which is undefined under the ANSI spec. Note: This
3508 is specified by ANSI, but it's not very popular with programmers.
3509 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3510 macro in the SBCL sources for an example of a workaround which you
3511 might use to make portable ANSI-standard code which does what you want.
3512 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3513 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3514 when it's used as a non-toplevel form. (This is required in order
3515 to implement the DEFCONSTANT-EQX macro.)
3516 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3517 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3518 funny things, due to bugs in the implementation of EVAL-WHEN
3519 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3520 non-toplevel DEFCONSTANTs are uncommon.)
3521 * The core file version number and fasl file version number have been
3522 incremented, because the old noncompliant DEFCONSTANT behavior involved
3523 calling functions which no longer exist, and because I also took the
3524 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3525 * fixed bug 1 (error handling before read-eval-print loop starts), and
3526 redid debugger restarts and related debugger commands somewhat while
3528 ** The QUIT debugger command is gone, since it did something
3529 rather different than the SB-EXT:QUIT command, and since it never
3530 worked properly outside the main toplevel read/eval/print loop.
3531 Invoking the new TOPLEVEL restart provides the same functionality.
3532 ** The GO debugger command is also gone, since you can just invoke
3533 the CONTINUE restart directly instead.
3534 ** The TOP debugger command is also gone, since it's redundant with the
3535 FRAME 0 command, and since it interfered with abbreviations for the
3537 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3539 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3540 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3541 some porting work by Martin Atzmueller)
3542 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3543 with a boatload of patches which he ported to SBCL. Now that those
3545 ** The system tries to make sure that its low-priority messages
3546 are prefixed by semicolons, to help people who like to use
3547 syntax highlighting in their ILISP buffer. (This patch
3548 was originally due to Raymond Toy.)
3549 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3550 patches originally written by Raymond Toy.
3551 ** The compiler understands coercion between single-value and
3552 multiple-VALUES type expressions better, getting rid of some very
3553 weird behavior, thanks to patches originally by Robert MacLachlan
3554 and Douglas Crosher.
3555 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3556 lambda lists, thanks to a patch originally by Pierre Mai.
3557 ** The system no longer bogusly warns about "abbreviated type
3559 ** The compiler gets less confused by inlining and RETURN-FROM,
3560 thanks to some patches originally by Tim Moore.
3561 ** The system no longer hangs when dumping circular lists to fasl
3562 files, thanks to a patch originally from Douglas Crosher.
3563 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3564 undefined function error.
3565 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3566 in structure slot initforms) for now, documented workaround instead:-|
3567 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3568 * fixed bug 5: added stubs for various Gray stream functions called
3569 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3570 installed, at least appropriate type errors are generated
3571 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3572 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3573 consistently in DEFMETHOD forms.
3574 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3575 it doesn't seem to affect SBCL after all
3576 * The C runtime system now builds with better optimization and many
3577 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3579 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3581 * The system is now under CVS at SourceForge (instead of the
3582 CVS repository on my home machine).
3583 * The new signal handling code has been tweaked to treat register
3584 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3585 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3586 downstream, e.g. in debug-int.lisp, has implicit dependencies
3587 on the unsignedness of integer representation of machine words,
3588 and that caused the system to bomb out with infinite regress
3589 when trying to recover from type errors involving signed values,
3590 e.g. (BUTLAST '(1 2 3) -1).)
3591 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3592 (This was one of the bugs Peter Van Eynde reported back in July.)
3593 * The system now uses code inspired by Colin Walters' O(N)
3594 implementation of MAP (from the cmucl-imp@cons.org mailing
3595 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3596 inline the MAP operation, and there is more than one
3597 sequence argument to the MAP call (so that it can't just
3598 do ETYPECASE once and for all based on the type of the
3599 single sequence argument). (The old non-inline implementation
3600 of the general M-argument sequence-of-length-N case required
3601 O(M*N*N) time when any of the sequence arguments were LISTs.)
3602 * The QUIT :UNIX-CODE keyword argument has been renamed to
3603 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3604 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3606 * Raymond Wiker's patches to port dynamic loading from Linux to
3607 FreeBSD have been added.
3608 * The BUGS file is now more nearly up to date, thanks in large part
3609 to Martin Atzmueller's review of it.
3610 * The debugger now flushes standard output streams before it begins
3611 its output ("debugger invoked" and so forth).
3612 * The core version number and fasl file version number have both
3613 been incremented, because of incompatible changes in the layout
3615 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3616 used to not be called for a saved Lisp image.)
3617 * Martin Atzmueller's version of a patch to fix a compiler crash,
3618 as posted on sbcl-devel 13 September 2000, has been installed.
3619 * Instead of installing Martin Atzmueller's patch for the
3620 compiler transform for SUBSEQ, I deleted the compiler transform,
3621 and transforms for some similar consing operations.
3622 * A bug in signal handling which kept TRACE from working on OpenBSD
3624 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3625 away by constant folding
3626 * The system now defines its address space constants in one place
3627 (in the Lisp sources), and propagates them automatically elsewhere
3628 (through GENESIS and the sbcl.h file). Therefore, patching the
3629 address map is less unnecessarily tedious and error-prone. The
3630 Lisp names of address space constants have also been systematized.
3631 * CVS tags like dollar-Header-dollar have been removed from
3632 the sources, because they have never saved me trouble and
3633 they've been source of trouble working with patches and other
3634 diff-related operations.
3635 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3636 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3638 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3640 * The system has been ported to OpenBSD.
3641 * The system now compiles with a simple "sh make.sh" on the systems
3642 that it's supported on. I.e., now you no longer need to tweak
3643 text in the target-features.lisp-expr and symlinks in src/runtime/
3644 by hand, the make.sh takes care of it for you.
3645 * The system is no longer so grossly inefficient when compiling code
3646 involving vectors implemented as general (not simple) vectors (VECTOR T),
3647 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3648 which dares to use the various sequence functions on non-simple
3649 vectors, takes less of a performance hit.
3650 * There is now a primitive type predicate VECTOR-T-P
3651 to test for the (VECTOR T) type, so that e.g.
3652 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3653 can now be compiled with some semblance of efficiency. (The old code
3654 turned the type declaration into a full call to %TYPEP at runtime!)
3655 * AREF on (VECTOR T) is still not fast, since it's still compiled
3656 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3657 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3658 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3659 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3660 without making full calls to SUBTYPEP at runtime.
3661 (Some analogous efficiency issues for non-simple vectors specialized to
3662 element types other than T, or for non-simple multidimensional arrays,
3663 have not been addressed. They could almost certainly be handled the
3664 same way if anyone is motivated to do so.)
3665 * The changes in array handling break binary compatibility, so
3666 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3667 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3668 instead of (VALUES T T).
3669 * By following the instructions that Dan Barlow posted to sbcl-devel
3670 on 2 July 2000, I was able to enable primitive dynamic object
3671 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3672 functionality is not implemented (since it calls ld to resolve
3673 library references automatically, requiring RUN-PROGRAM for its
3674 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3675 try to resolve library references) is now supported.
3676 * The system now flushes the standard output streams when it terminates,
3677 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3678 them at several other probably-convenient times, e.g. in each pass of
3679 the toplevel read-eval-print loop, and after evaluating a form given
3680 as an "--eval" command-line option. (These changes were motivated by a
3681 discussion of stream flushing issues on cmucl-imp in August 2000.)
3682 * The source transform for TYPEP of array types no longer assumes
3683 that an array whose element type is a not-yet-defined type
3684 is implemented as an array of T, but instead punts, so that the
3685 type will be interpreted at runtime.
3686 * There is now some support for cross-compiling in make.sh: each of
3687 the phases of make.sh has its own script. (This should be transparent
3688 to people doing ordinary, non-cross-compile builds.)
3689 * Since my laptop doesn't have hundreds of megabytes of memory like
3690 my desktop machine, I became more motivated to do some items on
3691 my to-do list in order to reduce the size of the system a little:
3692 ** Arrange for various needed-only-at-cold-init things to be
3693 uninterned after cold init. To support this, those things have
3694 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3695 symbols with such names are now uninterned after cold init).
3696 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3697 fasl files for cold load.
3698 ** Remove the old compiler structure pooling code (which used to
3699 be conditional on the target feature :SB-ALLOC) completely.
3700 ** Redo the representation of some data in cold init to be more compact.
3701 (I also looked into supporting byte compiled code at bootstrap time,
3702 which would probably reduce the size of the system a lot, but that
3703 looked too complicated, so I punted for now.)
3704 * The maximum signal nesting depth in the src/runtime/ support code has
3705 been reduced from 4096 to 256. (I don't know any reason for the very
3706 large old value. If the new smaller value turns out to break something,
3707 I'll probably just bump it back up.)
3708 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3710 * Many, many bugs reported by Peter Van Eynde have been added to
3711 the BUGS list; some have even been fixed.
3712 * While enabling dynamic object file loading, I tried to make the
3713 code easier to understand, renaming various functions and variables
3714 with less ambiguous names, and changing some function calling
3715 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3716 * While trying to figure out how to do the OpenBSD port, I tried to
3717 clean up some of the code in src/runtime/. In particular, I dropped
3718 support for non-POSIX signal handling, added various comments,
3719 tweaked the code to reduce the number of compilation warnings, and
3720 renamed some files to increase consistency.
3721 * To support the new automatic configuration functionality in make.sh,
3722 the source file target-features.lisp-expr has been replaced with the
3723 source file base-target-features.lisp-expr and the machine-generated
3724 file local-target-features.lisp-expr.
3725 * fixed a stupid quoting error in make.sh so that using CMU CL
3726 "lisp -batch" as cross-compilation host works again
3728 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3730 * DESCRIBE no longer tries to call itself recursively to describe
3731 bound/fbound values, so that it no longer fails on symbols which are
3732 bound to themselves (like keywords, T, and NIL).
3733 * DESCRIBE now works on generic functions.
3734 * The printer now prints less-screwed-up representations of closures
3735 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3736 * A private symbol is used instead of the :EMPTY keyword previously
3737 used to mark empty slots in hash tables. Thus
3738 (DEFVAR *HT* (MAKE-HASH-TABLE))
3739 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3740 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3741 now does what ANSI says that it should. (You can still get
3742 similar noncompliant behavior if bang on the hash table
3743 implementation with all the symbols you get back from
3744 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3745 This breaks binary compatibility, since tests for equality to
3746 :EMPTY are wired into things like the macroexpansion of
3747 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3749 * There's now a minimal placeholder implementation for CL:STEP,
3750 as required by ANSI.
3751 * An obscure bug in the interaction of the normal compiler, the byte
3752 compiler, inlining, and structure predicates has been patched
3753 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3754 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3756 * Missing ordinary arguments in a macro call are now detected even
3757 when the macro lambda list contains &KEY or &REST.
3758 * The debugger no longer complains about encountering the top of the
3759 stack when you type "FRAME 0" to explicitly instruct it to go to
3760 the top of the stack. And it now prints the frame you request even
3761 if it's the current frame (instead of saying "You are here.").
3762 * As specified by ANSI, the system now always prints keywords
3763 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3765 * The default initial SIZE of HASH-TABLEs is now smaller.
3766 * Type information from CLOS class dispatch is now propagated
3767 into DEFMETHOD bodies, so that e.g.
3768 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3770 is now basically equivalent to
3771 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3772 (DECLARE (TYPE SINGLE-FLOAT X))
3774 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3775 operation, without having to do run-time type dispatch.
3776 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3777 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3778 * Since I don't seem to be making big file reorganizations very often
3779 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3780 is overflowing my ability to conveniently back them up), I've finally
3781 checked the system into CVS. (The CVS repository is on my home system,
3782 not at SourceForge -- putting it on SourceForge might come later.)
3783 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3784 high-level GC-NOTIFY-FOO functions send their output. (There's
3785 still very little control of where low-level verbose GC functions
3786 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3787 less than it used to -- the GC-NOTIFY-FOO functions are now under
3788 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3789 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3790 in only one place in the source code, and propagates it automatically
3791 everywhere that it's needed. Thus e.g. when I bump the version from
3792 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3793 * The C source files now include boilerplate legalese and documentation
3794 at the head of each file (just as the Lisp source files already did).
3795 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3796 to his page will be replaced with a link to his new CLiki service.
3798 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3800 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3801 * The build process now looks for GNU make under the default name "gmake",
3802 instead of "make" as it used to. If GNU make is not available as "gmake"
3803 on your system, you can change this default behavior by setting the
3804 GNUMAKE environment variable.
3805 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3806 can build without error under CMU CL.
3808 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3810 * There is now a partial SBCL user manual (with some new text and some
3811 text cribbed from the CMU CL manual).
3812 * The beginnings of a profiler have been added (starting with the
3813 CMU CL profiler and simplifying and cleaning up). Eventually the
3814 main interface should be through the TRACE macro, but for now,
3815 it's still accessed through vaguely CMU-CL-style functions and macros
3816 exported from the package SB-PROFILE.
3817 * Some problems left over from porting CMU CL to the new
3818 cross-compilation bootstrap process have been cleaned up:
3819 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3820 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3821 part when getting rid of the compiler *BACKEND* stuff.)
3822 ** The value of *NULL-TYPE* was screwed up, because it was
3823 being initialized before the type system knew the final
3824 definition of the 'NULL type. This screwed up several key
3825 optimizations in the compiler, causing inefficiency in all sorts
3826 of places. (I found it because I wanted to understand why
3827 GET-INTERNAL-RUN-TIME was consing.)
3828 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3829 PROCLAIM FTYPE information. Unfortunately this broke binary
3830 compatibility again, since now the forms output by DEFGENERIC
3831 to refer to functions which didn't exist in 0.6.3.
3832 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3833 can use the new (as of 0.6.3) transform for SOME into MAP into
3835 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3836 values to (INTEGER 0 1000000), so that the time code will no longer
3837 occasionally get blown up by Linux returning 1000000 microseconds
3838 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3839 its output conform to the ANSI spec. (Alas, this makes its output
3840 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3841 * A full call to MAP NIL with a single sequence argument no longer conses.
3842 * fixes to problems pointed out by Martin Atzmueller:
3843 * The manual page no longer talks about multiprocessing as though
3844 it were currently supported.
3845 * The ILISP support patches have been removed from the distribution,
3846 because as of version 5.10.1, ILISP now supports SBCL without us
3847 having to maintain patches.
3848 * added a modified version of Raymond Toy's recent CMU CL patch for
3849 EQUALP comparison of HASH-TABLE
3851 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3853 * The system still can't cross-compile itself with
3854 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3855 entails), but at least it can compile more of itself that way
3856 than it used to be able to, and various buglets which were uncovered
3857 by trying to cross-compile itself that way have now been fixed.
3858 * This release breaks binary compatibility again. This time
3859 at least I've incremented the FASL file format version to 2, so that the
3860 problem can be detected reliably instead of just causing weird errors.
3861 * various new style warnings:
3862 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3863 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3864 ** using the lexical binding of a variable named in the *FOO* style
3865 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3866 as specified by ANSI.
3867 * *RANDOM-STATE* is no longer automatically initialized from
3868 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3869 default behavior of the system is to repeat its behavior every time
3870 it's run. If you'd like to change this behavior, you can always
3871 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3872 old convention there was no comparably easy way to get the system to
3873 repeat its behavior every time it was run.
3874 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3875 FTYPE declarations has been removed, in favor of their ANSI
3876 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3877 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3878 the types of their sequence arguments can be determined at compile time.
3879 This is done through a new open code expansion for MAP which eliminates
3880 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3881 argument types can be determined at compile time.
3882 * The optimizer now transforms COERCE into an identity operation when it
3883 can prove that the coerced object is already of the correct type. (This
3884 can be a win for machine generated code, including the output of other
3885 optimization transforms, such as the MAP transform above.)
3886 * Credit information has been moved from source file headers into CREDITS.
3887 * Source file headers have been made more standard.
3888 * The CASE macro now compiles without complaining even when it has
3891 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3893 * (Note that the way that the PCL macroexpansions were rewritten
3894 to accommodate the change in DEFGENERIC below breaks binary
3895 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3896 not run under sbcl-0.6.2. Once we get out of alpha releases,
3897 i.e. hit release 1.0.0, we'll probably try to maintain binary
3898 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3899 and sbcl-1.4.4. Until then, however, it might be fairly common
3900 for maintenance releases to break binary compatibility.)
3901 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3903 * The reporting of unhandled signals has been changed to print some
3904 explanatory text as well as the report form. (Previously only
3905 the report form was printed.)
3906 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3907 it defines, so that the compiler no longer issues undefined function
3908 warnings for compiled-but-not-yet-loaded generic functions.
3909 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3910 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3911 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3912 section of the ANSI spec are supported.
3913 * The "" nickname for the "KEYWORD" package has been removed.
3914 The reader still handles symbol tokens which begin with a package marker
3915 as keywords, but it doesn't expose its mechanism for doing so in the
3916 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3917 * The system now issues STYLE-WARNINGs for contradictory TYPE
3918 proclamations. (Warnings for contradictory FTYPE proclamations would
3919 be nice too, but those can't be done usefully unless the type system
3920 is made smarter about FUNCTION types.)
3921 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3922 systematized, so that "*target-*.lisp is supposed to exist only on the
3923 target and imply that there's a related file which exists on the
3924 host, and *host-*.lisp is supposed to exist only on the host and imply
3925 that there's a related file which exists on the target. This involves a
3926 lot of renaming. Hopefully the acute confusion caused by the renaming
3927 will be justified by the reduction in chronic confusion..
3928 ** runtime-type.lisp -> early-target-type.lisp
3929 ** target-type.lisp -> late-target-type.lisp
3930 ** early-host-format.lisp -> early-format.lisp
3931 ** late-host-format.lisp -> late-format.lisp
3932 ** host-error.lisp -> misc-error.lisp
3933 ** early-error.lisp -> early-target-error.lisp
3934 ** late-error.lisp -> late-target-error.lisp
3935 ** host-defboot.lisp -> early-defboot.lisp
3936 ** code/misc.lisp -> code/target-misc.lisp
3937 ** code/host-misc.lisp -> code/misc.lisp
3938 ** code/numbers.lisp -> code/target-numbers.lisp
3939 ** code/early-numbers.lisp -> numbers.lisp
3940 ** early-host-type.lisp -> early-type.lisp
3941 ** late-host-type.lisp -> late-type.lisp
3942 ** host-typep.lisp -> typep.lisp
3943 ** load.lisp -> target-load.lisp
3944 ** host-load.lisp -> load.lisp
3945 ** host-disassem.lisp -> disassem.lisp
3946 ** host-insts.lisp -> insts.lisp
3947 ** byte-comp.lisp -> target-byte-comp.lisp
3948 ** host-byte-comp.lisp -> byte-comp.lisp
3949 ** host-signal.lisp -> signal.lisp
3950 ** host-defstruct.lisp -> defstruct.lisp
3951 ** late-target-type.lisp -> deftypes-for-target.lisp
3952 Furthermore, several other previously target-only files foo.lisp (e.g.
3953 hash-table.lisp and random.lisp) have been split into a target-and-host
3954 foo.lisp file and a target-only target-foo.lisp file, with their key type
3955 definitions in the target-and-host part, so that the cross-compiler will
3956 know more about target types.
3957 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3958 gone away. In their place are various *BACKEND-FOO* variables
3959 corresponding to the slots of the old structure.
3960 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3961 into the target SBCL has been fixed.
3962 * The chill.lisp system for loading cold code into a running SBCL
3964 * Support for the CMU CL "scavenger hook" extension has been removed.
3965 (It was undocumented and unused in the CMU CL sources that SBCL was
3966 derived from, and stale in sbcl-0.6.1.)
3967 * Various errors in the cross-compiler type system were detected
3968 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3969 (enabling various consistency checks). Many of them were fixed,
3970 but some hard problems remain, so the compiler is back to
3971 running without *TYPE-SYSTEM-INITIALIZED* for now.
3972 * As part of the cross-compiler type system cleanup, I implemented
3973 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3974 * I have started adding UNCROSS calls throughout the type system
3975 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3976 on cross-compiler input files will be able to go away, and various
3978 * CONSTANTP now returns true for quoted forms (as explicitly required
3981 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3983 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3984 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3985 (MAKE-LIST -1). In the longer term, it ought to become true
3986 that declarations are assertions even at SAFETY 1. For now, it's not
3987 quite true even at SAFETY 3, but it's at least more nearly true..
3988 (Note that this change seems to increases the size of the system by
3989 O(5%) and to decrease the speed of the compiler by 20% or more.)
3990 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3991 for the problem of printing dozens of lines of distracting information
3992 about low-level system machinery as part of the top stack frame
3993 on entry to the debugger when an undefined function was called.
3994 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3995 is not reset to COMMON-LISP-USER.
3996 * Compilation of stuff related to dyncount.lisp has been made conditional
3997 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3998 smaller. The various dyncount-related symbols have been moved into
3999 a new "SB-DYNCOUNT" package.
4000 * tty-inspect.lisp has been renamed to inspect.lisp.
4001 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4002 feature has gone away. (When we eventually port to other flavors of
4003 libc and/or Unix, we'll try to make the differences between flavors
4004 invisible at the user level.)
4005 * Various other *FEATURES* tags, and/or their associated conditionals,
4006 have been removed if obsolescent, or given better documentation, or
4007 sometimes given more-mnemonic names.
4009 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4011 * tidied up "make.sh" script
4012 * tidied up system directory structure
4013 * better "clean.sh" behavior
4014 * added doc/FOR-CMUCL-DEVELOPERS
4015 * many many small tweaks to output format, e.g. removing possibly-confusing
4016 trailing #\. character in DESCRIBE-INSTANCE
4017 * (EQUALP #\A 'A) no longer signals an error.
4018 * new hashing code, including EQUALP hashing
4019 * tidied up Lisp initialization and toplevel
4020 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4021 * command line argument processing
4022 * added POSIX-GETENV function to deal with Unix-ish environment variables
4023 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4024 terminating SBCL on EOF
4025 * non-verbose GC by default
4026 * There is no more "sbcl" shell script; the sbcl file is now the C
4027 runtime executable (just like CMU CL).
4028 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4030 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4031 symbol and package information as Lisp native symbol and package objects
4032 instead of strings naming symbols and strings naming packages. This way,
4033 whenever packages are renamed (as in warm init), debug information is
4034 transformed along with everything else.
4035 * tweaked the optimization policy declarations which control the building
4036 of SBCL itself. Now, among other things, the system no longer saves
4037 source location debugging information. (This helps two problems at once
4038 by reducing SBCL size and by keeping SBCL from trying to look for its
4039 sources -- which may not exist -- when reporting errors.)
4040 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4041 debugging and testing purposes
4042 * cleaned up printing, making the printer call PRINT-OBJECT for
4043 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4044 methods, giving nearly-ANSI behavior
4045 * converted almost all special variables to use *FOO* naming convention
4046 * deleted PARSE-TIME functionality, since it can be done portably
4047 * moved some files out of cold init into warm init
4048 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4050 * regularized formatting of source files
4051 * added an install.sh script
4052 * fixed ridiculous memory usage of cross-compiler by making
4053 compiler/alloc.lisp not try to do pooling unless it can hook
4054 itself into the GC of the cross-compilation host. Now the system
4055 builds nicely on my old laptop.
4056 * added :SB-ALLOC in target-features.lisp-expr
4057 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4058 was not implemented)
4059 * re-did condition handling and note reporting in the compiler. Notes
4060 are no longer handled by signalling conditions. Style warnings
4061 and warnings are handled more correctly and reported in such a way
4062 that it's easy to find one or the other in your output (so that you
4063 can e.g. figure out which of many problems caused COMPILE-FILE to
4065 * changed the severity of several compiler warnings from full WARNING
4066 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4067 compiler note reporting so that it doesn't use the condition system
4068 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4069 COMPILE-FILE command)
4070 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4071 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4072 no longer supported, since their implementation was deeply intertwingled
4073 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4074 * removed bogus "support" for compiler macros named (SETF FOO), and
4075 removed the compiler macro for SETF INFO (but only after making a fool
4076 of myself on the cmucl-imp mailing list by posting a bogus patch for
4077 DEFINE-COMPILER-MACRO..)
4078 * Compiled files containing forms which have side effects on the Lisp
4079 reader (such as DEFPACKAGE forms) are now handled more correctly.
4080 (Compiler queuing of top level lambdas has been suppressed by setting
4081 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4082 * deleted various currently-unused source files, e.g. gengc.lisp. They
4083 may be added back at some point e.g. when porting to other architectures,
4084 but until they are it's distracting to distribute them and to try to
4086 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4087 there were so many of them they're just distractions, and UNCROSS is
4088 known to be able to handle the current sources
4089 * moved PROFILE functionality into TRACE, so that it will be clear
4090 how the wrapping and unwrapping of functions when you profile them
4091 interacts with the wrapping and unwrapping of functions when you
4092 trace them. (Actually, the functionality isn't there yet, but at least
4093 the interface specification is there. Hopefully, the functionality will
4094 arrive with some future maintenance release.)
4095 * removed host-oops.lisp
4096 * changed signature of QUIT function to allow UNIX-CODE argument
4097 * fixed READ-SEQUENCE bug
4098 * tweaked verbose GC output so that it looks more like the progress
4099 output that ANSI specifies for functions like LOAD
4100 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4101 * added <http://sbcl.sourceforge.com> to the banner information printed by