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 * improvement: sb-sprof traces call stacks to an arbitrary depth on
9 x86/x86-64, rather than the previous fixed depth of 8
10 * bug fix: non-ascii command-line arguments are processed correctly
11 (thanks to Yaroslav Kavenchuk)
12 * bug fix: TRACE :ENCAPSULATE NIL (and function end breakpoints)
14 * bug fix: non-required arguments were not passed correctly when a method
15 defined using DEFMETHOD was called from a mop-generated method using
16 CALL-NEXT-METHOD (reported by Pascal Costanza)
17 * bug fix: recursion is now permitted in accessors through
18 SLOT-UNBOUND. (reported by Pascal Costanza)
19 * bug fix: an error was signaled at startup if the HOME environment
20 variable was defined, but had an empty value (reported by Peter Van Eynde)
21 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
22 signal a PROGRAM-ERROR, not a BUG.
23 * bug fix: SB-POSIX:READDIR works when built with large file support.
24 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
26 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
27 for better automatic indentation support. (thanks to Matt Pillsbury)
28 * optimization: loading generic functions no longer takes O(n^2) time,
29 proportional to the amount of methods in the generic function
30 (reported by Todd Sabin and Jeremy Brown)
31 * optimization: the FIND and POSITION family of sequence functions
32 are significantly faster on arrays whose element types have been
35 changes in sbcl-1.0 relative to sbcl-0.9.18:
36 * improvement: experimental support for threading on FreeBSD/x86.
37 (thanks to NIIMI Satoshi)
38 * improvement: runtime option --dynamic-space-size can be used
39 to set the size of the dynamic space reserved on startup.
40 * improvement: floating point modes in effect are now saved in
41 core, and restored on startup.
42 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
43 startup, not time since first call to GET-INTERNAL-REAL-TIME.
44 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
45 threads are not running after *SAVE-HOOKS* have run.
46 * improvement: writes to CLOS instance slots are type-checked in code
47 compiled with (SAFETY 3)
48 * improvement: floating-point exception handling on FreeBSD (thanks to
50 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
51 (thanks to Zach Beane)
52 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
54 * improvement: added support for the Shift-JIS external format.
55 (contributed by NIIMI Satoshi)
56 * improvement: callbacks are supported on Linux/PPC. (thanks to
58 * bug fix: compiler bug triggered by a (non-standard) VALUES
59 declaration in a LET* was fixed. (reported by Kaersten Poeck)
60 * bug fix: file compiler no longer confuses validated and already
61 dumped structurres. (reported by Kaersten Poeck)
62 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
63 fill-pointer signals a type-error as required. (thanks to
65 * bug fix: disassemly of funcallable instances works.
66 * bug fix: single stepping on PPC.
67 * bug fix: fix thread-safety problems in the type system (generally
68 manifesting as nonsensical errors like "STRING is a bad type specifier
69 for sequences" or "The value 1 is not of type FIXNUM")
70 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
71 (reported by Josip Gracin)
72 * bug fix: an error is signaled for attempts to displace arrays with
73 incompatible element types (thanks to Mario Mommer)
74 * bug fix: more correct handling of wide characters in debug info
75 (bug reported by Attila Lendvai and fixed by Juho Snellman)
76 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
77 and don't cause extra consing
78 * optimization: MAP and MAP-INTO are significantly faster on vectors
79 whose elements types have been declared.
80 * Improvements to SB-SPROF:
81 ** Support for allocation profiling
82 ** Reduced profiling overhead, especially for long profiling runs
83 * Improvements to the Windows port:
84 ** floating point exceptions are now reported correctly.
85 ** stack exhaustion detection works partially.
86 ** more accurate GET-INTERNAL-REAL-TIME.
87 ** better RUN-PROGRAM behaviour in case of errors in the child
89 ** PROBE-FILE now simplifies pathnames correctly.
90 ** DIRECTORY now works correctly with :WILD-INFERIORS.
91 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
92 1st 1970 (although time timezone data will be unreliable).
93 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
95 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
96 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
97 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
99 * bug fix: two potential GC deadlocks affecting threaded builds.
100 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
101 returns T (reported by Anton Kazennikov)
102 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
103 with non-variable places
104 * bug fix: remove a race condition in the setting of
105 funcallable-instance functions, this should make threaded CLOS
106 code more stable against memory faults.
107 * bug fix: corruption of specials when unbinding is interrupted by an
108 asynchronous unwind (reported by Hannu Koivisto)
109 * improvement: the debugger will now also display local variables that
110 are only used once, for code compiled with a DEBUG optimization quality
113 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
114 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
115 * incompatible change: External-format support for FFI calls. The
116 SB-ALIEN:C-STRING no longer implies an ASCII
117 external-format. Instead the string is subject to external-format
118 conversion. Additionally return values of type SB-ALIEN:C-STRING
119 are by default of type (SIMPLE-ARRAY CHARACTER), not
120 SIMPLE-BASE-STRING. For an alien type that behaves like the old
121 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
122 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
123 * incompatible change: SB-EXT package no longer contains the
124 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
125 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
127 * incompatible change: the single-stepper is no longer available
128 on Alpha, Mips and Sparc platforms.
129 * minor incompatible change: the direct superclasses of
130 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
131 not (STANDARD-OBJECT FUNCTION). This makes the
132 class-precedence-lists of GENERIC-FUNCTION and
133 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
135 * minor incompatible change: the default stream external format on
136 non-unicode SBCL is no longer detected from the locale, but is always
138 * new feature: Add a version of evaluator that uses an interpreter instead
139 of the compiler. EVAL still uses the compiler by default, to switch it
140 to use the interpreter, set the value of the variable
141 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
142 * minor incompatible change: the single-stepper REPL has been merged
143 with the normal debugger (see the "Stepping" heading of the debugger help
144 for more details). The debugger command STEP will no longer switch
145 to the single-stepper REPL.
146 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
147 for a type now works.
148 * bug fix: loading of default sysinit file works. (thanks to Leonid
150 * bug fix: better detection of circularities in the file-compiler.
151 (reported by Marco Monteiro)
152 * bug fix: the CL pathname functions now work with files that have
153 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
154 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
155 systems with Unix98 pty semantics.
156 * bug fix: ASDF-INSTALL will now work with bsd tar.
157 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
159 * bug fix: timers expiring in dead threads no longer cause a
160 type-error (reported by Paul "Nonny Mouse").
161 * bug fix: thanks to more lightweight single-stepper instrumentation,
162 code compiled with (DEBUG 3) will compile and execute significantly faster,
163 and will have more accurate type-inferencing than before
164 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
165 whose bindings are modified
166 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
167 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
168 ** replace dummy implementations of CL:MACHINE-INSTANCE and
169 CL:SOFTWARE-VERSION with proper version
171 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
172 * feature: implemented the READER-METHOD-CLASS and
173 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
174 as specified by AMOP.
175 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
177 * optimization: faster LOGCOUNT implementation on x86 and x86-64
178 (thanks to Lutz Euler)
179 * optimization: hashing of general arrays and vectors has been
180 improved. (reported by Any Fingerhut)
181 * enhancement: SB-INTROSPECT is now able to find definitions of
182 profiled functions. (thanks to Troels Henriksen)
183 * enhancement: compiler-macro expansion applies now to FUNCALL forms
184 as well, allowing compiler-macros for SETF-functions to expand.
185 * enhancement: step-instrumentation no longer wraps known
186 single-value functions with multiple-value context, allowing
187 better type inference.
188 * fixed bug #337: use of MAKE-METHOD in method combination now works
189 even in the presence of user-defined method classes. (reported by
190 Bruno Haible and Pascal Costanza)
191 * fixed bug #339(c): if there are applicable methods not part of any
192 long-form method-combination group, call INVALID-METHOD-ERROR.
193 (reported by Bruno Haible)
194 * fixed bug #361: the :FUNCTION initarg in the protocol for
195 initialization of methods can now be used to override
196 internally-produced optimized functions. (reported by Bruno
198 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
199 system-provided lambda expression no longer cause warnings about
200 unbound #:|pv-table| symbols.
201 * bug fix: improved the handling of type declarations and the
202 detection of violations for keyword arguments with non-constant
204 * bug fix: potentially erronous calls to PATHNAME and
205 MERGE-PATHNAMES were being flushed in some cases.
206 (reported by Richard Kreuter)
207 * bug fix: compiled calls to TYPEP were mishandling obsolete
208 instances. (reported by James Bielman and Attila Lendvai)
209 * bug fix: format strings with ~> without matching ~< no longer
210 trigger an AVER, but signal an understandable error instead.
211 (reported by Antonio Martinez)
212 * bug fix: specifying an output-file without a pathname-type for
213 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
214 of input-file instead of "fasl". (reported by Robert Dodier)
215 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
216 (reported by James Y Knight).
217 * bug fix: compiler-macros-function did not consider the environment
218 argument for shadowing by local functions.
219 * bug fix: compiler-macros expansion was inhibited by local INLINE
221 * bug fix: inline expansions of known functions were subject to
222 step-instrumentation in high debug policies, leading to problems
224 * bug fix: compiler failed to differentiate between different CONS
226 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
227 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
228 * bug fix: #368: incorrect use of expressed vs. upgraded array
230 * bug fix: #306a: more precise unions of array types.
231 * thread-safety improvements:
232 ** CONDITION-WAIT could return early on Linux, if the thread was
233 interrupted and subsequently continued with SIGCONT.
234 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
236 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
237 * added support for the ucs-2 external format. (contributed by Ivan
239 * minor incompatible change: pretty printing of objects of type
240 (cons symbol) is, in the default pprint-dispatch-table, now
241 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
243 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
244 called later in a class's lifetime, possibly as late as when the
245 first instance of the class is created. Previously,
246 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
247 class became finalizeable.
248 * fixed bug: FILE-POSITION sometimes returned inconsistent results
249 for multibyte external-format streams. (thanks to "vbzoli")
250 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
251 with :ALLOCATION :CLASS inherited from superclasses of the
253 * fixed bug: anonymous classes can now be created using the :NAME
254 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
255 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
256 * fixed bug: core-files saved with :EXECUTABLE T can again be
257 executed when SBCL_HOME isn't set. (reported by James Knight)
258 * fixed bug: toplevel LOCALLY forms with declarations could
259 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
260 * fixed bug: printing from several different threads using different
261 values of *print-case* could cause invalid output, due to
262 some internal special variables of the printer not being bound
263 thread-locally (reported by Max Mikhanosha)
264 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
265 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
266 (reported by Pascal Costanza)
267 * fixed bug: FINALIZE-INHERITANCE is called from
268 REINITIALIZE-INSTANCE on classes when the class has previously
269 been finalized, as required by AMOP.
270 * minor code generation optimizations:
271 ** better register allocation in CLOS dispatching functions
272 ** overflow detection when coercing signed bytes to fixnums on x86-64
273 ** is now implemented with one IMUL instruction instead of three shifts
274 ** more efficient bit-vector access on x86 and x86-64
275 ** more efficient access to raw structure slots on x86 and x86-64
276 * fixed some bugs revealed by Paul Dietz' test suite:
277 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
280 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
281 * feature: thread support on Solaris/x86, and experimental thread support
283 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
285 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
286 default method for SLOT-UNBOUND.
287 * minor incompatible change: prevent the user from specializing the
288 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
289 somewhat counter to the intent of the protocol, I (CSR) think, and
290 additionally it just doesn't work in SBCL as currently
291 implemented, thanks to optimizations (that are always valid for
292 the other three SLOT-VALUEish functions, but not for the setter).
293 * minor incompatible change: the :SB-LDB feature is now enabled by
294 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
295 the low-level debugger.
296 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
297 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
298 down to an SB-SYS:FD-STREAM, allowing direct communication in
299 more cases. Temporary files and pipes are still used as a fallback
301 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
302 on the non-reentrant C functions localtime(3) and gmtime(3).
303 * bug fix: global optimization policy was not visible in LOCALLY and
305 * bug fix: class objects can be used as specializers in methods.
306 (reported by Pascal Costanza)
307 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
308 with a type error. (reported by blitz_ on #lisp)
309 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
310 failed with a type error. (reported by Pascal Bourguignon)
311 * bug fix: merging pathnames against defaults with :DIRECTORY
312 starting with '(:RELATIVE :BACK) should preserve the :BACK.
313 (reported by James Y Knight)
314 * bug fix: saving large (>2GB) cores on x86-64 now works
315 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
316 constant index and a (simple-array (signed-byte 32)) array
317 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
318 error (patch by Robert J. Macomber)
319 * feature: TIME now displays the amount of run-time spent in GC
320 * bug fix: The debugger now does a better job of respecting
321 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
322 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
323 workaround for bug 403.)
324 * bug fix: floating point exception handling now works on FreeBSD
325 (thanks to NIIMI Satoshi)
326 * fixed some bugs revealed by Paul Dietz' test suite:
327 ** MISC.641: LET-conversion were not supposed to work in late
329 * improvements to the Win32/x86 port:
330 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
331 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
333 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
334 * new feature: source path information is generated for macro-expansion
335 errors for use in IDE's like Slime (thanks to Helmut Eller)
336 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
337 * bug fix: compiler does not loop forever on an invalid type in
339 * improvement: compilation of most CLOS applications is significantly
341 * optimization: added a limited bytecode compiler for simple toplevel
342 forms, speeding up compilation and FASL loading
343 * bug fix: the statistical profiler now properly distinguishes anonymous
346 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
347 * minor incompatible change: in sbcl-0.9.11 (but not earlier
348 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
349 into subclasses of the finalized class. Now user calls to
350 FINALIZE-INHERITANCE finalize just the one class, and calls by the
351 system return before any subclasses are finalized.
352 * minor incompatible change: The reader no longer ignores errors
353 regarding non-existent packages in #+ and #- feature tests.
354 * new feature: command line options --no-sysinit, --no-userinit to
355 inhibit loading the corresponding init files
356 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
357 for sending data through UDP sockets (thanks to François-René Rideau)
358 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
359 error is signalled instead (thanks to François-René Rideau)
360 * minor incompatible change: Error signalling behaviour of lexical
361 operations violating package locks has changed slightly. Refer to
362 documentation on package locks for details.
363 * enhancement: EVAL can process IF-expressions without resorting to the
365 * bug fix: Some operations involving SIN, COS, or TAN weren't being
366 constant-folded properly due to missing out-of-line functions.
367 (reported by Mika Pihlajamäki)
368 * bug fix: LISTEN sometimes returned T even in cases where no data was
369 immediately available from the stream
370 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
371 were reversed. (reported by Levente Mészáros)
372 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
373 when given a vector displaced to an adjustable array. (reported
375 * bug fix: derivation of float boundaries from numbers outside the
376 appropriate float range (reported by John Wiseman)
377 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
379 * bug fix: precomputing cache entries for generic functions with
380 some subclasses of specializers as yet invalid does not attempt to
381 fill a cache line with a negative offset. (reported by Levente
383 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
384 allow condition class objects as arguments to DOCUMENTATION and
385 (SETF DOCUMENTATION); only find and set documentation for
386 structure names for the STRUCTURE doc-type. (suggested by Gary
388 * improvements to the Win32/x86 port:
389 ** (user-homedir-pathname) and default initialization file
390 locations now know about the user's "Documents and Settings"
391 directory (thanks to Yaroslav Kavenchuk)
392 ** run-program is implemented (thanks to Mike Thomas)
393 ** sockets support (thanks to Timothy Ritchey)
394 ** better backtrace support (thanks to Alastair Bridgewater)
395 ** sb-grovel supported
396 ** asdf-install and sb-posix work somewhat
397 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
398 * improvements to the Solaris/x86 port:
399 ** works on Solaris 11/Solaris Express
400 ** floating-point exception handling support
401 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
402 * fixed some bugs revealed by Paul Dietz' test suite:
403 ** REMOVE-METHOD returns its generic function argument even when
404 the method is not one of the generic functions' methods.
405 ** objects declared as MEMBER types can be admissible arguments to
407 ** printing characters should simply be printed by the FORMAT ~:C
409 ** compiler failure when compiling functions with hairy constant
410 defaults for optional parameters.
411 ** compiler produces wrong code when MAYBE-INLINE-expanding a
412 function, which is already optimized.
414 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
415 * new platform: experimental support for SBCL x86/Darwin, including
416 MacOS X 10.4.5 on Intel.
417 * new feature: Unicode character names are now known to the system
418 (through CHAR-NAME and NAME-CHAR).
419 * new feature: the filesystem location of SBCL's core file is
420 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
421 * minor incompatible change: the contrib modules SB-POSIX and
422 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
423 this change is to make it easier to distribute
424 location-independent binaries.
425 * bug fix: as implied by AMOP, standardized classes no longer have
426 slots named by external symbols of public packages. (reported by
428 * optimization: calling functions via constant symbols -- as in
429 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
430 via the function object as in (FUNCALL #'FOO).
431 * optimization: on x86 and x86-64, the calling convention for the
432 case of dealing with an unknown number of values has been altered
433 to be friendlier to the prediction heuristics implemented,
434 particularly with reference to CALL and RET pairing. (thanks to
435 Alastair Bridgewater)
436 * optimization: on x86, the code for access to thread-local symbol
437 values and binding/unbinding of thread-local symbols is smaller.
438 (thanks to Alastair Bridgewater)
439 * enhancement: CONSTANTP is now able to determine constantness of
440 more complex forms, including calls to constant-foldable standardized
441 functions and some special forms beyond QUOTE.
442 * fixed bug: occasional GC crashes on Solaris/x86
443 * optimization: x86-64 supports stack allocation of results of simple
444 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
445 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
446 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
447 (thanks to James Knight)
448 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
449 by the standard (thanks to Stephen Compall)
451 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
452 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
453 be used for bundling the runtime and the core file into one
454 executable binary. This feature is not currently supported on all SBCL
455 platforms. (thanks to James Bielman and NIIMI Satoshi)
456 * new feature: a generational or ephemeral garbage collector is now
457 the default on the PowerPC platform (both Linux and Darwin). The
458 old Cheney (stop and copy) collector is a build-time option.
459 (thanks to Cyrus Harmon, after Raymond Toy)
460 * minor incompatible change: the method by which SBCL finds its
461 contributed modules has changed; it no longer relies on symbolic
462 links from an $SBCL_HOME/systems directory, but searches directly
463 in the subdirectories of $SBCL_HOME.
464 * enhancement: the dynamic heap size on the Linux/PPC platform is
465 markedly larger, even using the older Cheney garbage collector.
466 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
467 certain complicated string types. (reported by Gary King)
468 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
469 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
470 * fixed bug: DOCUMENTATION on structures no longer signals an
471 error if no structure type of the right name exists. (reported by
473 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
474 applicable non-standard primary or :AROUND method on
475 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
476 no longer cause unbound variable errors. (reported by Kevin Reid)
477 * fixed bug: metacircle resolution in cases where methods have slots
478 added before the slots from STANDARD-METHOD. (reported by Jean
480 * fixed bug: the Power PC assembly code for calling into lisp
481 sometimes computed its return address wrongly (depending on the
482 15th bit of the address, fixed at link time). This used to
483 manifest itself in a segmentation violation while building PCL.
484 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
485 many others over the years)
486 * fixed bug: ROOM no longer reports silly numbers for stack usage if
487 the stack pointer (treated as a signed integer) is negative.
488 (thanks to Peter van Eynde)
490 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
491 * new platform: experimental support for the Windows operating
492 system has been added. (thanks to Alastair Bridgewater)
493 * fixed several bugs in and robustified the PPC FFI (including
494 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
495 * bug fix: allow non-simple string symbol names (reported by Paul
497 * bug fix: interrupt handling on NetBSD (thanks to Richard M
499 * bug fix: saving a core corrupted callbacks on x86/x86-64
500 * bug fix: closed a loophole in metacircularity detection and
501 grounding in the PCL implementation of CLOS.
502 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
503 (reported by tomppa on #lisp)
504 * bug fix: clear the direction flag on Lisp -> C transitions, as
505 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
506 (reported by Andrej Grozin and Hendrik Maryns)
507 * optimization: major improvements to GC efficiency on GENCGC platforms
508 * optimization: faster implementation of EQUAL
509 * optimization: emit more efficient opcodes for some common
510 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
511 * optimization: several other minor code-generation improvements on x86-64
512 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
514 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
515 * minor incompatible change: (SETF CLASS-NAME) and (SETF
516 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
517 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
518 by ANSI as a generic function, and for consistency (SETF
519 GENERIC-FUNCTION-NAME) is treated likewise.
520 * fixed bug #233.b: make constraint propagation notice when a variable
521 value is changed after it is referenced but before it is used
522 * fixed bug #296: no more arbitrary behaviour from filenames with
523 odd characters as --load arguments. (reported by Adam Warner)
524 * fixed bug #390: :CHARACTER-SET pathname components now work as
525 expected. (reported by Tim Daly Jr)
526 * fixed bug #391: complicated :TYPE intersections in slot
527 definitions no longer cause an error in PCL internals.
528 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
529 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
530 returning the number of octets which would be written to the
531 file-stream. (thanks to Robert J. Macomber)
532 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
533 arguments to the internal stream functions. (thanks to David
535 * bug fix: stack exhaustion is now handled on the x86/SunOS
536 platform. (thanks to Daisuke Homma)
537 * bug fix: more accurate ROOM results on GENCGC platforms
538 * optimization: improved type inference for arithmetic-for
539 index variables in LOOP
540 * optimization: faster floating-point SQRT on x86-64
541 * fixed some bugs revealed by Paul Dietz' test suite:
542 ** DOCUMENTATION returns NIL instead of "" for method combinations
543 that don't have a docstring
545 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
546 * minor incompatible change: (SETF CLASS-NAME) and (SETF
547 GENERIC-FUNCTION-NAME) are no longer generic functions, and
548 therefore are not customizeable by user code (as seems to be at
549 least permitted and maybe required by AMOP). As a consolation,
550 however, the SBCL implementation of these functions now calls
551 REINITIALIZE-INSTANCE as specified by AMOP.
552 * bug fix: slot-definition documentation is propagated as per ANSI
553 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
554 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
555 argument, as do eslotd initialization methods. (from Pascal
556 Costanza's "Closer" project)
557 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
558 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
560 * bug fix: it is now possible to have more than one subclass of
561 STANDARD-GENERIC-FUNCTION without causing stack overflow.
562 (reported by Bruno Haible, Pascal Costanza and others)
563 * bug fix: the dependent update protocol now works for generic
564 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
566 * bug fix: condition-class instances corresponding to
567 DEFINE-CONDITION forms are now created eagerly. (reported by
568 Kalle Olavi Niemitalo on comp.lang.lisp)
569 * bug fix: floating point printing is more accurate in some
570 circumstances. (thanks to Simon Alexander)
571 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
572 merged with *DEFAULT-PATHNAME-DEFAULTS*.
573 * bug fix: callbacks on OS X now preserve stack-alignment.
574 (thanks to Cyrus Harmon)
575 * enhancement: the x86-64 disassembler is much better at
576 disassembling SSE instructions. (thanks to Lutz Euler)
577 * enhancement: CHECK-TYPE on a variable now allows type-inference
578 (thanks to Vincent Arkesteijn)
579 * optimization: improved performance of EUC-JP external format.
580 (thanks to NIIMI Satoshi)
581 * optimization: performance improvements to IO on file streams of
582 :ELEMENT-TYPE CHARACTER
583 * optimization: much faster memory allocation on x86-64
584 * optimization: faster garbage collections (latency and throughput)
586 * optimization: faster bignum subtraction and division on x86 and x86-64
587 (thanks to Lutz Euler)
588 * optimization: more accurate type-inference for loops that use a
589 floating point index variable or a negative step.
591 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
592 * bug fix: add a workaround to SBCL looping infinitely at startup on
593 Linux kernels with apparently buggy implementations of personality().
594 (thanks to Svein Ove Aas)
595 * bug fix: Unicode symbols are correctly printed in LDB backtraces
596 (thanks to David Lichteblau)
597 * bug fix: local bindings shadow global symbol macros for
598 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
599 * optimization: non-open coded uses of numeric comparison operators
600 (e.g. >) no longer cons when called with more than one parameter
601 on platforms supporting dynamic-extent allocation.
602 * enhancement: saving cores with foreign code loaded is now supported
603 on MIPS/Linux in addition to the previously supported platforms.
604 * bug fix: FIND-RESTART now tests for activity, not applicability when given
605 a restart object as identifier. (reported by Helmut Eller for CMUCL)
606 * bug fix: division by zero in sb-sprof when no samples were collected
607 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
608 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
609 (thanks to Alexander Kjeldaas)
610 * new feature: ignoring the return values of destructive functions
611 like NREVERSE signals a compile-time style-warning.
612 (thanks to Kevin Reid)
613 * new feature: the alignment of alien structure fields can be
614 explicitly specified. (thanks to Cyrus Harmon)
615 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
616 and dump core on SIGQUIT
618 ** incompatible change: threads do not inherit values of specials
619 from their parents (see manual)
620 ** bug fix: threads stacks belonging to dead threads are freed by the
621 next exiting thread, no need to gc to collect thread stacks anymore
622 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
623 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
624 ** bug fix: dynamic variable and thread start related gc lossage
625 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
627 * fixed some bugs revealed by Paul Dietz' test suite:
628 ** SUBTYPEP is slightly more accurate on heinously complicated
629 CONS types where some of the members have uncertain (in the
630 NIL, NIL sense) type relationships to each other.
632 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
633 no longer purified unless :PURIFY T is explicitly specified.
634 ** Non-purified cores are significantly smaller than before
636 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
637 * new feature: timers based on Zach Beane's excellent timer package
638 * added support for the following external formats: koi8-u,
639 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
640 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
641 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
642 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
643 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
644 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
645 * incompatible change: a threaded SBCL will no longer revert to
646 non-threaded mode on non-NPTL systems, but refuse to start entirely.
647 * bug fix: interrupts are disabled until startup is complete; no
648 more sigsegvs when receiving a signal to soon
649 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
651 * bug fix: add a workaround for the memory randomization features in
652 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
653 workaround will only be in effect on systems with the proc filesystem
655 * bug fix: printing objects of type HASH-TABLE signals a
656 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
658 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
659 have been running for over 50 days. (reported by Gilbert Baumann)
660 * bug fix: the logic for getting names of functions gets less
661 confused when confronded with alternate-metaclass
662 funcallable-instances. (reported by Cyrus Harmon)
663 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
664 consistent, even on internal alternate-metaclass objects.
665 * bug fix: flush closure information collected by physical
666 environment analysis prepass before the main pass. (bug reported
668 * bug fix: compiler pack phase does not modify a hashtable, which is
669 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
670 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
671 may be expanded. (reported by Alexander <malishev> on c.l.l)
672 * bug fix: no more style-warnings when compiling code using LAST
673 under high-SPEED low-SPACE settings. (reported by David Wallin
675 * fixed some incompatibilities between SBCL's MOP and the MOP
677 ** the METAOBJECT class is now implemented;
678 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
679 STANDARD-OBJECT, as required;
680 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
681 now compatible (as required by VALIDATE-SUPERCLASS); there
682 remains a consistency requirement over the presence or absence
683 of the FUNCTION class in the superclass list of finalized
684 classes; see the manual for more details;
685 ** the :AROUND method for COMPUTE-SLOTS on
686 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
687 requested slot ordering.
689 ** bug fix: parent thread now can be gc'ed even with a live
691 ** bug fix: binding a special with PROGV to no value is not
692 the same as that symbol not having been bound (thanks to
694 ** bug fix: binding specials is thread safe (thanks to
696 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
697 and SB-SPROF do not die with 'no handler for signal XX in
698 interrupt_handle_now(..)' anymore
699 ** bug fix: WITH-TIMEOUT works with multiple running threads
700 * fixed some bugs revealed by Paul Dietz' test suite:
701 ** ENSURE-GENERIC-FUNCTION should take a method class object for
702 the :method-class keyword argument.
704 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
705 * new port: the Solaris operating system on x86 processors is now
706 mostly supported, though some rough edges in the environment
707 remain. (thanks to Daisuke Homma)
708 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
709 space for its heap. (thanks to Thiemo Seufer)
710 * optimization: SBCL on MIPS platforms now supports dynamic-extent
711 closures. (thanks to Thiemo Seufer)
712 * minor incompatible change: eof selects abort in the debugger.
713 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
714 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
716 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
717 deprecated, and will go away in a future revision of SBCL.
718 * minor incompatible change: GC-ON and GC-OFF are no longer
719 implemented with a counter, it does not matter how many times gc
720 is switched on or off
721 * bug fix: discriminating functions for generic function classes
722 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
723 make invalid assumptions about method precedence order. (reported
725 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
726 platforms. (thanks to Thiemo Seufer)
727 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
728 * bug fix: correct pathname printing with printer escaping is on.
729 (thanks to Kevin Reid)
730 * bug fix: complex VOP definitions in "user-space" no longer trigger
731 package locks. (reported by Zach Beane)
732 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
733 for funcallable-instance functions, and is no different from
735 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
736 characters in the current readtable. (reported by Nicholas Neuss)
737 * bug fix: SUBTYPEP on various CONS types returns more a more
738 accurate acknowledgment of its certainty.
739 * optimizations: REMOVE-DUPLICATES now runs in linear time on
740 lists in some cases. This partially fixes bug 384.
741 * flush all standard streams before prompting in the REPL and the
743 * bug fix: signal handling and triggering gc do not conflict
744 directly anymore, in particular a high frequency sb-sprof does
745 not prevent gc from running
746 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
747 approximation for timezone and DST information between the
748 universal time epoch and the smallest negative 32-bit time_t.
749 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
750 year 1899 so long as the encoded time is non-negative.
751 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
752 the correct answer for non-constant rotate amounts when performing
753 an inline 32-bit rotation.
755 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
756 there is only one thread in the session
757 ** bug fix: memory leak for streams created in one thread and
758 written to in another
759 ** bug fix: lockup when compiled with gcc4
760 ** bug fix: race that allows the gc to be triggered when gc is
762 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
763 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
765 ** bug fix: finished threads are gc'ed properly
766 * fixed some bugs revealed by Paul Dietz' test suite:
767 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
768 has arguments with a different set of applicable methods from
769 the orignal arguments.
770 ** The type error thrown by MAP now has a correct expected-type
772 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
773 name a compiled function.
774 ** Three MISC tests where a large bignum was improperly coerced to
775 a float that couldn't represent that bignum during type
776 derivation were fixed.
777 ** SUBTYPEP can now handle types involving the negation of a
778 list-form FUNCTION type.
779 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
780 as (COMPLEX (AND RATIO (NOT FIXNUM))).
781 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
783 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
784 * New feature: Experimental support for bivalent streams: streams
785 opened with :element-type :default now allow character and binary
786 (unsigned-byte 8) I/O
787 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
788 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
790 * bug fix: no more highly sporadic "couldn't check whether ~S is
791 readable" when reading a stream and an interrupt hits in the middle
792 of a select system call
793 * compiler better recognizes complex arrays (reported by Eduardo
795 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
796 for variables on SBCL built with linkage-tables. (reported by Luis
798 * various error reporting improvements.
799 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
800 (Thanks to Thiemo Seufer)
801 * bug fix: nested reader invokations work correctly
802 * bug fix: it is possible to have simultaneous references to foreign
803 code and foreign data with the same name.
805 ** added x86-64 support
806 ** incompatible change: the threading api now works with thread
807 objects instead of thread ids
808 ** bug fix: threads are protected from signals and interruption when
809 starting up or going down
810 ** bug fix: a race where an exiting thread could lose its stack to gc
811 ** bug fix: don't halt on infinite error in threads if possible
812 ** fixed numerous gc deadlocks introduced in the pthread merge
813 ** bug fix: fixed thread safety issues in read and print
814 ** bug fix: debugger doesn't hang on session lock if interrupted at
815 an inappropriate moment
816 ** bug fix: run-program is now thread safe(r)
817 ** bug fix: inner with-recursive-lock no longer releases the mutex
818 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
819 * fixed some bugs revealed by Paul Dietz' test suite:
820 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
821 EXPECTED-TYPE slots filled.
822 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
824 ** MAKE-CONDITION accepts classes as type-designators.
825 ** COMPILE may never return NIL.
826 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
827 range before calling Unix time functions
829 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
830 * numerous signal handling fixes to increase stability
831 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
832 * minor incompatible change: we now correctly canonize default
833 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
834 as per the MOP, rather than the historical (INITARG INITFUNCTION
835 INITFORM). (reported by Bruno Haible)
836 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
837 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
838 for more information.
839 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
840 pathname is a directory pathname.
841 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
842 * dynamic space size on PPC has been increased to 768Mb. (thanks to
844 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
845 :METACLASS argument in addition to a class name. (reported by
846 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
847 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
848 (reported by Bruno Haible; patch from Gerd Moellmann)
849 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
851 * bug fix: more cleanups to the floating point exception handling on
852 x86-64 (thanks to James Knight)
853 * bug fix: the compiler does not try to constant fold calls of
854 COERCE and BIT-* functions when they return freshly constructed
855 objects. (reported by Nikodemus Siivola and Paul Dietz)
856 * optimization: improved the MIPS versions of generic (in the
857 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
858 * optimization: direct conversion of (unsigned-byte 32) to floats on
859 the PowerPC platform.
860 * optimization: structure instances with raw slots now use less
861 memory, and probably show better memory locality. (thanks to
863 * optimization: DYNAMIC-EXTENT declarations for lists and closures
864 are treated as requests for stack allocation on the x86-64,
865 Alpha-32, PPC and SPARC platforms.
866 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
867 the wrong file descriptor; implementation of SOCKET-OPEN-P.
868 (thanks to Tony Martinez)
870 ** gcing a dead thread can no longer lead to lockups
871 ** threads block signals until they are set up properly
872 ** errno is no longer shared by threads
873 ** interrupt-thread restores the eflags register on x86
874 ** fixed some lockups due to gc/thread interaction
875 * fixed some bugs revealed by Paul Dietz' test suite:
876 ** invalid dotted lists no longer raise a read error when
878 ** signal an error if a symbol that names a declaration is used
879 as the name of a type, or vice versa
880 ** allow using the (declare (typespec var*)) abbreviation for
881 (declare (type typespec var*)) with all type specifiers
882 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
883 ** always return NIL from PPRINT-POP when OBJECT is NIL
884 ** don't signal errors when pretty-printing malformed LABELS,
885 FLET or MACROLET forms
886 ** declarations in a DOLIST body are also in scope for the
888 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
890 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
893 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
894 * fixed cross-compiler leakages that prevented building a 32-bit
895 target with a 64-bit host compiler.
896 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
897 opened with :IF-EXISTS :OVERWRITE.
898 * fixed bug 281: error for an invalid qualifier in a short-form method
899 combination method is not signalled until the faulty method is called.
900 * bug fix: iteration variable type inferrer failed to deal with open
901 intervals. (reported by Alan Shields)
902 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
903 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
904 or more runtime options were provided to the sbcl binary.
905 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
906 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
907 * optimization: there's now a fast-path for fixnum arguments in the
908 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
909 * optimization: the code generated on x86-64 is more compact thanks
910 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
911 * bug fix: floating-point exception handling now partly works on
912 x86-64. (Thanks to James Knight)
913 * improvement to the MIPS backend from Thiemo Seufer: C-style
914 64-bit long long arguments and return values to alien functions
916 * fixed some bugs revealed by Paul Dietz' test suite:
917 ** the type-error signalled from WARN has a filled-in DATUM slot.
918 ** the type-error required when a stream is not associated with
919 a file has the stream as its datum.
920 ** type-errors on single-floats on x86-64 no longer have
921 :INVALID-OBJECT as the datum
922 ** the type-errors signalled for invalid function names now have
923 a correct expected type
924 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
925 for structures defined with a :TYPE. Documentation strings for
926 typed structures are no longer immediately discarded
927 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
928 broadcast streams and synonym streams. FILE-LENGTH now also works
929 on broadcast streams.
931 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
932 * incompatible change: the --noprogrammer option, deprecated since
933 version 0.7.5, has been removed. Please use the equivalent
934 --disable-debugger option instead.
935 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
937 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
938 inoperational for a while now) has been completely removed.
939 * null lexical environments are now printed as #<NULL-LEXENV>,
940 significantly reducing the amount of clutter in typical backtraces.
941 * documentation on weak pointers, finalization, and after GC hooks
942 has been added to the manual.
943 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
944 as other specialized array types, is much faster. SUBSEQ and
945 COPY-SEQ on such arrays have also been sped up.
946 * optimization: EQL is now more efficient when at least other argument
947 is known to be of type (OR FIXNUM (NOT NUMBER)).
948 * fixed bug: the runtime is now less vulnerable to changes in the
949 size of the SBCL object on OS X, and virtual memory is reserved for
950 all spaces that need to be at a fixed address.
951 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
952 * fixed bug: finalizers and after GC hooks that cause consing are now
954 * fixed bug: compiler error messages and summaries are now printed to
955 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
956 * fixed inference of the upper bound of an iteration variable.
957 (reported by Rajat Datta).
958 * fixed bug 211e: calling local functions with duplicated constant
959 keyword argument no longer causes a bogus style warning about an
961 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
962 ftype declaration to be disregarded. (reported by Dave Roberts)
963 * fixed bug 373: caused by erronous compilation of references to alien
964 variables in the runtime on ppc/darwin.
965 * fixed bug 376: CONJUGATE type deriver.
966 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
967 (reported by Baughn on #lisp)
968 * a cross-compiler bug on non-x86 platforms has been identified and
969 fixed. (thanks to Bruno Haible)
970 * improvements to the MIPS runtime code for increased stability.
971 (thanks to Thiemo Seufer)
972 * increased the maximimum compact environment size to allow
973 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
974 * partial workaround for bug 135: don't name the function we're
975 calling for hairy cases of EVAL, so as not to accumulate one environment
976 entry per call to EVAL. (reported by Kevin Reid)
977 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
978 * optimization: type testing for non-vector arrays should be faster.
979 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
981 * bug fix: setting 31st element of a bit vector to zero did not work
983 * bug fix: redefining a class definition which failed due to a
984 previous accessor / function clash now works (but see BUGS entry
985 #380 for more problems in this area). (thanks to Zach Beane)
986 * the long form of DEFINE-METHOD-COMBINATION disables method group
987 checking when given a single method group with pattern *. (thanks
989 * on x86 compiler supports stack allocation of results of simple
990 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
991 * support for building and running on Mac OS X 10.4 "Tiger" has been added
992 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
993 * fixed some bugs related to Unicode integration:
994 ** the restarts for recovering from input and output encoding
995 errors only appear when there is in fact such an error to
997 * fixed some bugs revealed by Paul Dietz' test suite:
998 ** MISC.549 and similar: late transformation of unsafe type
999 assertions into derived types caused unexpected code
1001 ** SCALE-FLOAT type deriver is less wrong.
1002 ** type derivers for EXP, LOG and similar functions compute result
1003 types for complex arguments better.
1004 ** (MISC.563) CONJUGATE type deriver works for very restricted
1006 ** out-of-line type testers for character strings are available.
1007 ** EQUAL compiler transform understands specialness of objects
1009 ** accessing double-floats stored on the stack now works on x86-64.
1010 ** debugger internals could sometimes create invalid lispobjs,
1011 resulting in GC crashes.
1012 ** MISC.548: type check weakening can convert required type into
1014 ** initialization forms for bindings are not in scope of free special
1017 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1018 * incompatible change: thread support for non-NPTL systems has
1019 been removed - locking is buggy and unreliable. A threaded
1020 SBCL build will now warn at startup and refuse to create
1021 new threads, unless futex support is detected in the kernel
1022 * incompatible change: the top level REPL now has only an ABORT
1023 restart associated with it, not TOPLEVEL and ABORT as it used to.
1024 TOP and TOPLEVEL are now available as debugger commands for
1025 returning to the top level.
1026 * incompatible change: forms evaluated in the REPL now use the
1027 global optimization policy.
1028 * incompatible change: user- and system-initialization files are
1029 no longer processed with LOAD, but by READ and EVAL; hence the
1030 global optimization policy, startup package, readtable, etc,
1032 * The .fasl file format number has been incremented because of
1033 various incompatible changes.
1034 * internal entry point details and argument counts no longer appear
1035 in backtraces unless explicitly requested by setting
1036 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1037 * built-in and standard functions no longer have names like "top
1038 level local call to FOO".
1039 * fixed bug 32: functions defined in non-null lexical environments
1040 now have more legible printed representation
1041 * fixed bug 33: functions defined in non-null lexical environemnts
1042 are now more amenable to inspection by INSPECT.
1043 * workaround for bug 354: XEPs no longer appear in backtraces unless
1044 explicitly requested.
1045 * fixed bug: receiving the signal which results from attempting to
1046 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1047 SIGBUS on other *BSDs) on architectures where the C stack is also
1048 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1049 to debugger confusion or wild instability. (reported by Cheuksan
1051 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1052 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1053 (reported by Lutz Euler)
1054 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1055 compiled code. (reported by Paul Dietz)
1056 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1057 specialized parameter to a method to something that is not TYPEP
1058 the specializer is now possible.
1059 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1060 face of package deletion.
1061 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1062 pathname. (reported by David Tolpin on c.l.l)
1063 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1064 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1065 STRING-STREAM and FILE-STREAM, respectively.
1066 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1067 than previously; in addition, its test suite is now run on build.
1068 (thanks to Luke Gorrie)
1069 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1070 * optimization: added a immediate representation for single-floats
1072 * fixed some bugs revealed by Paul Dietz' test suite:
1073 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1074 ** attempting to create a package with a colliding nickname causes
1075 correctable errors to be signalled.
1076 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1077 ** several x86-64 backend bugs related to sign-extension of immediate
1080 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1081 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1082 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1084 * fixed loading of multiply forward-referenced layouts.
1085 (thanks to Cheuksan Wang)
1086 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1087 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1088 (see COMPILE-FILE documentation for details of :PRINT option)
1089 * fixed bugs 19 and 317: fixed-format floating point printing is
1090 more accurate. This also fixes a bug reported by Adam Warner
1091 related to the ~@F format directive.
1092 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1094 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1095 dispatch table if the from-char is a dispatch macro character.
1096 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1097 by the wordsize no longer ignore the last word. (reported by Lutz
1099 * fixed bug in type checking of dynamic-extent variables. (reported
1101 * optimization: sequence traversal functions use their freedom to
1102 coerce function designators to functions.
1103 * optimization: code with many calls to CLOS methods specialized on
1104 CLOS classes has had redundant type checks removed.
1105 * optimization: added declarations to speed up operations that access
1106 the internal character database (for example STRING-UPCASE)
1107 * optimization: comparison operations between floats and sufficiently small
1108 fixnums no longer create extra rationals
1109 * fixed some bugs related to Unicode integration:
1110 ** portions of multibyte characters at the end of buffers for
1111 character-based file input are correctly transferred to the
1112 start of the buffer at the next read.
1113 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1114 passing it through to OPEN.
1115 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1116 argument given to internal calls to OPEN.
1117 ** the built-in comment readers (introduced by character sequences
1118 ";" and "#|") are more forgiving to encoding errors; they will
1119 STYLE-WARN and then attempt to resync the stream at a character
1120 boundary. (thanks to Teemu Kalvas)
1121 * fixed some bugs in the x86-64 port:
1122 ** Negative short int return values from c-calls are sign-extended
1124 ** The stack is aligned to 16-bytes for c-calls, as required by
1125 the ABI. (thanks to Cheuksan Wang)
1126 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1127 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1128 for enums. (thanks to Vincent Arkesteijn)
1129 ** Multiple small optimizations and bugfixes for floating point
1131 * fixed some bugs revealed by Paul Dietz' test suite:
1132 ** Space, Tab, Linefeed, Return and Page have the invalid
1133 secondary constituent character trait.
1134 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1136 ** WITH-INPUT-FROM-STRING should only update the index place on
1138 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1139 works more reliably.
1140 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1141 with one-dimensional arrays with fill pointers.
1142 ** TYPE-OF failed on a complex with an integer realpart and a
1144 ** compiler failure during type inference for the code of form
1145 (IF (EQL X (THE ...)) ...) (MISC.535).
1147 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1148 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1149 platform. The port supports 61-bit fixnums, large memory spaces
1150 and reloading shared object files.
1151 * enhancement: saving cores with foreign code loaded is now
1152 supported on ppc/Darwin in addition to the previously supported
1154 * enhancement: the statistical profiler now walks deeper into the
1155 call stack for more meaningful call-graphs and accrued time
1156 reports (x86/x86-64 only). It also now reports time spent in
1158 * enhancement: it is now possible to trace most individual methods
1159 of a generic function in addition to tracing the generic function
1161 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1163 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1164 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1165 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1166 produces an error. (thanks to Vincent Arkesteijn)
1167 * bug fix: NAMESTRING on pathnames with :WILD components in their
1168 directories works correctly. (thanks to Artem V. Andreev)
1169 * fixed bug 125: compiler preserves identity of closures. (reported
1171 * bug fixed: functions with &REST arguments sometimes failed with
1172 "Undefined function" when compiled with (DEBUG 3). (reported by
1174 * bug fix: overflow during compiling of setting element of a bit
1175 vector with constant index and value. (reported by Timmy Douglas)
1176 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1177 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1178 * on x86 compiler partially supports stack allocation of dynamic-extent
1180 * GO and RETURN-FROM do not check the extent of their exit points
1181 when compiled with SAFETY 0.
1182 * fixed some bugs related to Unicode integration:
1183 ** encoding and decoding errors are now much more robustly
1184 handled; it should now be possible to recover even from invalid
1185 input or output to the terminal. (thanks to Teemu Kalvas)
1186 ** provided a first cut at implementing STRING-TO-OCTETS and
1187 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1188 ** altered the SB-MD5 contributed module slightly, changing the
1189 interface just enough for it to be supportable for builds where
1190 lisp characters are not eight bits.
1191 * fixed some bugs revealed by Paul Dietz' test suite:
1192 ** the FORMATTER-generated functions for ~V[ conditionals require
1193 the correct number of arguments.
1194 ** READ-FROM-STRING returns the mandated second value when applied
1195 to displaced strings.
1196 ** the #\Rubout and #\Backspace characters are treated as invalid
1197 constituent characters by the tokenizer.
1199 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1200 * new feature: reloading changed shared object files with
1201 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1202 * new feature: references to foreign variables and functions
1203 can now be compiled and loaded before the corresponding shared
1204 object file is loaded, as long as the foreign definitions are
1205 available at runtime.
1206 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1207 supported. (thanks to Dan Debertin)
1208 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1209 just keywords. (thanks to Vincent Arkesteijn)
1210 * fixed bug #331: structure-class instances corresponding to
1211 DEFSTRUCT forms are now created eagerly.
1212 * fixed bug #345: backtraces from calls to undefined functions work
1213 on x86 as well. Related bug #61 is now also partially fixed on x86
1214 (backtraces from throws to unknown catch tags.)
1215 * bug fix: lambda-list parsing is now stricter vrt. order and number
1216 of lambda-list keywords.
1217 * bug fix: as specified by AMOP, an error is signalled if a
1218 class-option appears multiple times in a DEFCLASS form. (reported
1220 * bug fix: RANDOM can be compiled when the compiler derives the type
1221 of its numeric argument as a disjoint set of small integers.
1222 (reported by Paul Dietz)
1223 * bug fix: starting a core saved with shared objects loaded when
1224 those objects are not available no longer causes threaded SBCL to
1225 hang. (reported by Sean Ross)
1226 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1228 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1229 argument, which must always be NIL. (reported by Kalle Niemitalo)
1230 * bug fix: printing 1.0d+23 no longer results in an error.
1231 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1232 * bug fix: structure slot setters preserve evaluation order. (thanks
1234 * fixed some bugs related to Unicode integration:
1235 ** RUN-PROGRAM can allow its child to take input from a Lisp
1236 stream. (reported by Stefan Scholl)
1237 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1238 locales. (reported by Ken Causey)
1239 * fixed some bugs revealed by Paul Dietz' test suite:
1240 ** INCF, DECF and REMF evaluate their place form as specified in
1242 ** FORMATTER expands ~{ ~} iteration directives with V or #
1243 parameters correctly.
1244 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1245 consequent uses no arguments correctly.
1246 ** the system has a partial understanding of the (COMPLEX RATIO)
1248 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1251 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1252 * new feature: a build-time option (controlled by the :SB-UNICODE
1253 keyword feature, enabled by default) for building the system with
1254 support for the entire 21-bit character space defined by the
1256 * new feature: the system now has rudimentary external-format
1257 support; the primary user-visible change at this time is that
1258 characters with the high bit set (such as lower-case-e-acute) will
1259 print correctly to a terminal in a UTF-8 environment.
1260 * minor incompatible change: BASE-CHAR no longer names a class;
1261 however, CHARACTER continues to do so, as required by ANSI.
1262 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1263 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1264 has been moved to the SB-EXT package (temporarily re-exported from
1266 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1267 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1269 * fixed bug #351: better error-handlind and reporting for malformed
1271 * fixed bug #350: bignum-printing is now more memory-efficient,
1272 allowing printing of very large bignums, eg. (expt 2 10000000).
1273 (reported by Bruno Haible)
1274 * fixed bug #302: better primitive-type selection for intersection
1276 * fixed bug #308: non-graphic characters now all have names, as
1277 required. (reported by Bruno Haible)
1278 * bug fix: redefining a class with different superclasses now correctly
1279 removes it from the direct-subclasses of its previous superclasses.
1280 (reported by David Morse)
1281 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1282 new value now works. (reported by Bruno Haible)
1283 * bug fix: correct canonicalization of multiple non-standard slot
1284 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1285 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1286 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1287 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1288 now exists, an signals an error.
1289 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1290 during macroexpansion for non-iteration variables as well. (reported
1291 by Bruno Haible for CMUCL)
1292 * bug fix: Cyclic structures and unprintable objects in compiler
1293 messages no longer cause errors. (reported by Bruno Haible)
1294 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1295 and READ-FROM-STRING all now return a primary value of NIL if
1296 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1297 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1298 (reported by Bruno Haible for CMUCL)
1299 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1300 do not designate a condition. (reported by Bruno Haible for
1302 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1303 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1304 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1305 permitting resolution in favour of any of the conflicting symbols.
1306 (reported by Bruno Haible for CMUCL)
1307 * bug fix: EQUAL compiler optimizations is less aggressive on
1308 strings which can potentially compare true despite having distinct
1309 specialized array element types.
1310 * bug fix: unit enumerations can be defined without dividing by
1311 zero. (thanks to Vincent Arkesteijn)
1312 * FORMAT compile-time argument count checking has been enhanced.
1313 (report from Bruno Haible for CMUCL)
1314 * a partial workaround for the bug 262: the compiler does not try to
1315 inline-expand a local function doing RETURN-FROM from a deleted
1316 BLOCK. (thanks to Peter Denno for the bug report and to David
1317 Wragg for the simple test case)
1318 * fixed some bugs revealed by Paul Dietz' test suite:
1319 ** RENAME-PACKAGE allows all package designators as new package
1321 ** constraint propagation and lambda variable substitution are
1322 more cautious in dealing with partially deleted code.
1323 ** compiler sometimes lost reoptimization passes.
1324 ** CERROR, given a condition as condition designator, passes its
1325 remaining arguments to the continue format control without
1327 ** Case-altering FORMAT directives work correctly on non-ASCII
1329 ** The REAL type specifier handles bounds outside the floating
1330 point ranges without signalling FLOATING-POINT-OVERFLOW.
1331 ** Functions with IR1-transformations can create intercomponent
1332 references to global functions.
1333 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1335 ** FORMAT ~R treats a nil value for its first parameter correctly.
1337 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1338 * enhancement: saving cores with foreign code loaded is now
1339 supported on x86/NetBSD and sparc/Linux in addition to the previously
1340 supported platforms.
1341 * bug fix: on some platforms repeated installations caused multiple
1342 copies of HTML documentation to be installed -- should not happen
1343 any more. (reported by Stefan Scholl)
1344 * bug fix: parsing self-recursive alien record types multiple times
1345 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1346 original patch by Helmut Eller for CMUCL)
1347 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1348 (thanks to Richard Kreuter)
1349 * bug fix: defining classes whose accessors are methods on existing
1350 generic functions in other (locked) packages no longer signals
1351 bogus package lock violations. (reported by François-René Rideau)
1352 * bug fix: special variables as DEFMETHOD parameters no longer have
1353 associated bogus type declarations. (reported by David Wragg and
1355 * bug fix: read-write consistency on streams of element-type
1356 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1357 * bug fix: redefiniton of the only method of a generic function with
1358 no DEFGENERIC no longer emits a full WARNING. In addition,
1359 redefinition of generic functions with no DEFGENERIC to an
1360 incompatible lambda list now signals an error. (thanks to Zach
1362 * bug fix: DEFGENERIC now works even when there's a function of the
1363 same name in an enclosing lexical environment. (thanks to Zach
1365 * fixed compiler failure, caused by instrumenting code during
1366 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1367 * optimization: added loop analysis and improved register allocation
1368 * fixed some bugs revealed by Paul Dietz' test suite:
1369 ** POSITION on displaced vectors with non-zero displacement
1370 returns the right answer.
1371 ** (SIMPLE-STRING) is a valid type specifier for sequence
1373 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1375 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1376 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1378 ** The justification version of the FORMAT ~< directive treats
1379 non-zero minpad parameter correctly.
1381 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1382 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1383 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1384 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1385 the supported interface.
1386 * new feature: Single-stepping of code compiled with DEBUG 2 or
1387 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1388 * new feature: saving cores with foreign code loaded is now
1389 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1390 Timothy Moore's work for CMUCL)
1391 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1392 parameters to * instead of NIL if no initform is supplied.
1393 (reported by Johan Bockgaard)
1394 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1395 correctly when FUNCALL appears as the car of the form. Note:
1396 despite this FUNCALL forms are not currently subject to
1397 compiler-macro expansion. (port of Raymond Toy's fix for the
1398 same from CMUCL, reported by Johan Bockgaard)
1399 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1400 (thanks for Teemu Kalvas)
1401 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1402 case instead of terminating the iteration (thanks for Julian
1403 Squires, Sean Champ and Raymond Toy)
1404 * bug fix: incorrect expansion of defgeneric that caused a style
1405 warning. (thanks for Zach Beane)
1406 * bug fix: slot accessor effective method computation works properly
1407 for classes with multiple non-standard applicable methods on
1408 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1409 * on x86 compiler supports stack allocation of results of LIST and
1410 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1411 CMUCL implementation by Gerd Moellmann)
1412 * fixed some bugs revealed by Paul Dietz' test suite:
1413 ** FORMAT strings with both the ~<~:;~> form of the justification
1414 directive and pretty-printing directives cause an error.
1416 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1417 * incompatible change: the internal functions
1418 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1419 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1420 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1421 instead of the old functions.
1422 * new feature: on platforms where "dladdr" is available foreign
1423 function names now appear in backtraces. (based on Helmut Eller's
1425 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1426 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1428 * documentation: ASDF manual is now distributed with SBCL.
1429 * bug fix: SBCL can now load its contributed modules with REQUIRE
1430 even if the system-provided entries have been removed from
1431 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1433 * bug fix: backtraces involving undefined functions or assembly
1434 routines are more informative. (thanks to Brian Downing)
1435 * bug fix: mutually referent alien structures now work correctly.
1436 (reported by Rick Taube)
1437 * bug fix: structures defined by WITH-ALIEN can be referred to
1438 within other definitions in the same WITH-ALIEN.
1439 * bug fix: division operators (MOD, TRUNCATE and the like) with
1440 constant zero divisors and integer dividends no longer generate
1442 * bug fix: provide default methods for INPUT-STREAM-P and
1443 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1444 * bug fix: improve the detection and resolution of MOP metacycles
1445 (where effective method computation depends on the generic function
1446 having its effective method computed). (reported by Bruno Haible)
1447 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1448 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1449 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1450 * bug fix: pathnames with relative directory components can be
1451 represented relative to default pathnames.
1452 * optimization: in taking the GCD of bignums, reduce the two bignums
1453 to approximately the same size (using Euclid's algorithm) before
1454 applying the more sophisticated binary GCD. (thanks to Juho
1456 * optimization: COUNT on bitvectors now operates word-at-a-time.
1457 * optimization: ASH with a positive, but not necessarily constant,
1458 (leftwards) shift, when in a modular context, is compiled to a
1460 * fixed some bugs revealed by Paul Dietz' test suite:
1461 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1462 if the corresponding argument is NIL.
1463 ** FORMAT directives accepting a minpad parameter treat negative
1465 ** literal commas as character parameters to FORMAT directives are
1467 ** literal spaces directly after ~<Newline> directives within a
1468 format-logical-block (~:< ~@:>) do not induce :FILL-style
1469 conditional newlines.
1470 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1471 as specified: it no longer includes conditional newlines).
1472 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1474 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1475 :PER-LINE-PREFIX argument does not evaluate to a string.
1477 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1478 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1479 the manual for details; for now, package locks can be disabled by
1480 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1481 if no major problems are found then it is likely that they will be
1482 compiled in unconditionally.
1483 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1484 unsupported operators on all platforms. To load a shared library
1485 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1486 object file, link it into a shared library outside of SBCL and
1487 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1489 * bug workaround: ROOM T and the GC were not working together
1490 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1491 reporting facility aren't preserved. That reporting has been
1492 disabled, so now until and unless someone figures out how to make it
1493 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1494 * minor incompatible change: as threatened around sbcl-0.8.0, the
1495 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1496 The symbols are also exported from SB-PCL for backwards
1497 compatibility, but more so than before SB-PCL should be treated as
1498 an implementation-internal package.
1499 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1501 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1502 It is known as of this release not to work on the Alpha, however.
1503 * fixed bug #167: errors signalled due to illegal syntax in method
1504 bodies are now more legible.
1505 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1506 designators and can hence be used with TYPEP.
1507 * fixed bug #333: CHECK-TYPE now ensures that the type error
1508 signalled, if any, has the right object to be accessed by
1509 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1510 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1511 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1513 * fixed bug #334: programmatic addition of slots using specialized
1514 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1515 and :ALLOCATION :CLASS slots.
1516 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1517 float exponents. (rereported by Peter Seibel)
1518 * fixed a bug: #\Space (and other whitespace characters) are no
1519 longer considered to be macro characters in standard syntax by
1520 GET-MACRO-CHARACTER.
1521 * fixed bug: initialization of condition class metaobjects no longer
1522 causes an instance of the condition to be created. (reported by
1524 * fixed bug: it is now possible to have slots such that
1525 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1526 description is neither :INSTANCE nor :CLASS.
1527 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1528 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1529 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1530 non-standard SLOT-DEFINITION-ALLOCATION values.
1531 * fixed bugs: various race conditions handling exiting threads.
1532 CL-PPCRE's thread torture test now passes.
1533 * fixed arguably-a-bug: GC time proportional to number of threads in
1534 system even when most of them are idle
1535 * optimization: improved performance of BIT and SBIT on bit-vectors.
1536 * .fasl file incompatibility: The fasl file version number has
1537 been incremented because of changes associated with package locks.
1539 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1540 * minor incompatible change: the system no longer provides
1541 optimization hints (conditions which are TYPEP
1542 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1543 mode; these hints are still emitted when higher SPEED optimization
1545 * new contrib module: a sampling profiler (profiling by statistical
1546 sampling, rather than by instrumenting functions) is available as
1547 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1548 * the behaviour of the standard function ED is now customizeable by
1549 third parties through a hook variable: see ED's documentation
1550 string for information on the protocol.
1551 * the compiler no longer emits efficiency notes for (FUNCALL X)
1552 when the type of X is uncertain under default optimization
1554 * fixed bug 276: mutating a binding of a specialized parameter to a
1555 method to something that is not TYPEP the specializer is now
1557 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1558 exhibit the required behaviour on the x86 platform. (thanks to
1559 Peter van Eynde, Eric Marsden and Bruno Haible)
1560 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1561 even for difficult arguments. (reported by Peter Graves)
1562 * fixed bug 141a: the backquote printer now descends quoted
1564 * fixed another bug in backquote printing: no more destructive
1565 modification of the form's list structure. (reported by Brian
1567 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1568 move between its address being taken and the call to
1569 interrupt_thread, fixing a crashing race condition.
1570 * the SB-POSIX contrib implementation has been adjusted so that it
1571 no longer exhibits ridiculously poor performance when constructing
1572 instances corresponding to C structs.
1574 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1575 * minor incompatible change: the sb-grovel contrib now treats C
1576 structures as alien (in the sense of SB-ALIEN) objects rather than
1577 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1578 has implications for memory management of client code
1579 (sb-grovel-returned objects must now be manually managed) and for
1580 type safety (alien objects now have full types).
1581 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1582 used to control emission of compiler diagnostics, rather than the
1583 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1584 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1585 quality should be considered deprecated.
1586 * install.sh now installs the user manual as well
1587 * (not quite a new documentable feature, but worth considering in
1588 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1589 beginnings of a semantically meaningful condition hierarchy is
1590 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1592 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1593 designator as the defaults argument.
1594 * fixed bug: Displaced arrays whose displaced-to array has become
1595 too small now cause ARRAY-DIMENSION to signal an error, providing
1596 for safer bounds-checking. (reported by Bruno Haible)
1597 * fixed bug: DEFCLASS slot definitions with identical :READER and
1598 :WRITER names now signal a reasonable error. (reported by Thomas
1600 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1602 * fixed bug: Invalid slot specification errors now print correctly.
1603 (thanks to Zach Beane)
1604 * fixed bug 320: Shared to local slot value transfers in class
1605 redefinitions now happen corrently from superclasses as
1606 well. (reported by Bruno Haible)
1607 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1609 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1610 improper lists correctly. (reported by Bruno Haible)
1611 * fixed bug 313: source-transform for <fun-name> was erroneously
1612 applied to a call of a value of a variable with name <fun-name>.
1613 (reported by Antonio Menezes Leitao)
1614 * fixed bug 307: The obsolete instance protocol ensures that
1615 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1616 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1617 remove dead unknown-values globs from the middle of the stack.
1618 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1619 easier to resume long computations after using *BREAK-ON-SIGNALS*
1620 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1621 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1622 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1623 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1624 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1625 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1627 * fixed a (fairly theoretical) bug in string printing: if
1628 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1629 does not have array-element-type equal to the most general string
1631 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1632 function name. (thanks to Nikodemus Siivola)
1633 * fixed bug: compile-time format string checker failed on
1634 non-closed ~{. (reported by Thomas F Burdick)
1635 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1636 don't warn on legal loop code involving more than one aggregate
1637 boolean. (thanks to Nikodemus Siivola)
1638 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1639 in keyword position to MAKE-PACKAGE was not being reported as
1641 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1642 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1644 * fixed bugs 280 and 312: the checking for multiple definitions in a
1645 file is less likely to become confused by uses of inline
1647 * fixed bug: the #S reader macro performs the keyword coercion
1648 specified for slot names. (reported by Kalle Niemitalo)
1649 * fixed bug: lambda lists may contain symbols whose names start with
1650 & but are not lambda-list-keywords; their occurrence triggers a
1652 * fixed bug 321: define-method-combination argument lambda lists do
1653 not cause invalid code to be generated when &optional or &aux
1654 variables have default values. (reported by Bruno Haible)
1655 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1656 have CLOS classes; this appears also to have fixed reported
1657 instability in stack exhaustion detection.
1658 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1659 MUFFLE-WARNING when no associated restart is present is now
1661 * optimization: rearranged the expansion of various defining macros
1662 so that each expands into only one top-level form in a
1663 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1665 * optimization: used a previously little-used slot in symbols to
1666 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1668 * fixed some bugs revealed by Paul Dietz' test suite:
1669 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1671 ** Fixed an optimization bug related to inheritance of initforms
1672 from local to shared slots.
1673 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1674 ** CAST optimizer forgot to flush argument derived type.
1675 ** print/read consistency on floats is now orders of magnitude
1676 more likely. (thanks also to Bruno Haible for a similar report
1678 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1679 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1680 for objects of type REAL. Make it so.
1681 ** SXHASH is specified (infelicitously) to respect similarity,
1682 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1683 it so. (thanks to Markus Ziegler)
1684 ** on the Alpha, the compiler succeeds in compiling functions
1685 returning a known number of arguments greater than 63.
1686 ** fixed handling of invalid NIL arguments in keyword position
1688 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1690 ** provided a readably-printable representation for RANDOM-STATE
1692 ** ensured that pathnames, where they have a namestring, always
1693 print using #P"..." syntax.
1695 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1696 * Support for the forthcoming 2.0 version of the NetBSD kernel
1697 running on x86 hardware has been added. (thanks to Perry
1698 E. Metzger most immediately, and others for their past work)
1699 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1700 current release version; SBCL's OpenBSD support had been broken
1701 since about the time of OpenBSD's switch to ELF binary format.)
1702 * [placeholder for DX summary]
1703 ** user code with &REST lists declared dynamic-extent, under high
1704 speed or space and low safety and debug optimization policy.
1705 * The manual has been converted to Texinfo format and the debugger
1706 chapter from the cmucl manual has been added.
1707 * A facility has been added to extract documentation strings from
1708 sbcl and store them as Texinfo-formatted snippets for inclusion in
1709 the manual (via Texinfo's @include directive)
1710 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1711 the test case to Dave Roberts)
1712 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1713 pointers. (reported by Sean Ross)
1714 * bug fix: PROFILE output is printed nicely even for large numerical
1715 values. (thanks to Zach Beane)
1716 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1717 greater than 32 handle EOF correctly.
1718 * bug fix: on X86 an immediate argument of the IMUL instruction is
1719 correctly printed in disassembly. (thanks to Lutz Euler)
1720 * bug fix: class slots in redefined classes preserve their old
1721 values. (thanks to Bruno Haible and Nikodemus Siivola)
1722 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1723 succeeds. (reported by Marco Baringer)
1724 * fixed some bugs revealed by Paul Dietz' test suite:
1725 ** READ-SEQUENCE now works on ECHO-STREAMs.
1726 ** RATIONALIZE works more according to its specification. (thanks
1729 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1730 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1731 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1732 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1733 should matter to you only if you rebind the printer control
1734 variables and then find you want different bindings in the
1735 debugger than in the ordinary execution of your program.)
1736 * The runtime build system has been tweaked to support building
1737 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1738 assembler and linker. (thanks to Nikodemus Siivola)
1739 * Unbound, undefined, undeclared variables now trigger full
1740 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1741 more often programmer error than deliberate exploitation of undefined
1743 * optimization: the hash algorithm for strings has changed to one
1744 that is less vulnerable to spurious collisions. (thanks to Juho
1746 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1747 needless bounds checking. (thanks to Juho Snellman)
1748 * optimization: implemented multiplication as a modular
1749 (UNSIGNED-BYTE 32) operation on the PPC backend.
1750 * fixed some bugs revealed by Paul Dietz' test suite:
1751 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1753 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1754 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1755 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1757 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1759 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1761 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1762 their output stream on EOF from read.
1763 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1764 have been read to end-of-file.
1765 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1767 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1768 description of determination of which consecutive characters
1770 ** Printing the "Space" character with escaping on now yields "#\\ ",
1771 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1772 ** Reading floating-point numbers with *READ-BASE* set to a number
1773 less than 10 works correctly.
1774 ** Reading floating-point numbers with *READ-BASE* set to a number
1775 more than 10 works correctly.
1776 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1777 the readtable currently in effect.
1779 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1780 * minor incompatible change: parsing of namestrings on a physical
1781 (Unix) host has changed; numbers after the final #\. in a
1782 namestring are no longer interpreted as a version field. This is
1783 intented to be largely invisible to the user, except that the
1784 meaning of the namestring "*.*.*" has changed: it now refers to a
1785 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1786 should usually be replaced by
1787 (make-pathname :name :wild :type :wild :version :wild)
1788 with the added benefit that this is more likely to be portable.
1789 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1790 option to OPEN now signals an error if the file being opened
1791 exists; this may have an impact on existing code.
1792 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1793 signals received, on the PowerPC platforms (both Linux and
1794 Darwin). (thanks to Pierre Mai for pointing out the location of
1796 * several fixes on OS X: The system now builds and runs cleanly on
1797 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1798 (thanks to Brian Mastenbrook)
1799 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1800 second-resolution: integer multiples of 1/3600 between -24 and 24.
1801 (thanks to Vincent Arkesteijn)
1802 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1803 type when called with 1 argument; PEEK-CHAR checked type of
1804 PEEK-TYPE only after having read first character from a
1805 stream. (reported by Peter Graves)
1806 * bug fix: the garbage collector now has much better locality
1807 behaviour, and in particular no longer treats all memory as being
1808 exhausted when many small objects point to each other in a deeply
1810 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1811 recognized as being TYPEP their class.
1812 * bug fix: the PUSHNEW documentation string has been corrected.
1813 (thanks to Vincent Arkesteijn)
1814 * bug fix: defaulting of the value for the last of an atypically
1815 large number of multiple values being bound was not being
1816 performed correctly on the Alpha or PPC platforms
1817 * optimization: implemented multiplication as a modular
1818 (UNSIGNED-BYTE 32) operation on the x86 backend.
1819 * optimization: SEARCH on simple-base-strings can now be open-coded.
1820 (see also contrib/compiler-extras.lisp for inspiration for
1821 teaching the compiler about the Boyer-Moore algorithm).
1822 * value, returned by MAX (and MIN) called with several EQUALP, but
1823 not EQL, arguments now does not depend on compiler settings.
1824 * fixed some bugs revealed by Paul Dietz' test suite:
1825 ** in stack analysis liveness information is propagated from
1826 non-local entry points.
1827 ** pathwise CAST removing failed when the CAST node did not start
1829 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1830 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1832 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1833 host is already defined.
1834 ** RENAME-FILE works on streams instead of signalling an internal
1836 ** PEEK-CHAR uses the current readtable when determining whether
1837 or not a character is whitespace.
1838 ** MERGE-PATHNAMES handles the case when the pathname does not
1839 specify a name while the default-pathname specifies a version.
1840 ** Pathnames now stand a better chance of respecting print/read
1842 ** Attempting to use standardized file system operators with a
1843 pathname with invalid :DIRECTORY components signals a
1845 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1847 ** DIRECTORY on logical pathnames is more correct.
1848 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1849 signal a TYPE-ERROR if their argument is not a stream.
1850 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1851 designator argument does not designate a stream.
1852 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1853 examining the synonym.
1854 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1856 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1857 element-type larger than ([UN]SIGNED-BYTE 32).
1859 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1860 * When built with the :SB-FUTEX feature, threaded builds now take
1861 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1862 for faster/more reliable mutex and condition variable support.
1863 * Incompatible change (but one you probably shouldn't have been using
1864 anyway): the interface and code for arbitrating between multiple
1865 threads in the same user session has been redesigned.
1866 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1867 error when called without an explicit environment argument.
1868 (thanks to Nikodemus Siivola)
1869 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1870 confused about their position. (thanks to Adam Warner and Gerd
1872 * bug fix: Pretty printing backquoted forms with unquotations in the
1873 argument list position of various code constructs such as LAMBDA
1874 now works correctly. (reported by Paul Dietz)
1875 * bug fix: Pretty printing unquotations no longer loses all
1876 stream position information.
1877 * optimization: performance of string output streams is now less
1878 poor for multiple small sequence writes.
1879 * optimization: performance of CSUBTYPEP in the presence of complex
1880 expressions involving CONS and NOT many times has been improved.
1881 (reported by Paul Dietz)
1882 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1884 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1885 streams for the REPL, for the debugger, and so on. (thanks to
1887 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1888 * fixed some bugs revealed by Paul Dietz' test suite:
1889 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1890 is ignored if it is too small, rather than propagating through
1891 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1892 ** extremely complex negations of CONS types were not being
1893 sufficiently canonicalized, leading to inconsistencies in
1895 ** VALUES tranformer lost derived type.
1897 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1898 * fixed a bootstrapping bug: the build process no longer assumes
1899 that the various BOOLE-related constants have the same value in
1900 host and target lisps. (noted by Paul Dietz' test suite on an
1901 SBCL binary built from CLISP)
1902 * The system can now be dynamically linked on the MIPS platform,
1903 which enables dynamic loading of foreign code from Lisp. (thanks
1904 to Ralf Baechle for discussions on the MIPS ABI)
1905 * The system now records debugging information for its own source
1906 files in a filesystem-position-independent manner, relative to
1907 the "SYS" logical host.
1908 * fixed a compiler bug: MV-LET convertion did not check references
1909 to the "max args" entry point. (reported by Brian Downing)
1910 * tweaked disassembly notes to be less confident about proclaiming
1911 some instruction as an LRA. (thanks to Brian Downing)
1912 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1913 now each have their own history, command character, and other
1914 characteristics. (thanks to David Lichteblau)
1915 * fixed some bugs revealed by Paul Dietz' test suite:
1916 ** compiler failure in compiling LOGAND expressions including a
1918 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1919 shift greater than 32.
1920 ** FLUSH-DEST did not mark blocks for type check regeneration.
1921 ** HANDLER-CASE failed to accept declarations in handler clauses
1922 in some circumstances.
1924 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1925 * New code in contrib/sb-introspect (still probably not entirely
1926 stable yet) provides some support for smart Lisp development
1927 environments like SLIME.
1928 * The conditions signalled for errors occurring when loading .fasl
1929 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1930 in a way which should help ASDF recover gracefully.
1931 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1932 stuff has been cleaned up. If you code contrib/ stuff, this might
1933 affect you, and you can look at contrib/README, contrib/STANDARDS,
1934 and/or the 0.8.4.27 diff to check.
1935 * In full calls the compiler now does not generate checks for declared
1936 argument types for all arguments.
1937 * various threading fixes
1938 ** and some experimental patches which didn't make it into
1939 the main tree for this release, but which are shipped in
1940 contrib/experimental-thread.patch as a possible fix for some
1941 failures (deadlock, spinning...) in GC-intensive multithreaded
1943 * fixed PPC build problem (source code incompatibility of different
1944 library versions): added offsetof() hackery which attempts to divine
1945 where glibc maintainers put uc_mcontext today
1946 * fixed bug 282: compiler does not trust type assertions while passing
1947 arguments to a full call.
1948 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1949 &OPTIONAL VALUES type parameter.
1950 * fix bug 214: algorithm for noting rejected templates is now more
1951 similar to that of template seletion. (also reported by rydis on
1953 * fixed bug 141b: printing backquoted information readably and prettily
1954 inserts a space where necessary.
1955 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1956 (CHARACTER 10) are now reported as errors, rather than propagated
1957 as unknown types. (reported by piso on #lisp)
1958 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1959 with values NIL and :ERROR. (thanks to Milan Zamazal)
1960 * fixed bug 191c: CLOS now does proper keyword argument checking as
1961 described in CLHS 7.6.5 and 7.6.5.1.
1962 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1963 longer raise an error; further, using a list as a for-as-arithmetic
1964 counter now raises a meaningful error.
1965 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1966 now understood by sequence creation functions such as MAKE-SEQUENCE
1968 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1969 a STRING-INPUT-STREAM.
1970 * compiler enhancement: SIGNUM is now better able to derive the type
1972 * type declarations inside WITH-SLOTS are checked. (reported by
1974 * fixed some bugs revealed by Paul Dietz' test suite:
1975 ** incorrect optimization of TRUNCATE for a positive first
1976 argument and negative second.
1977 ** compiler failure in let-convertion during flushing dead code.
1978 ** compiler failure while deriving type of TRUNCATE on an
1979 interval, containing 0.
1980 ** ASH of a negative bignum by a negative bignum count now returns
1982 ** intersection of CONS types now canonicalizes properly, fixing
1983 inconsistencies in SUBTYPEP.
1985 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1986 * incompatible change: The --disable-debugger command line
1987 option now clobbers the debugger at a more fundamental
1988 level, by redefining #'INVOKE-DEBUGGER instead of by
1989 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1990 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1991 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1992 enter the debugger REPL and then suffer recursive errors
1993 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1994 while under the new system, BREAK in --disable-debugger mode
1995 terminates the system just as an unhandled error would.
1996 * fixed compiler performance when processing loops with a step >1;
1997 * bug fix: DOCUMENTATION now retrieves generic function
1998 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1999 support has been systematized, and now supports the methods
2000 specified by ANSI, along with a default method and a method for
2001 slot documentation. (reported by Nathan Froyd)
2002 * bug fix: effective methods associated with a generic function are
2003 no longer cached over a change of that generic function's method
2004 combination. (reported by Andreas Fuchs)
2005 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2006 names a non-existent directory. (thanks to Andreas Fuchs)
2007 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2008 the PPC platform, lead to stack corruption; this has been fixed.
2009 (reported by Rainer Joswig)
2010 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2011 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2012 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2013 arithmetic in general, is now much more reliable on the PPC
2015 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2016 platform now returns the right answer.
2017 * optimization: restored some effective method precomputation in
2018 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2019 precomputation is now tunable.
2020 * optimization: compiler-internal data structure use has been
2021 reviewed, and changes have been made that should improve the
2022 performance of the compiler by about 20%.
2023 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2024 simple-base-strings and simple-bit-vectors is improved.
2025 * optimization: the optimization of 32-bit logical and arithmetic
2026 functions introduced in version 0.8.3 on the x86 has been
2027 implemented on the mips, ppc and sparc platforms; an
2028 implementation of the same facility, but for 64-bit arithmetic,
2029 has been added for the alpha.
2030 * microoptimization: the compiler is better able to make use of the
2031 x86 LEA instruction for multiplication by constants.
2032 * bug fix: in some situations compiler did not report usage of
2033 generic arithmetic in (SPEED 3) policy.
2034 * bug 145b fix: compiler used wrong type specifier while converting
2035 MEMBER-types to numeric.
2036 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2038 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2039 index argument now works properly on the Alpha platform.
2040 * bug fix: floating point exception treatment on the Alpha platform
2042 * bug fix: FILE-POSITION works much better on string input and
2043 output streams. (thanks to Nikodemus Siivola)
2044 * bug fix: many threading/garbage collection symptoms sorted.
2045 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2046 might be pseudo-atomic.
2047 * internal change: Stopping for GC is now done with signals not
2048 ptrace. GC is now done in whichever thread wanted it, instead of
2050 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2052 * bug fix: result form in DO is not contained in the implicit
2054 * incompatible change: ICR structure is changed; the value part of
2055 CONTINUATION is now called LVAR; corresponding functions are
2056 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2057 * added type deriver for ISQRT (thanks to Robert E. Brown).
2058 * bug fix: better support for loading from the command line when an
2059 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2061 * fixed some bugs revealed by Paul Dietz' test suite:
2062 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2063 ** ROUND and FROUND now give the right answer when given very
2064 small float arguments.
2065 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2067 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2068 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2069 ** LCM should return a non-negative integer.
2070 ** PARSE-INTEGER returned the index of a terminator instead of the
2071 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2072 ** PARSE-INTEGER returned an incorrect index being applied to a
2074 ** LCM with two arguments of 0 returns 0 rather than signalling
2076 ** unsigned addition of a 32-bit constant with the high bit set no
2077 longer causes an internal compiler error on the x86.
2078 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2079 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2080 with negative last argument.
2081 ** byte specifiers with zero size and position no longer cause
2082 an error during type derivation.
2083 ** bignum multiplication on the Alpha platform now returns the
2085 * porting: The system now builds on SuSE AMD64, although it still
2086 generates a 32-bit binary.
2087 * .fasl file incompatibility: The fasl file version number has
2088 been incremented (because of the changes to internal compiler
2089 data structures referred to above).
2091 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2092 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2093 more accurately, on the Darwin kernel running on PowerPC hardware.
2094 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2095 * Compiler code deletion notes now signal a condition of type
2096 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2097 an associated MUFFLE-WARNING restart.
2098 * The compiler now performs limited argument count validation of
2099 constant format strings in FORMAT, and where appropriate in ERROR,
2100 CERROR and WARN. (thanks to Gerd Moellmann)
2101 * New ASDF-INSTALL contrib can be used for automatic download and
2102 installation of third-party Lisp code from CCLAN or other sites
2104 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2105 which forces another thread to execute a function supplied by the
2107 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2108 now accept and act upon their :ELEMENT-TYPE keyword argument.
2109 (reported by Martin Atzmueller, Edi Weitz)
2110 * bug fix: FILE-POSITION now accepts position designators up to
2111 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2112 is the greater. (thanks to Patrik Nordebo)
2113 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2114 Kalle Olavi Niemitalo)
2115 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2116 printable. (reported by Eric Marsden)
2117 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2118 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2119 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2120 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2121 (thanks to Stig Sandoe)
2122 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2123 UNREAD-CHAR now correctly includes the unread character in the
2124 target sequence. (thanks to Gerd Moellmann)
2125 * bug fix in threaded builds: the system can now be suspended and
2126 resumed by shell job control with minimal disruption.
2127 * bug fixes in times and timezones >2038AD
2128 * better handling of "where is GNU make?" problem in build scripts
2129 (thanks to Nikodemus Siivola)
2130 * new optimization: inside a named function any reference to a
2131 function with the same name is considered to be a self-reference;
2132 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2133 optimization quality.
2134 * new optimization on x86: logical functions and + now have
2135 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2136 used when the result is truncated to 32 bits.
2137 * VALUES declaration is partially enabled.
2138 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2139 * fixed some bugs revealed by Paul Dietz' test suite:
2140 ** The system now obeys the constraint imposed by
2141 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2142 types form a lattice under type intersection.
2143 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2144 ** ASSOC now ignores NIL elements in an alist.
2145 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2146 and (1+ MOST-POSITIVE-FIXNUM) answers.
2147 ** The addition of a method with invalid qualifiers to a generic
2148 function does not cause an error to be signalled immediately;
2149 a warning is signalled, and the error is generated only on
2150 calling the generic function.
2151 * changed .fasl file version number, in part to add type codes for
2152 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2153 obscure ANSI requirements
2155 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2156 * fixed bug 148: failure to inline-expand a local function left
2157 garbage, confusing the compiler.
2158 * fixed bugs 3cd: structure slot readers perform type check if the
2159 slot can have an invalid value (i.e. it is either not initialized
2160 or can be written with a less specific slot writer).
2161 * bug fix: the compiler now traps array references to elements off
2162 the end of an array; previously, the bounds checking in some
2163 circumstances could go off-by-one.
2164 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2166 * type declarations for array element types now obey the description
2167 on the CLHS page "Declaration TYPE", as per discussions on
2168 sbcl-help around 2003-05-08. This means that a declaration
2169 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2170 declaration, all references to BAR will be asserted or assumed
2171 (with THE, so dependent on compiler policy) to involve objects of
2172 type FOO. Note that no such declaration is implied in
2173 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2174 * declared types of functions from the "Conditions"
2175 chapter. (reported by Paul Dietz)
2176 * bug fix: CERROR accepts a function as its first argument.
2177 * bug fix: NTH an NTHCDR accept a bignum as index
2178 arguments. (reported by Adam Warner)
2179 * optimization: character compare routines now optimize comparing
2180 against a constant character. (reported by Gilbert Baumann)
2181 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2182 argument now works properly on the MIPS platform.
2183 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2184 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2185 declared to have a functional type, does not cause an error
2187 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2188 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2189 * fixed bug: if last continuation of a deleted block has a
2190 destination, this destination should be deleted too. (reported by
2192 * fixed a bug in the bootstrap process: the host compiler's values
2193 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2194 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2195 test case from Patrik Nordebo)
2196 * improved the ability of the disassembler on the PPC platform to
2197 provide helpful disassembly notes.
2198 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2199 the class in more cases than previously.
2200 * bug fix: FILE-POSITION now understands :START and :END for
2201 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2202 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2203 8) data. (thanks to David Lichteblau)
2204 * bug fix: it is possible to add a method to a generic function
2205 without lambda list.
2206 * bug fix: reader failed to signal END-OF-FILE inside an
2207 object representation. (reported by Nikodemus Siivola)
2208 * fixed some bugs revealed by Paul Dietz' test suite:
2209 ** LAST and [N]BUTLAST should accept a bignum.
2210 ** condition slot accessors are methods.
2211 ** (VECTOR NIL) is a subtype of STRING.
2213 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2214 * minor incompatible change: some nonsensical specialized lambda
2215 lists (used in DEFMETHOD) which were previously ignored now signal
2217 * minor incompatible change: the system is now aware of the types of
2218 variables in the COMMON-LISP package, and will signal errors for
2219 most violations of these type constraints (where previously they
2220 were silently accepted).
2221 * minor incompatible change: COMPILE-FILE now uses the freedom
2222 afforded (ANSI 3.2.2.3) to use derived function types for
2223 functions defined in the same file. This also permits the system
2224 to warn on static type mismatches and function
2225 redefinition. (Currently it does not work with high DEBUG level.)
2226 * minor incompatible change: VALUES declaration is disabled.
2227 * When issuing notes, the compiler now signals a condition of type
2228 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2229 restart for use in user handlers. It is expected that the
2230 COMPILER-NOTE condition will eventually become a condition
2231 supertype to a hierarchy of note types, which will then be
2232 handleable in a similar fashion. However, at the moment, no such
2233 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2234 but it's an implementation detail, not a classification for the
2236 * Changes in type checking closed the following bugs:
2237 ** type checking of unused values (192b, 194d, 203);
2238 ** template selection based on unsafe type assertions (192c, 236);
2239 ** type checking in branches (194bc).
2240 * A short form of VALUES type specifier has ANSI meaning (it has
2241 increased the number of situations when SBCL cannot perform type
2243 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2244 keywords or constants is permissible.
2245 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2246 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2247 classes. (thanks to Antonio Martinez)
2248 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2249 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2250 argument to a pretty-print stream. Instead, it leaves any such
2251 operations to DESCRIBE-OBJECT methods.
2252 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2254 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2255 lambda lists. (thanks to David Lichteblau)
2256 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2257 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2258 (thanks to Gerd Moellmann)
2259 * fixed bug 246: increased compilation speed of long
2260 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2262 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2264 * DEFCONSTANT now throws a condition of type
2265 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2266 constant to a non-EQL value; CONTINUE and ABORT restarts
2267 respectively change and preserve the value.
2268 * fixed bug 63: The code walker, part of the implementation of CLOS,
2269 is now better at handling symbol macros.
2270 * bug fix: There is no longer an internal implementation type named
2271 CL:LENGTH. (reported by Raymond Toy)
2272 * bug fix: In macro-like defining macros/special operators the
2273 implicit block does not enclose the lambda list.
2274 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2275 enclosed in parentheses) are not suitable as type specifiers, and
2276 their use properly signals an error now.
2277 * bug fix: An argument count mismatch for a type specifier in code
2278 being compiled no longer causes an unhandled error at compile
2279 time, but signals a compile-time warning.
2280 * fixed simple vector readable printing
2281 * bug fix: DESCRIBE takes more care over whether the class
2282 precedence list slot of a class is bound before accessing it.
2283 (reported by Markus Krummenacker)
2284 * bug fix: FORMATTER can successfully compile pretty-printer format
2285 strings which use variants of the ~* directive inside.
2286 * bug fix: SEARCH now applies its TEST predicate to the elements of
2287 the arguments in the correct order. (thanks to Wolfhard Buss)
2288 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2289 in type inference. (thanks to Robert E. Brown)
2290 * bug fix: Reading in symbols with an explicit package name of ""
2291 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2292 * fixed some bugs revealed by Paul Dietz' test suite:
2293 ** NIL is now allowed as a structure slot name.
2294 ** Arbitrary numbers, not just REALs, are allowed in certain
2295 circumstances in LOOP for-as-arithmetic clauses.
2296 ** Multiple class redefinitions before slot access no longer
2297 causes a type error.
2298 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2299 association between the name and a class.
2300 ** Generic functions with non-standard method-combination and over
2301 five methods all of which return constants no longer return NIL
2302 after the first few invocations. (thanks to Gerd Moellmann)
2303 ** CALL-NEXT-METHOD with no arguments now passes the original
2304 values of the arguments, even in the presence of assignment.
2305 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2307 ** DISASSEMBLE works with closures and funcallable instances.
2308 ** ADD-METHOD now returns the generic function, not the new method.
2309 ** FIND-METHOD signals an error if the lengths of the specializers
2310 is incompatible with the generic function, even if the ERRORP
2312 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2313 which its argument is a member.
2314 ** DEFCLASS only redefines the class named by its class-name
2315 argument if that name is the proper name of the class;
2316 otherwise, it creates a new class.
2317 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2318 of the UNBOUND-SLOT condition to the name of the slot.
2319 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2320 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2321 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2322 SLOT-MAKUNBOUND in the specified fashion.
2324 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2325 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2326 cross-compilation host. As a consequence, we can now bootstrap our
2327 way up to SBCL starting with a bare gcc toolchain and human-readable
2328 source code (first the source to CLISP, then the source to SBCL).
2329 * A contributed module containing a partial implementation of the
2330 simple-streams interface has been included. (thanks to Rudi
2332 * A contributed module implementing the RFC1321 Message Digest
2333 Algorithm, known as MD5, has been included.
2334 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2335 no longer has any effect, as the code controlled by this feature
2336 has been deleted. (As far as we know, no-one has ever built using
2337 this feature, and its semantics were confused in any case).
2338 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2339 work on conditions (as required by the ANSI specification),
2340 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2341 expected behaviour on conditions. Users should note, however,
2342 that such behaviour is not required by the ANSI specification,
2343 and so use of this behaviour may render their code unportable.
2344 * fixed some bugs revealed by Paul Dietz' test suite:
2345 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2347 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2348 ** SXHASH on condition objects no longer returns NIL.
2349 ** :ALLOCATION :CLASS slots are better treated; their values are
2350 updated on class redefinition, and initforms inherited from
2351 superclasses are applied.
2352 ** REMOVE-METHOD returns its generic function argument even when
2353 no method was removed.
2354 ** SHARED-INITIALIZE now initializes the values of the requested
2355 slots, including those with :ALLOCATION :CLASS.
2356 ** ALLOCATE-INSTANCE now works on structure classes defined via
2357 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2359 ** SLOT-EXISTS-P now works on conditions, as well as structures
2361 ** MAKE-LOAD-FORM now has the required methods on
2362 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2363 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2364 keyword, and now implements the SLOT-NAMES argument.
2365 ** methods with &OPTIONAL arguments no longer allow too many
2366 arguments to be passed in the call without error.
2367 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2368 option is consistent with the required arguments of the generic
2369 function lambda list.
2370 * bug fix: REQUIRE accepts a string designator. (Thanks to
2372 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2373 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2374 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2375 * bug fix: The compiler now checks for duplicated variables in macro
2377 * bug fix: SETQ on globals returns the correct value.
2378 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2379 not a COMPILER-ERROR (followed by some other strange error on
2380 choosing the CONTINUE restart).
2381 * bug fix: make.sh and friends are now more consistent in the way that
2382 they look for GNU "make".
2384 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2385 * experimental native threads support (on x86 Linux >=2.4 only).
2386 This is not compiled in by default: you need to add :SB-THREAD to
2387 the target features. See the "Beyond ANSI" chapter of the manual
2389 * fix for longstanding nonANSIism: The old distinction between
2390 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2391 The return value from CL:FIND-CLASS is now a CLOS class, and
2392 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2393 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2394 * An interface to the MetaObject Protocol, as described in Kiczales,
2395 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2396 MIT Press, 1991, is available from the SB-MOP package.
2397 * incompatible change: the SB-PCL package should now be considered
2398 a private implementation detail, and no longer a semi-private MOP
2400 * minor incompatible change: due to rearrangement for threads, the
2401 control stack and binding stack are now allocated at arbitrary
2402 addresses instead of being hardcoded per-port. Users affected by
2403 this probably have to be doing advanced things with shared
2404 libraries, and will know who they are.
2405 * minor incompatible change: Previously, all --eval forms used were
2406 processed with READ before any of them were processed with EVAL.
2407 Now each --eval form is processed with both READ and EVAL before
2408 the next --eval form is processed. (Thus package operations like
2409 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2410 work as the user might reasonably expect.)
2411 * minor incompatible change: *STANDARD-INPUT* is now only an
2412 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2414 * minor incompatible change: Y-OR-N-P is now character-oriented,
2415 not line oriented. Also, YES-OR-NO-P now works without errors.
2416 (thanks to Antonio Martinez)
2417 * sb-aclrepl module improvements: an integrated inspector, added
2418 repl features, and a bug fix to :trace command.
2419 * Known functions, which cannot be open coded by the backend, are
2420 considered to be able to check types of their arguments. (fixing
2421 a bug report by Nathan J. Froyd)
2422 * fixed a bug in computing method discriminating functions: It is
2423 now possible to define methods specialized on classes which have
2424 forward-referenced superclasses. (thanks to Gerd Moellmann)
2425 * fixed evaluation order in optional entries (reported by Gilbert
2427 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2428 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2429 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2430 required-by-AMOP NAME argument, as well as CLASS and
2431 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2432 * fixed bug 20: DEFMETHOD can define methods using names that are
2433 not the proper names of classes to designate class specializers.
2434 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2435 Unix file descriptors, instead of blowing up. (thanks to Antonio
2437 * Garbage collection refactoring: user-visible change is that a
2438 call to the GC function during WITHOUT-GCING will not do garbage
2439 collection until the end of the WITHOUT-GCING. If you were doing
2440 this you were probably losing anyway.
2441 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2442 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2443 TYPEP the latter but not the former.
2444 * The compiler issues a full WARNING for calls to undefined functions
2445 with names from the CL package.
2446 * MAP-INTO for a vector destination is open coded. (reported by
2447 Brian Downing on c.l.l)
2448 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2449 documentation string.
2450 * fixed some bugs revealed by Paul Dietz' test suite:
2451 ** COPY-ALIST now signals an error if its argument is a dotted
2453 ** Condition slots are now accessed more correctly in the presence
2454 of multiple initargs for a given slot.
2455 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2456 exclude restarts of the same name associated with a different
2458 ** DEFCLASS of forward-referenced classes with another
2459 forward-referenced class in the superclasses list no longer
2461 ** Condition slots are now initialized once each, not multiple
2462 times. (thanks to Gerd Moellmann)
2463 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2465 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2466 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2467 arguments contain duplicated elements.
2468 ** RESTART-CASE understands local macros.
2469 ** RESTART-CASE associates exactly its own restarts with a condition.
2470 ** ENDP in safe mode checks its argument to be of type LIST.
2471 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2472 in question is unbound.
2473 ** Optimized MAKE-INSTANCE functions no longer cause internal
2474 assertion failures in the presence of duplicate initargs.
2475 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2476 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2477 propagated even in the ctor optimized implementation of
2479 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2481 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2482 * a better implementation of SXHASH on (simple) bit vectors,
2483 measured both in execution speed and in distribution of results
2484 over the positive fixnums, has been installed. Likewise, a better
2485 implementation of EQUAL for simple bit vectors is now available.
2486 * fixed CEILING optimization for a divisor of form 2^k.
2487 * fixed bug 240 (emitting extra style warnings "using the lexical
2488 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2489 by Antonio Martinez)
2490 * fixed SXHASH, giving different results for NIL depending on type
2491 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2492 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2493 variable. (found by Rolf Wester)
2494 * SBCL does not ignore type declarations for special
2495 variables. (reported by rif on c.l.l 2003-03-05)
2496 * some bug fixes in contrib/sb-aclrepl/
2497 * fixed some bugs revealed by Paul Dietz' test suite:
2498 ** a bug in the CONS type specifier, whereby the CAR and CDR
2499 types got intertwined, has been fixed;
2500 ** the type system is now able to reason about the interaction
2501 between INTEGER and RATIO types more completely;
2502 ** APPEND, [N]REVERSE and NRECONC check that those their
2503 arguments, which must be proper lists, are really so;
2504 ** An array specialized to be unable to hold elements has been
2505 implemented, as required -- yes, really -- by ANSI;
2506 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2507 on malformed property lists;
2509 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2510 * incompatible packaging change: in line with Unix convention,
2511 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2512 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2513 * REQUIRE and PROVIDE are now optionally capable of doing something
2514 useful. See the documentation string for REQUIRE.
2515 * infrastructure for a managed SBCL contrib system: contributed
2516 modules in this release include:
2517 ** the ASDF system definition facility;
2518 ** an interface to the BSD Sockets API;
2519 ** an ACL-like convenience interface to the repl;
2520 (thanks to Kevin Rosenberg)
2521 ** an implementation of ROTATE-BYTE, with efficient implementation
2523 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2524 gives a closer approximation to the right answer than previously.
2525 (thanks to Raymond Toy)
2526 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2527 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2528 optional environment arguments, as required by ANSI.
2529 * fixed bugs in other functions taking environment objects, allowing
2530 calls with an explicit NIL environment argument to be compiled
2532 * fixed bug 228: primary return values from
2533 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2534 COMPILE or FUNCTION.
2535 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2536 :TYPE will no longer signal errors on innocuous objects.
2537 * fixed bug 231b: SETQ is better at respecting type declarations in
2538 the lexical environment.
2539 * fixed a bug in DEFCLASS: classes named by symbols with no or
2540 unprintable packages can now be defined.
2541 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2542 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2543 * fixed compiler failure related to checking types of functions.
2544 (reported by Robert E. Brown)
2545 * the compiler is now much more consistent in its error-checking
2546 treatment of bounding index arguments to sequence functions: in
2547 (SAFETY 3) code, errors will be signalled in almost all cases if
2548 invalid sequence bounding indices are passed to functions defined
2549 by ANSI to operate on sequences.
2550 * fixed a bug in the build procedure: documentation of SBCL-specific
2551 packages is now preserved and available in the final Lisp image.
2552 * lifted FDEFINITION lookup out of loops in the implementation of
2553 many list operations. (thanks to Robert E. Brown)
2554 * fixed a bug in the reader: the #n# reader macro now works for
2555 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2556 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2557 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2558 E. Brown and Tony Martinez respectively)
2559 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2560 SLOT-BOUNDP: in particular, it is now able to optimize them much
2561 better, and is now not vulnerable to having packages renamed.
2562 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2563 always called when a slot is not present in an instance. (thanks
2565 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2566 longer signal an error on unreading a character following EOF on
2567 the previous constituent stream. (thanks to Tony Martinez)
2568 * fixed some bugs revealed by Paul Dietz' test suite:
2569 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2570 not just nonnegative fixnums;
2571 ** the logical bit-array operators such as BIT-AND now accept an
2572 explicit NIL for their "opt-arg" argument (to indicate a
2573 freshly-consed result bit-array);
2574 ** ELT now signals an error on an invalid sequence index in safe
2576 ** the type system is now cleverer about negations of numeric
2577 types, and consequently understands the BIGNUM and RATIO types
2579 ** the type system is now cleverer about the interaction between
2580 INTEGER and RATIO types: while bugs still remain, many more
2581 cases are accurately computed;
2582 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2583 if it is in the last clause;
2584 ** CONSTANTLY now correctly returns a side-effect-free function in
2586 ** DECLARE is no longer treated as a special-operator; in
2587 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2588 * incremented fasl file version number due to the change in the
2589 DEFSTRUCT-SLOT-DESCRIPTION structure.
2591 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2592 * minor incompatible change: code processed by the "interpreter" or
2593 EVAL now has a compilation optimization policy of (DEBUG 2)
2594 (changed from (DEBUG 1)) to improve debuggability of interactive
2595 development, and to allow the use of the debug RETURN command in
2597 * an experimental implementation of the RETURN command for the
2598 debugger has been included. (thanks to Frederik Kuivinen)
2599 * fixed bug 62: constraints were not propagated into a loop.
2600 * fixed bug in embedded calls of SORT (reported and investigated by
2602 * fixed some bugs revealed by Paul Dietz' test suite:
2603 ** printing and reading of arrays with some dimensions having
2604 length 0 (thanks to Gerd Moellmann);
2605 ** BOA constructor with &AUX argument without a default value does
2606 not cause a type error;
2607 ** CONSTANTP now returns true for all self-evaluating objects.
2609 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2610 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2611 accessors that are related by inheritance, as specified in the
2612 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2613 Valtteri Vuorikoski)
2614 * The compiler is now able to inline functions that were defined in
2615 a complex lexical environment (e.g. inside a MACROLET).
2616 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2618 * Support for the upcoming FreeBSD-5.0 release has been included.
2619 (thanks to Dag-Erling Smorgrav)
2620 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2621 effect when it is not in a toplevel context.
2622 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2623 stand a better chance of being correct. (thanks to Gerd
2625 * fixed bug in COERCE, which now signals an error on coercing a
2626 rational to a bounded real type which excludes the expected
2628 * The compiler is now able to derive types more accurately from the
2629 COERCE and COMPILE functions.
2630 * fixed bug 223: functional binding is considered to be constant
2631 only for symbols in the CL package.
2632 * fixed bug 231: SETQ did not check the type of a variable being set
2633 (reported by Robert E. Brown)
2634 * A new optimization for MAKE-INSTANCE has been included, fixing
2635 various bugs (including relating to :ALLOCATION :CLASS slots and
2636 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2638 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2639 ** As required by ANSI, LOOP now disallows anonymous collection
2640 clauses such as COLLECT I in conjunction with aggregate boolean
2641 clauses such as THEREIS (= I 1);
2642 ** LOOP now signals an error when any variable is reused in the
2643 same loop (including the potentially useful construct analogous
2644 to WITH A = 1 WITH A = (1+ A);
2645 ** IT is only a special loop symbol within the first clause of a
2646 conditional loop clause;
2647 ** LOOP with a typed iteration variable over a hashtable now
2648 signals a type error iff it should.
2649 * fixed some other bugs revealed by Paul Dietz' test suite:
2650 ** FILE-STREAM now names the class previously known as FD-STREAM;
2651 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2652 argument) no longer signals an error;
2653 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2654 of the slot symbol, rather than using the current package
2655 ((:CONC-NAME "") continues to intern the slot's name in the
2657 * incremented fasl file version number, because of the incompatible
2658 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2659 of the new implementation of DEFINE-COMPILER-MACRO.
2661 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2662 * Support for building SBCL for MIPS platforms running in
2663 little-endian mode has now been checked in, and basic
2664 functionality on said platforms verified.
2665 * minor incompatible change: PCL now records the pathname of a file
2666 in which methods and the like are defined, rather than its
2668 * minor incompatible change: TRUENAME now considers the truename of
2669 a file naming a directory to be the pathname with :DIRECTORY
2670 component indicating that directory.
2671 * minor incompatible change: a NAMED clause in the extended form of
2672 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2673 reason for the previous behaviour is unclear.
2674 * more systematization and improvement of CLOS and MOP conformance
2675 in PCL (thanks to Gerd Moellman and Pierre Mai):
2676 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2678 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2679 primary methods with no specializers;
2680 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2682 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2683 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2684 FORWARD-REFERENCED-CLASSes; error reporting on
2685 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2687 ** SXHASH on CLOS instances now uses a slot internal to the
2688 instance to return different numbers on distinct instances,
2689 while preserving the same return value through invocations of
2691 ** DEFMETHOD signals errors when methods with longer incongruent
2692 lambda lists are added to generic functions;
2693 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2694 CLASS, as specified in AMOP;
2695 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2696 on the order returned by the primary method for classes of
2697 class STANDARD-CLASS;
2698 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2699 * fixed some bugs shown by Paul Dietz' test suite:
2700 ** DOLIST puts its body in TAGBODY;
2701 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2703 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2704 value producing form;
2705 ** if more variables are given to PROGV than values, extra
2706 variables are bound and made to have no value;
2707 ** NSUBSTITUTE on list arguments gets the right answer with
2709 ** ELT signals an error of type TYPE-ERROR when the index argument
2710 is not a valid sequence index;
2711 ** LOOP signals (at macroexpansion time) an error of type
2712 PROGRAM-ERROR when duplicate variable names are found;
2713 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2714 ** FUNCALL of special-operators now cause an error of type
2716 ** PSETQ now works as required in the presence of side-effecting
2717 symbol-macro places;
2718 ** NCONC accepts any object as its last argument;
2719 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2721 ** loop-for-as-package does not require a package to be explicitely
2723 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2725 * fixed bug 166: compiler preserves "there is a way to go"
2726 invariant when deleting code.
2727 * fixed bug 172: macro lambda lists with required arguments after
2728 &REST arguments now cause an error to be signalled. (thanks to
2730 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2731 bug. (thanks to Matthew Danish)
2732 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2734 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2735 arguments are lexically rebound. (thanks to Gerd Moellmann and
2737 * fixed bug 194: error messages are now more informative when there
2738 is no primary method applicable in a call to a generic
2739 function. (thanks to Gerd Moellmann)
2740 * fixed bug in command line argument checking (thanks to Julian
2742 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2743 (thanks to Matthew Danish)
2744 * incremented fasl file version number, because of the
2745 SXHASH-related changes in the layout of CLOS data structures
2747 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2748 * minor incompatible change: The runtime (the Unix executable named
2749 "sbcl") is now much pickier about the .core files it will load.
2750 Essentially it now requires .core files to descend from the same
2751 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2752 as the runtime does. (The intent is to prevent the crashes which
2753 can occur, and which can even be reported as mysterious failures,
2754 when people patch the sources or change the build parameters
2755 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2756 sbcl and .core files.)
2757 * fixed bug: VALUES-LIST is no longer optimized away.
2758 * fixed bug 142: The FFI conversion of C string values to Lisp
2759 string values no longer conses excessively. (thanks to Nathan
2760 Froyd porting Raymond Toy's fix to CMU CL)
2761 * began to systematize and improve MOP conformance in PCL (thanks to
2762 Nathan Froyd, Gerd Moellman and Pierre Mai):
2763 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2765 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2766 ** FINALIZE-INHERITANCE is now called on class finalization;
2767 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2768 argument precedence order.
2769 * fixed bug 202: The compiler no longer fails on functions whose
2770 derived types contradict their declared type.
2771 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2772 so it can be non-toplevel.
2773 * The fasl file version number has changed (because of the new
2774 implementation of DEFMACRO).
2775 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2776 safe code, that any length requirement by their type-specifier
2777 argument is valid. The exceptions to this are described in bug
2779 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2780 functions now check the types of their inputs as required by ANSI.
2781 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2782 introduced symbol is DECLAREd to be SPECIAL.
2783 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2784 * fixed bug: :COUNT argument to sequence functions may be negative
2785 * fixed bug: body of DO-SYMBOLS may contain declarations
2786 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2787 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2788 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2789 symbol macro only once
2790 * fixed printing of call frame when argument list is unavailable
2791 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2792 * compiler no longer signals WARNING on unknown keyword
2795 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2796 * A beta-quality port to the mips architecture running Linux,
2797 based on the old CMUCL backend, has been made. It has been tested
2798 on a big-endian kernel, and works sufficiently well to be able to
2799 rebuild itself; it has not been tested in little-endian mode.
2800 * fixed an inconsistency between gencgc.c and purify.c which made
2801 dumping/loading .core files unreliable
2802 * fixed bug 120a: The compiler now deals correctly with IFs where
2803 the consequent is the same as the alternative, instead of
2804 misderiving the return type. (thanks to Alexey Dejneka)
2805 * fixed bug 113: Logical pathnames are now dumpable (the logical
2806 host is resolved at load-time, throwing an error if it is not
2808 * fixed bug 174: FORMAT's error message is slightly clearer when a
2809 non-printing character is used in a format directive.
2810 * fixed several bugs in compiler checking of type declarations, i.e.
2811 violations of the Python "declarations are assertions" principle
2812 (thanks to Alexey Dejneka)
2813 * fixed several bugs in PCL's error checking (thanks to Gerd
2815 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2817 * fixed bug in compilation of functions as first class values
2818 (thanks to Antonio Martinez-Shotton)
2819 * The compiler's handling TYPE-ERRORs which it can prove will
2820 inevitably happen at runtime has been cleaned up and corrected
2821 in several ways. (thanks to Alexey Dejneka)
2822 * improved argument type checking for various basic arithmetic
2823 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2824 lavished on them in the past that they can be compiled in many
2825 ways in different special cases
2826 * fixed bug 181: compiler checks validity of user supplied type
2828 * cleaned up code flushing in optimization: Function calls which
2829 should signal errors for safety purposes (e.g. which ANSI says
2830 should signal errors when their arguments are of incorrect type)
2831 are no longer optimized away.
2832 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2833 * incremented fasl file version number, because changes in the
2834 implementation of sequence functions like COERCE caused
2835 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2836 in old inline expansions) to become undefined. (Actually these
2837 changes were later undone, so we might very well be binary
2838 compatible with 0.7.7 after all, but leaving the version number
2839 incremented seemed like the simplest and most conservative
2842 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2843 * An alpha-quality port to the parisc architecture running Linux,
2844 based on the old CMUCL backend, has been made. This, even more so
2845 than the other backends, should be considered still a work in
2846 progress; known problems include that the Linux kernel in 64-bit
2847 mode does not propagate the correct sigcontext structure to
2848 userspace, and consequently SBCL on a parisc64 kernel will not
2850 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2851 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2852 suppressed.) Also now that the compiler is looking at declarations
2853 in the environment, it checks optimization declarations as well,
2854 and suppresses inlining when (> DEBUG SPEED).
2855 * More fixes have been made to treatment of floating point exception
2856 treatment and other Unix signals. In particular, floating point
2857 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2858 * The detection and handling of control stack exhaustion (infinite
2859 or very deeply nested recursion) has changed. Stack exhaustion
2860 detection is now done by write-protecting pages at the OS level
2861 and applies at all optimization settings; when found, a
2862 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2863 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2864 be caught using IGNORE-ERRORS.
2865 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2866 correctly when transforming recursive LABELS functions to LETs.
2867 (thanks to Alexey Dejneka)
2868 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2869 that are names of constants or global variables.
2870 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2871 alien routines with docstrings.
2872 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2873 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2875 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2876 (thanks to Eric Marsden)
2877 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2878 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2879 * Bug fix: The SPARC backend can now compile functions involving
2880 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2881 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2882 to a functions expecting a PATHNAME-DESIGNATOR.
2883 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2884 lambda lists. (thanks to Alexey Dejneka)
2885 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2886 functions so that EXPORTing the name of the class doesn't cause
2887 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2888 to Antonio Martinez for reporting this)
2889 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2890 constant-folded without failing an assertion. (thanks to Einar
2891 Floystad Dorum for reporting this)
2892 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2893 types are now dealt with more correctly by the compiler.
2894 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2895 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2896 behaviour analogous to RENAME-FILE. This puts its behaviour more
2897 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2899 * The fasl file version number has changed again. (because of the
2900 bug fix involving the names of PCL MAKE-INSTANCE functions)
2902 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2903 * bug fix: Floating point exceptions are treated much more
2904 consistently on the x86/Linux and PPC/Linux platforms.
2905 * Array initialization with :INITIAL-ELEMENT is now much faster for
2906 cases when the compiler cannot open code the array creation, but
2907 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2908 array accesses have also seen a speed increase.
2909 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2910 specified. (This was at the root of some bad interactions between
2911 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2913 * bug fix: Internal error arguments for undefined functions are now
2914 computed correctly on the PPC/Linux platform.
2915 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2916 Raymond Toy's patch for CMU CL)
2917 * Support for the Solaris 9 operating environment has been included
2918 (thanks to Daniel Merritt)
2919 * A very ugly but hopefully complete draft of the missing FFI chapter
2920 of the manual has been created by reformatting the corresponding
2921 CMU CL manual chapter into (currently very ugly and incoherent)
2922 DocBook and bringing it up to date for SBCL behavior. Thus, the
2923 manual is now essentially complete, at least by my extreme
2924 once-and-only-once standards, whereby it's acceptable to refer to
2925 the doc strings of SB-EXT functions as the primary documentation.
2926 * The fasl file version number has changed again, due to cleanup of
2927 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2928 is no longer a static symbol.)
2930 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2931 * SBCL now builds with OpenMCL (version 0.12) as the
2932 cross-compilation host; also, more progress has been made toward
2933 bootstrapping under CLISP.
2934 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2936 * bug 158 fixed: The compiler can now deal with integer loop
2937 increments different from 1; fixing this turned out also to fix
2939 * bug 169 fixed: no more bogus warnings about using lexical bindings
2940 despite the presence of perfectly good SPECIAL declarations (thanks
2941 to David Lichteblau)
2942 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2943 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2944 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2946 * bug 180 fixed: Method combination specifications no longer ignore
2947 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2948 * bug fix: Structure type predicate functions now check their argument
2949 count as they should.
2950 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2951 correctly. (thanks to Pierre Mai)
2952 * minor incompatible change: The --noprogrammer option is deprecated
2953 in favor of the new --disable-debugger option, which is very similar.
2954 (The major difference is that it takes effect at a slightly different
2955 time at startup, causing handling of errors in --sysinit and
2956 --userinit files will be affected differently.) The
2957 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2958 been added to allow this functionality to be controlled from ordinary
2959 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2960 maintainers, who might want to run non-interactive scripts to
2961 build SBCL cores which will later be used interactively.)
2962 * minor incompatible change: The LOAD function no longer, when given
2963 a wild pathname to load, loads all files matching that pathname.
2964 Instead, an error of type FILE-ERROR is signalled.
2966 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2967 * bug 147 fixed: The compiler preserves its block link/count
2968 invariants more correctly now so that it doesn't crash. (thanks
2970 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2972 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2973 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2974 no-value-for-this-character as NIL (as specified by ANSI).
2975 * HOST-NAMESTRING on physical pathnames now returns a string that is
2976 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2977 (thanks to Christophe Rhodes)
2978 * The Alpha port handles icache flushing more correctly. (thanks to
2980 * More progress has been made toward bootstrapping under CLISP. (thanks
2981 to Christophe Rhodes)
2982 * The fasl file format has changed again, because dynamic loading
2983 on OpenBSD (which has non-ELF object files) motivated some cleanups
2984 in the way that foreign symbols are transformed and passed around.
2985 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2986 is no longer treated as whitespace by the reader, but instead as
2987 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2988 |A
\7fB|, instead of A as it used to.
2990 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2991 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2992 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2993 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2994 this as of 0.7.1.45, but was left out of the previous news section
2995 (thanks to Dan Barlow)
2996 * SBCL now runs on the Solaris operating system on SPARC architectures
2997 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2998 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2999 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3000 * SPARC backend cleanups, allowing builds of cores optimized for V8
3001 and V9 SPARCS, and also emission of code targeted to a particular
3002 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3004 * SBCL is closer to bootstrapping under CLISP, thanks to various
3005 fixes by Christophe Rhodes.
3006 * The fasl file format has changed again, to allow the compiler's
3007 INFO database to support symbol macros.
3008 * The user manual (in doc/) is formatted into HTML more nicely.
3009 (thanks to coreythomas)
3010 * The system is smarter about SUBTYPEP relationships, especially
3011 those involving NOT types (including types such as ATOM which are
3012 represented internally using NOT types). Thus SUBTYPEP is less
3013 likely to return (VALUES NIL NIL) in general, and in particular
3014 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3016 * The fasl file format has changed again, because the internal
3017 representation of types now includes a new slot to support the new
3018 SUBTYPEP-of-NOT-types logic.
3019 * (not a change in the main branch of SBCL, but a related prototype
3020 which can hopefully be merged into the main branch of SBCL in the
3021 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3022 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3023 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3024 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3025 dispatch macro character combinations. (thanks to Alexey Dejneka)
3026 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3027 that superficially look like logical namestrings correctly.
3028 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3030 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3031 default supplied-p) for &optional and &key arguments. (thanks to
3034 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3035 * incompatible change: The compiler is now less aggressive about
3036 tail call optimization, doing it only when (> SPACE DEBUG) or
3037 (> SPEED DEBUG). (This is an incompatible change because there are
3038 programs which relied on the old CMU-CL-style behavior to optimize
3039 away their unbounded recursion which will now die of stack overflow.)
3040 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3041 for non-GENCGC systems has been increased to 20M (since that
3042 seems much closer to the likely performance optimum for modern
3043 systems than the old 4M value was)
3044 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3045 and *DEBUG-PRINT-LEVEL*
3046 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3047 of CMU CL's support for SPARC, and various endianness and other
3048 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3049 * new syntactic sugar for the Unix command line: --load foo.bar is now
3050 an alternate notation for --eval '(load "foo.bar")'.
3052 ** The system now detects stack overflow and handles it gracefully,
3053 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3054 optimization settings. (This is a good thing in general, and
3055 its introduction in this version should be particularly timely
3056 for anyone whose code fails because of suppression of tail
3058 ** The system now hunts for the C variable "environ" in a more
3059 devious way, to avoid segfaults when the C library version differs
3060 between compile time and run time. (thanks to Christophe Rhodes)
3061 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3062 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3063 ** The compiler no longer issues bogus style warnings for undefined
3064 classes in the same source file as the DEFCLASSes which defined
3065 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3067 ** fixes in CONDITION class precedence list for undefined function
3068 errors (thanks to Alexei Dejneka)
3069 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3070 correctly. (thanks to Dan Barlow)
3071 ** portability fixes aiming at bootstrapping under CLISP (thanks
3072 to Dave McDonald and Christophe Rhodes)
3073 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3074 ** fixes in type translation and and type inference (thanks to
3076 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3077 ** various fixes in the new ports (thanks to Dan Barlow)
3078 * several changes related to debugging:
3079 ** suppression of tail recursion, as noted above
3080 ** stack overflow detection, as noted above
3081 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3082 is now the default. (For some time encapsulation has been more
3083 reliable than the breakpoint-based :ENCAPSULATE NIL
3084 implementation, at least on X86 systems; and I just noticed that
3085 encapsulation also seems closer to the spirit of the ANSI
3088 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3090 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3091 up properly again. (There was a packaging bug in 0.7.0 which
3092 left their definitions in SB-SYS::LOAD-FOREIGN and
3093 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3094 vital for most things which interface to C-level interfaces,
3095 like extensions working with sockets or databases or
3096 Perl-compatible regexes or whatever, and the need to fix
3097 this bug is the main reason that 0.7.1 was released so
3099 ** DEFGENERIC is now choosier about the methods it redefines, so that
3100 reLOADing a previously-LOADed file containing DEFGENERICs does
3101 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3102 cycle now works as it should. (thanks to Alexey Dejneka)
3103 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3104 fixed by Christophe Rhodes. (He actually submitted this patch
3105 months ago, and I delayed until after 0.7.0.)
3106 ** Bug 111 (internal compiler confusion about runtime checks on
3107 FUNCTION types) was fixed by Alexey Dejneka.
3108 * Some internal cleanups (getting rid of variables which aren't
3109 needed now that the byte interpreter is gone) caused the fasl
3110 file format number to change again.
3112 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3113 * major incompatible change: The default fasl file extension, i.e. the
3114 default extension for files produced by COMPILE-FILE, has changed
3115 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3117 ** There are many changes in the implementation of the compiler.
3118 SBCL is now essentially a compiler-only implementation of ANSI
3119 Common Lisp. EVAL still "interprets" a few special cases, but
3120 almost all the interesting cases are handled by creating
3121 a LAMBDA expression, calling COMPILE on it, then calling
3122 FUNCALL on the result.
3123 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3124 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3125 Since the code is newer, there might still be some new bugs
3126 (though not as many as before Martin Atzmueller's fixes:-). But
3127 the new code is substantially simpler and clearer, and hopefully
3128 any remaining bugs will be simpler, less fundamental, and more
3129 fixable then the bugs in the old code.
3130 ** The revised compiler is still a little unsteady on its feet.
3132 *** The debugging information it produces (particularly the names
3133 of FUNCTION objects) is sometimes much less useful than what
3134 the old compiler produced.
3135 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3136 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3137 within a MACROLET) has been temporarily weakened.
3138 ** There are new compiler optimizations for various functions:
3139 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3140 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3141 *** the math functions TRUNCATE, FLOOR, and CEILING
3142 *** the function-of-all-trades COERCE
3143 Mostly these should be transparent, but there's one
3144 potentially-annoying problem (bug 117): when the compiler
3145 inline-expands a function and does type analysis on the result,
3146 it can create control paths which have type mismatches, and
3147 when it can't prove that those control paths aren't taken,
3148 it will issue WARNINGs about the type mismatches. This is
3149 a particular problem in practice for the new sequence functions.
3150 It's not clear how this should be fixed, and for now, a
3151 workaround is given in the entry for 117 in the BUGS file.
3152 ** (Because of the interaction between the two previous items --
3153 occasional inlining problems and new inline expansions -- some
3154 of the new sequence function optimizations won't really kick in
3155 completely until debugging information, and then inlining, are
3156 straightened out in some future version.)
3157 * minor incompatible changes:
3158 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3159 DIRECTORY no longer implicitly promotes NIL slots of its
3160 pathname argument to :WILD. In particular, when you ask for the
3161 contents of a directory (which you used to be able to do without
3162 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3163 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3164 ** changes in behavior that ANSI explicitly defines to be
3165 implementation dependent:
3166 *** The new compiler-only implementation still conforms with ANSI,
3167 but acts a little different than before. Besides the obvious
3168 changes in performance tradeoffs (that the cost per form passed
3169 to EVAL has gone up, and the cost per form executed by EVAL
3170 has gone down), the behavior of the system changes a little
3171 because there are no longer any interpreted function objects.
3172 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3173 e.g. doing COMPILE on the output of interactive DEFUN is
3175 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3177 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3178 has changed from (:CL) to NIL.
3179 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3180 e.g. #\Newline and #\Tab, don't have names specified in the
3181 ANSI Common Lisp standard, is now based on their ASCII symbolic
3182 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3183 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3184 are no longer used for output.
3185 ** changes in internal implementation constants:
3186 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3187 4 million. (If your application spends a lot of time GCing and
3188 you have a lot of RAM, you might want to experiment with
3189 increasing it even more.)
3190 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3191 However, almost all old code should still continue to work without
3192 immediate update, as SB-C-CALL is now a (deprecated) nickname
3194 ** Old operator names in the style DEF-FOO are now deprecated in
3195 favor of new corresponding names DEFINE-FOO, for consistency with
3196 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3197 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3198 internal symbols, but a few supported extensions like
3199 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3200 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3201 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3202 etc. as you get deeper into recursive calls to the debugger
3203 command loop, instead of the old "5]", "5]]", "5]]]"
3204 sequence. (I was motivated to do this when squabbles between
3205 ILISP and SBCL left me very deeply nested in the debugger. In the
3206 short term, this change will probably provoke more ILISP/SBCL
3207 squabbles, but hopefully it will be an improvement in the long run.)
3208 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3209 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3210 function, since declaiming return types involving aliens is
3211 (1) annoyingly messy to do by hand and (2) vital to efficient
3212 compilation of code which calls such functions.
3213 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3214 longer reexported by the SB-EXT package. They're solely useful
3215 for alien code, so it seems more logical that you should get
3216 them from the SB-ALIEN package, not in SB-EXT.
3217 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3218 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3219 features. Instead, the code that they used to control is always
3220 built into the system.
3221 * many other bug fixes
3222 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3223 advantage of the new EVAL-WHEN stuff and to clean them up in
3224 general, and they are now more ANSI-compliant in a number of
3225 ways. Martin Atzmueller is responsible for a lot of this.
3226 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3228 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3229 *** correct ERROR type for various file operations
3230 *** some fixes for Lisp streams
3231 *** DEFMETHOD syntax checking
3232 *** changing old weird representation of debug information as
3233 strings (which, among their other deficiencies, don't transform
3234 correctly when you rename packages, and don't change their
3235 print representation when you change things like *PACKAGE*
3236 and *PRINT-LENGTH*) to symbols and lists of symbols
3237 He also made several improvements and fixed several bugs in DESCRIBE.
3238 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3240 *** misbehavior of WRITE-STRING/WRITE-LINE
3241 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3242 and several other LOOP problems as well
3243 *** DIRECTORY when similar filenames are present
3244 *** DEFGENERIC with :METHOD options
3245 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3246 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3247 *** argument ordering in FIND with :TEST option
3248 *** mishandled package designator argument in APROPOS-LIST
3249 *** various problems in the backquote readmacro
3250 *** a bug in APROPOS
3251 *** probably some others that I'm not describing very well here,
3252 since the CVS log documents them by reference to sbcl-devel
3253 messages, and the SourceForge archives aren't working well.:-(
3254 ** Dan Barlow improved the Alpha port (and is making progress on the
3255 PPC port, for those of you who think different).
3256 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3257 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3258 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3259 continues to work on the SPARC port (for those of you in a position
3260 to look down upon our little PC-compatible boxes from a great height).
3261 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3262 and uses that copy, rather than the current dynamic value, when
3263 it's trying to decide whether to truncate output. Thus e.g.
3264 (let ((*print-lines* 50))
3265 (pprint-logical-block (stream nil)
3267 (let ((*print-lines* 8))
3268 (print (aref possiblybigthings i) stream)))))
3269 should now truncate the logical block only at 50 lines, instead of
3270 often truncating it at 8 lines, as it did before.
3271 * The doc/cmucl/ directory, containing old CMU CL documentation
3272 from the time of the fork, is no longer part of the base system.
3273 SourceForge has shut down its anonymous FTP service, and with it
3274 my original plan for distributing the old CMU CL documentation
3275 there. For now, if you need these files you can download an old
3276 SBCL source release and extract them from it.
3277 * The fasl file version number changed again, for dozens of reasons,
3278 some of which are apparent above.
3280 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3281 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3282 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3283 patch, so that DISASSEMBLE output is much nicer.
3284 * The code in the SB-PROFILE package now seems reasonably stable.
3285 I still haven't decided what the final interface should look like
3286 (I'd like PROFILE to interact cleanly with TRACE, since both
3287 facilities use function encapsulation) but if you have a need
3288 for profiling now, you can probably use it successfully with
3289 the current CMU-CL-style interface.
3290 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3291 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3292 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3293 link now returns the dangling link itself, and for similar
3294 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3295 link itself. (In these cases the old code signalled an error and
3296 looped endlessly, respectively.) Thus, DIRECTORY now works even
3297 in the presence of dangling and cyclic symbolic links.
3298 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3299 is now a supported extension again, since the consensus on
3300 sbcl-devel was that it can be useful for ordinary development
3301 work, not just for debugging SBCL itself.
3302 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3303 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3304 that currently-defined functions might be redefined later with
3305 different return types.
3306 * Hash tables can be printed readably, as inspired by CMU CL code
3307 of Eric Marsden and SBCL code of Martin Atzmueller.
3308 * better error handling in CLOS method combination, thanks to
3309 Martin Atzmueller porting Pierre Mai's CMU CL patches
3310 * more overflow fixes for >16Mbyte I/O buffers
3311 * A bug in READ has been fixed, so that now a single Ctrl-D
3312 character suffices to cause end-of-file on character streams.
3313 In particular, now you only need one Ctrl-D at the command
3314 line (not two) to exit SBCL.
3315 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3317 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3318 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3319 does the right thing.
3320 * The implementation of some type tests, especially for CONDITION
3321 types, is now tidier and maybe faster, due to CMU CL code
3322 originally by Douglas Crosher, ported by Martin Atzmueller.
3323 * Some math functions have been fixed, and there are new
3324 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3325 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3326 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3327 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3328 likely to be built into sbcl-0.7.0.)
3329 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3330 CL, ported to SBCL by Martin Atzmueller)
3331 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3332 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3333 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3334 * There's a new slam.sh hack to shorten the edit/compile/debug
3335 cycle for low-level changes to SBCL itself, and a new
3336 :SB-AFTER-XC-CORE target feature to control the generation of
3337 the after-xc.core file needed by slam.sh.
3338 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3339 COMPILE-FILE is no longer supported, so that now every function
3340 gets an entry point, so that block compilation looks a little
3341 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3342 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3343 returns the number of bytes consed since the system started,
3344 rather than the number consed since the first time the function
3345 was called. (The new definition parallels ANSI functions like
3346 CL:GET-INTERNAL-RUN-TIME.)
3347 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3348 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3349 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3350 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3351 * Fasl file version numbers are now independent of the target CPU,
3352 since historically most system changes which required version
3353 number changes have affected all CPUs equally. Similarly,
3354 the byte fasl file version is now equal to the ordinary
3357 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3358 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3359 is no longer recognized. I apologize for this, because it was
3360 listed in SB-EXT as a supported extension, but I found that
3361 its existing behavior was poorly specified, as well as incorrectly
3362 specified, and it looked like too much of a mess to straighten it
3363 out. I have enough on my hands trying to get ANSI stuff to work..
3364 * many patches ported from CMU CL by Martin Atzmueller, with
3365 half a dozen bug fixes in pretty-printing and the debugger, and
3366 half a dozen others elsewhere
3367 * fixed bug 13: Floating point infinities are now supported again.
3368 They might still be a little bit flaky, but thanks to bug reports
3369 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3370 as flaky as they were.
3371 * The --noprogrammer command line option is now supported. (Its
3372 behavior is slightly different in detail from what the old man
3373 page claimed it would do, but it's still appropriate under the
3374 same circumstances that the man page talks about.)
3375 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3376 are now supported, and enabled by default. Thus, the compiler can
3377 handle many floating point and complex operations much less
3378 inefficiently. (Thus e.g. you can implement a complex FFT
3380 * The compiler now detects type mismatches between DECLAIM FTYPE
3381 and DEFUN better, and implements CHECK-TYPE more correctly, and
3382 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3383 thanks to patches from Martin Atzmueller.
3384 * various fixes to make the cross-compiler more portable to
3385 ANSI-conforming-but-different cross-compilation hosts (notably
3386 Lispworks for Windows, following bug reports from Arthur Lemmens)
3387 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3388 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3389 fixed thanks to Pierre Mai's CMU CL patches.
3390 * fixes to keep the system from overflowing internal counters when
3391 it tries to use i/o buffers larger than 16M bytes
3392 * fixed bug 45a: Various internal functions required to support
3393 complex special functions have been merged from CMU CL sources.
3394 (When I was first setting up SBCL, I misunderstood a compile-time
3395 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3396 * improved support for type intersection and union, fixing bug 12
3397 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3398 more obscure bugs as well
3399 * some steps toward byte-compiling non-performance-critical
3400 parts of the system, courtesy of patches from Martin Atzmueller
3401 * Christophe Rhodes has made some debian packages of sbcl at
3402 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3403 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3404 stable, but are nonetheless usable. When he's ready, I'd be happy
3405 to add them to the SourceForge "File Releases" section. (And if
3406 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3407 * new fasl file format version number (because of changes in
3408 internal representation of (OR ..) types to accommodate the new
3409 support for (AND ..) types, among other things)
3411 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3412 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3414 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3416 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3417 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3418 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3419 * DESCRIBE now works on structure objects again.
3420 * Most function call argument type mismatches are now handled as
3421 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3422 know whether the function will be redefined before the call is
3423 executed. (The compiler could flag local calls with full WARNINGs,
3424 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3425 it doesn't keep track of enough information to know whether calls
3426 are local in this sense.)
3427 * Compiler output is now more verbose, with messages truncated
3428 later than before. (There should be some supported way for users
3429 to override the default verbosity, but I haven't decided how to
3430 provide it yet, so this behavior is still controlled by the internal
3431 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3432 src/compiler/ir1util.lisp.)
3433 * Fasl file format version numbers have increased again, because
3434 support for the Gray streams extension changes the layout of the
3435 system's STREAM objects.
3436 * The Gray subclassable streams extension now works, thanks to a
3437 patch from Martin Atzmueller.
3438 * The full LOAD-FOREIGN extension (not just the primitive
3439 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3440 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3441 but like most other programs, it defaults to copying the Unix
3442 environment from the original process instead of starting the
3443 new process in an empty environment.
3444 * Extensions which manipulate the Unix environment now support
3445 an :ENVIRONMENT keyword option which doesn't smash case or
3446 do other bad things. The CMU-CL-style :ENV option is retained
3447 for porting convenience.
3448 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3449 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3451 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3453 * A patch from Martin Atzmueller seems to have solved the SIGINT
3454 problem, and as far as we know, signal-handling now works cleanly.
3455 (If you find any new bugs, please report them!)
3456 * The system no longer defaults Lisp source file names to types
3457 ".l", ".cl", or ".lsp", but only to ".lisp".
3458 * The compiler no longer uses special default file extensions for
3459 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3460 seems to expect a single default extension for all compiled code,
3461 and there's no compelling reason to try to stretch the standard
3462 to allow two different extensions.) Instead, byte-compiled files
3463 default to the same extension as native-compiled files.
3464 * Fasl file format version numbers have increased again, because
3465 a rearrangement of internal implementation packages made some
3466 dumped symbols in old fasl files unreadable in new cores.
3467 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3468 many fewer weird special cases.
3469 * Bug #17 (differing COMPILE-FILE behavior between logical and
3470 physical pathnames) has been fixed, and some related misbehavior too,
3471 thanks to a patch from Martin Atzmueller.
3472 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3473 by Tim Moore, ported to SBCL by Martin Atzmueller.
3474 * Martin Atzmueller fixed several filesystem-related problems,
3475 including bug #36, in part by porting CMU CL patches, which were
3476 written in part by Paul Werkowski.
3477 * More compiler warnings in src/runtime/ are gone, thanks to
3478 more patches from Martin Atzmueller.
3479 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3482 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3484 * DESCRIBE now works on CONDITION objects.
3485 * The debugger now handles errors which arise when trying to print
3486 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3488 * The build system now uses an additional file, customize-target-features.lisp,
3489 to allow local modifications to the target *FEATURES* list. (The point of
3490 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3491 debugging features enabled, without having to worry about propagating it
3492 into everyone's system when I do a "cvs update".) When no
3493 customize-target-features.lisp file exists, the target *FEATURES* list
3494 should be constructed the same way as before.
3495 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3496 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3497 warn about behavior which is undefined under the ANSI spec. Note: This
3498 is specified by ANSI, but it's not very popular with programmers.
3499 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3500 macro in the SBCL sources for an example of a workaround which you
3501 might use to make portable ANSI-standard code which does what you want.
3502 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3503 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3504 when it's used as a non-toplevel form. (This is required in order
3505 to implement the DEFCONSTANT-EQX macro.)
3506 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3507 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3508 funny things, due to bugs in the implementation of EVAL-WHEN
3509 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3510 non-toplevel DEFCONSTANTs are uncommon.)
3511 * The core file version number and fasl file version number have been
3512 incremented, because the old noncompliant DEFCONSTANT behavior involved
3513 calling functions which no longer exist, and because I also took the
3514 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3515 * fixed bug 1 (error handling before read-eval-print loop starts), and
3516 redid debugger restarts and related debugger commands somewhat while
3518 ** The QUIT debugger command is gone, since it did something
3519 rather different than the SB-EXT:QUIT command, and since it never
3520 worked properly outside the main toplevel read/eval/print loop.
3521 Invoking the new TOPLEVEL restart provides the same functionality.
3522 ** The GO debugger command is also gone, since you can just invoke
3523 the CONTINUE restart directly instead.
3524 ** The TOP debugger command is also gone, since it's redundant with the
3525 FRAME 0 command, and since it interfered with abbreviations for the
3527 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3529 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3530 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3531 some porting work by Martin Atzmueller)
3532 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3533 with a boatload of patches which he ported to SBCL. Now that those
3535 ** The system tries to make sure that its low-priority messages
3536 are prefixed by semicolons, to help people who like to use
3537 syntax highlighting in their ILISP buffer. (This patch
3538 was originally due to Raymond Toy.)
3539 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3540 patches originally written by Raymond Toy.
3541 ** The compiler understands coercion between single-value and
3542 multiple-VALUES type expressions better, getting rid of some very
3543 weird behavior, thanks to patches originally by Robert MacLachlan
3544 and Douglas Crosher.
3545 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3546 lambda lists, thanks to a patch originally by Pierre Mai.
3547 ** The system no longer bogusly warns about "abbreviated type
3549 ** The compiler gets less confused by inlining and RETURN-FROM,
3550 thanks to some patches originally by Tim Moore.
3551 ** The system no longer hangs when dumping circular lists to fasl
3552 files, thanks to a patch originally from Douglas Crosher.
3553 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3554 undefined function error.
3555 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3556 in structure slot initforms) for now, documented workaround instead:-|
3557 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3558 * fixed bug 5: added stubs for various Gray stream functions called
3559 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3560 installed, at least appropriate type errors are generated
3561 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3562 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3563 consistently in DEFMETHOD forms.
3564 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3565 it doesn't seem to affect SBCL after all
3566 * The C runtime system now builds with better optimization and many
3567 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3569 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3571 * The system is now under CVS at SourceForge (instead of the
3572 CVS repository on my home machine).
3573 * The new signal handling code has been tweaked to treat register
3574 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3575 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3576 downstream, e.g. in debug-int.lisp, has implicit dependencies
3577 on the unsignedness of integer representation of machine words,
3578 and that caused the system to bomb out with infinite regress
3579 when trying to recover from type errors involving signed values,
3580 e.g. (BUTLAST '(1 2 3) -1).)
3581 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3582 (This was one of the bugs Peter Van Eynde reported back in July.)
3583 * The system now uses code inspired by Colin Walters' O(N)
3584 implementation of MAP (from the cmucl-imp@cons.org mailing
3585 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3586 inline the MAP operation, and there is more than one
3587 sequence argument to the MAP call (so that it can't just
3588 do ETYPECASE once and for all based on the type of the
3589 single sequence argument). (The old non-inline implementation
3590 of the general M-argument sequence-of-length-N case required
3591 O(M*N*N) time when any of the sequence arguments were LISTs.)
3592 * The QUIT :UNIX-CODE keyword argument has been renamed to
3593 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3594 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3596 * Raymond Wiker's patches to port dynamic loading from Linux to
3597 FreeBSD have been added.
3598 * The BUGS file is now more nearly up to date, thanks in large part
3599 to Martin Atzmueller's review of it.
3600 * The debugger now flushes standard output streams before it begins
3601 its output ("debugger invoked" and so forth).
3602 * The core version number and fasl file version number have both
3603 been incremented, because of incompatible changes in the layout
3605 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3606 used to not be called for a saved Lisp image.)
3607 * Martin Atzmueller's version of a patch to fix a compiler crash,
3608 as posted on sbcl-devel 13 September 2000, has been installed.
3609 * Instead of installing Martin Atzmueller's patch for the
3610 compiler transform for SUBSEQ, I deleted the compiler transform,
3611 and transforms for some similar consing operations.
3612 * A bug in signal handling which kept TRACE from working on OpenBSD
3614 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3615 away by constant folding
3616 * The system now defines its address space constants in one place
3617 (in the Lisp sources), and propagates them automatically elsewhere
3618 (through GENESIS and the sbcl.h file). Therefore, patching the
3619 address map is less unnecessarily tedious and error-prone. The
3620 Lisp names of address space constants have also been systematized.
3621 * CVS tags like dollar-Header-dollar have been removed from
3622 the sources, because they have never saved me trouble and
3623 they've been source of trouble working with patches and other
3624 diff-related operations.
3625 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3626 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3628 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3630 * The system has been ported to OpenBSD.
3631 * The system now compiles with a simple "sh make.sh" on the systems
3632 that it's supported on. I.e., now you no longer need to tweak
3633 text in the target-features.lisp-expr and symlinks in src/runtime/
3634 by hand, the make.sh takes care of it for you.
3635 * The system is no longer so grossly inefficient when compiling code
3636 involving vectors implemented as general (not simple) vectors (VECTOR T),
3637 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3638 which dares to use the various sequence functions on non-simple
3639 vectors, takes less of a performance hit.
3640 * There is now a primitive type predicate VECTOR-T-P
3641 to test for the (VECTOR T) type, so that e.g.
3642 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3643 can now be compiled with some semblance of efficiency. (The old code
3644 turned the type declaration into a full call to %TYPEP at runtime!)
3645 * AREF on (VECTOR T) is still not fast, since it's still compiled
3646 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3647 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3648 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3649 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3650 without making full calls to SUBTYPEP at runtime.
3651 (Some analogous efficiency issues for non-simple vectors specialized to
3652 element types other than T, or for non-simple multidimensional arrays,
3653 have not been addressed. They could almost certainly be handled the
3654 same way if anyone is motivated to do so.)
3655 * The changes in array handling break binary compatibility, so
3656 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3657 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3658 instead of (VALUES T T).
3659 * By following the instructions that Dan Barlow posted to sbcl-devel
3660 on 2 July 2000, I was able to enable primitive dynamic object
3661 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3662 functionality is not implemented (since it calls ld to resolve
3663 library references automatically, requiring RUN-PROGRAM for its
3664 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3665 try to resolve library references) is now supported.
3666 * The system now flushes the standard output streams when it terminates,
3667 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3668 them at several other probably-convenient times, e.g. in each pass of
3669 the toplevel read-eval-print loop, and after evaluating a form given
3670 as an "--eval" command-line option. (These changes were motivated by a
3671 discussion of stream flushing issues on cmucl-imp in August 2000.)
3672 * The source transform for TYPEP of array types no longer assumes
3673 that an array whose element type is a not-yet-defined type
3674 is implemented as an array of T, but instead punts, so that the
3675 type will be interpreted at runtime.
3676 * There is now some support for cross-compiling in make.sh: each of
3677 the phases of make.sh has its own script. (This should be transparent
3678 to people doing ordinary, non-cross-compile builds.)
3679 * Since my laptop doesn't have hundreds of megabytes of memory like
3680 my desktop machine, I became more motivated to do some items on
3681 my to-do list in order to reduce the size of the system a little:
3682 ** Arrange for various needed-only-at-cold-init things to be
3683 uninterned after cold init. To support this, those things have
3684 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3685 symbols with such names are now uninterned after cold init).
3686 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3687 fasl files for cold load.
3688 ** Remove the old compiler structure pooling code (which used to
3689 be conditional on the target feature :SB-ALLOC) completely.
3690 ** Redo the representation of some data in cold init to be more compact.
3691 (I also looked into supporting byte compiled code at bootstrap time,
3692 which would probably reduce the size of the system a lot, but that
3693 looked too complicated, so I punted for now.)
3694 * The maximum signal nesting depth in the src/runtime/ support code has
3695 been reduced from 4096 to 256. (I don't know any reason for the very
3696 large old value. If the new smaller value turns out to break something,
3697 I'll probably just bump it back up.)
3698 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3700 * Many, many bugs reported by Peter Van Eynde have been added to
3701 the BUGS list; some have even been fixed.
3702 * While enabling dynamic object file loading, I tried to make the
3703 code easier to understand, renaming various functions and variables
3704 with less ambiguous names, and changing some function calling
3705 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3706 * While trying to figure out how to do the OpenBSD port, I tried to
3707 clean up some of the code in src/runtime/. In particular, I dropped
3708 support for non-POSIX signal handling, added various comments,
3709 tweaked the code to reduce the number of compilation warnings, and
3710 renamed some files to increase consistency.
3711 * To support the new automatic configuration functionality in make.sh,
3712 the source file target-features.lisp-expr has been replaced with the
3713 source file base-target-features.lisp-expr and the machine-generated
3714 file local-target-features.lisp-expr.
3715 * fixed a stupid quoting error in make.sh so that using CMU CL
3716 "lisp -batch" as cross-compilation host works again
3718 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3720 * DESCRIBE no longer tries to call itself recursively to describe
3721 bound/fbound values, so that it no longer fails on symbols which are
3722 bound to themselves (like keywords, T, and NIL).
3723 * DESCRIBE now works on generic functions.
3724 * The printer now prints less-screwed-up representations of closures
3725 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3726 * A private symbol is used instead of the :EMPTY keyword previously
3727 used to mark empty slots in hash tables. Thus
3728 (DEFVAR *HT* (MAKE-HASH-TABLE))
3729 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3730 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3731 now does what ANSI says that it should. (You can still get
3732 similar noncompliant behavior if bang on the hash table
3733 implementation with all the symbols you get back from
3734 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3735 This breaks binary compatibility, since tests for equality to
3736 :EMPTY are wired into things like the macroexpansion of
3737 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3739 * There's now a minimal placeholder implementation for CL:STEP,
3740 as required by ANSI.
3741 * An obscure bug in the interaction of the normal compiler, the byte
3742 compiler, inlining, and structure predicates has been patched
3743 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3744 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3746 * Missing ordinary arguments in a macro call are now detected even
3747 when the macro lambda list contains &KEY or &REST.
3748 * The debugger no longer complains about encountering the top of the
3749 stack when you type "FRAME 0" to explicitly instruct it to go to
3750 the top of the stack. And it now prints the frame you request even
3751 if it's the current frame (instead of saying "You are here.").
3752 * As specified by ANSI, the system now always prints keywords
3753 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3755 * The default initial SIZE of HASH-TABLEs is now smaller.
3756 * Type information from CLOS class dispatch is now propagated
3757 into DEFMETHOD bodies, so that e.g.
3758 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3760 is now basically equivalent to
3761 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3762 (DECLARE (TYPE SINGLE-FLOAT X))
3764 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3765 operation, without having to do run-time type dispatch.
3766 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3767 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3768 * Since I don't seem to be making big file reorganizations very often
3769 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3770 is overflowing my ability to conveniently back them up), I've finally
3771 checked the system into CVS. (The CVS repository is on my home system,
3772 not at SourceForge -- putting it on SourceForge might come later.)
3773 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3774 high-level GC-NOTIFY-FOO functions send their output. (There's
3775 still very little control of where low-level verbose GC functions
3776 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3777 less than it used to -- the GC-NOTIFY-FOO functions are now under
3778 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3779 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3780 in only one place in the source code, and propagates it automatically
3781 everywhere that it's needed. Thus e.g. when I bump the version from
3782 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3783 * The C source files now include boilerplate legalese and documentation
3784 at the head of each file (just as the Lisp source files already did).
3785 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3786 to his page will be replaced with a link to his new CLiki service.
3788 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3790 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3791 * The build process now looks for GNU make under the default name "gmake",
3792 instead of "make" as it used to. If GNU make is not available as "gmake"
3793 on your system, you can change this default behavior by setting the
3794 GNUMAKE environment variable.
3795 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3796 can build without error under CMU CL.
3798 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3800 * There is now a partial SBCL user manual (with some new text and some
3801 text cribbed from the CMU CL manual).
3802 * The beginnings of a profiler have been added (starting with the
3803 CMU CL profiler and simplifying and cleaning up). Eventually the
3804 main interface should be through the TRACE macro, but for now,
3805 it's still accessed through vaguely CMU-CL-style functions and macros
3806 exported from the package SB-PROFILE.
3807 * Some problems left over from porting CMU CL to the new
3808 cross-compilation bootstrap process have been cleaned up:
3809 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3810 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3811 part when getting rid of the compiler *BACKEND* stuff.)
3812 ** The value of *NULL-TYPE* was screwed up, because it was
3813 being initialized before the type system knew the final
3814 definition of the 'NULL type. This screwed up several key
3815 optimizations in the compiler, causing inefficiency in all sorts
3816 of places. (I found it because I wanted to understand why
3817 GET-INTERNAL-RUN-TIME was consing.)
3818 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3819 PROCLAIM FTYPE information. Unfortunately this broke binary
3820 compatibility again, since now the forms output by DEFGENERIC
3821 to refer to functions which didn't exist in 0.6.3.
3822 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3823 can use the new (as of 0.6.3) transform for SOME into MAP into
3825 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3826 values to (INTEGER 0 1000000), so that the time code will no longer
3827 occasionally get blown up by Linux returning 1000000 microseconds
3828 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3829 its output conform to the ANSI spec. (Alas, this makes its output
3830 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3831 * A full call to MAP NIL with a single sequence argument no longer conses.
3832 * fixes to problems pointed out by Martin Atzmueller:
3833 * The manual page no longer talks about multiprocessing as though
3834 it were currently supported.
3835 * The ILISP support patches have been removed from the distribution,
3836 because as of version 5.10.1, ILISP now supports SBCL without us
3837 having to maintain patches.
3838 * added a modified version of Raymond Toy's recent CMU CL patch for
3839 EQUALP comparison of HASH-TABLE
3841 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3843 * The system still can't cross-compile itself with
3844 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3845 entails), but at least it can compile more of itself that way
3846 than it used to be able to, and various buglets which were uncovered
3847 by trying to cross-compile itself that way have now been fixed.
3848 * This release breaks binary compatibility again. This time
3849 at least I've incremented the FASL file format version to 2, so that the
3850 problem can be detected reliably instead of just causing weird errors.
3851 * various new style warnings:
3852 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3853 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3854 ** using the lexical binding of a variable named in the *FOO* style
3855 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3856 as specified by ANSI.
3857 * *RANDOM-STATE* is no longer automatically initialized from
3858 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3859 default behavior of the system is to repeat its behavior every time
3860 it's run. If you'd like to change this behavior, you can always
3861 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3862 old convention there was no comparably easy way to get the system to
3863 repeat its behavior every time it was run.
3864 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3865 FTYPE declarations has been removed, in favor of their ANSI
3866 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3867 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3868 the types of their sequence arguments can be determined at compile time.
3869 This is done through a new open code expansion for MAP which eliminates
3870 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3871 argument types can be determined at compile time.
3872 * The optimizer now transforms COERCE into an identity operation when it
3873 can prove that the coerced object is already of the correct type. (This
3874 can be a win for machine generated code, including the output of other
3875 optimization transforms, such as the MAP transform above.)
3876 * Credit information has been moved from source file headers into CREDITS.
3877 * Source file headers have been made more standard.
3878 * The CASE macro now compiles without complaining even when it has
3881 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3883 * (Note that the way that the PCL macroexpansions were rewritten
3884 to accommodate the change in DEFGENERIC below breaks binary
3885 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3886 not run under sbcl-0.6.2. Once we get out of alpha releases,
3887 i.e. hit release 1.0.0, we'll probably try to maintain binary
3888 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3889 and sbcl-1.4.4. Until then, however, it might be fairly common
3890 for maintenance releases to break binary compatibility.)
3891 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3893 * The reporting of unhandled signals has been changed to print some
3894 explanatory text as well as the report form. (Previously only
3895 the report form was printed.)
3896 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3897 it defines, so that the compiler no longer issues undefined function
3898 warnings for compiled-but-not-yet-loaded generic functions.
3899 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3900 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3901 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3902 section of the ANSI spec are supported.
3903 * The "" nickname for the "KEYWORD" package has been removed.
3904 The reader still handles symbol tokens which begin with a package marker
3905 as keywords, but it doesn't expose its mechanism for doing so in the
3906 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3907 * The system now issues STYLE-WARNINGs for contradictory TYPE
3908 proclamations. (Warnings for contradictory FTYPE proclamations would
3909 be nice too, but those can't be done usefully unless the type system
3910 is made smarter about FUNCTION types.)
3911 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3912 systematized, so that "*target-*.lisp is supposed to exist only on the
3913 target and imply that there's a related file which exists on the
3914 host, and *host-*.lisp is supposed to exist only on the host and imply
3915 that there's a related file which exists on the target. This involves a
3916 lot of renaming. Hopefully the acute confusion caused by the renaming
3917 will be justified by the reduction in chronic confusion..
3918 ** runtime-type.lisp -> early-target-type.lisp
3919 ** target-type.lisp -> late-target-type.lisp
3920 ** early-host-format.lisp -> early-format.lisp
3921 ** late-host-format.lisp -> late-format.lisp
3922 ** host-error.lisp -> misc-error.lisp
3923 ** early-error.lisp -> early-target-error.lisp
3924 ** late-error.lisp -> late-target-error.lisp
3925 ** host-defboot.lisp -> early-defboot.lisp
3926 ** code/misc.lisp -> code/target-misc.lisp
3927 ** code/host-misc.lisp -> code/misc.lisp
3928 ** code/numbers.lisp -> code/target-numbers.lisp
3929 ** code/early-numbers.lisp -> numbers.lisp
3930 ** early-host-type.lisp -> early-type.lisp
3931 ** late-host-type.lisp -> late-type.lisp
3932 ** host-typep.lisp -> typep.lisp
3933 ** load.lisp -> target-load.lisp
3934 ** host-load.lisp -> load.lisp
3935 ** host-disassem.lisp -> disassem.lisp
3936 ** host-insts.lisp -> insts.lisp
3937 ** byte-comp.lisp -> target-byte-comp.lisp
3938 ** host-byte-comp.lisp -> byte-comp.lisp
3939 ** host-signal.lisp -> signal.lisp
3940 ** host-defstruct.lisp -> defstruct.lisp
3941 ** late-target-type.lisp -> deftypes-for-target.lisp
3942 Furthermore, several other previously target-only files foo.lisp (e.g.
3943 hash-table.lisp and random.lisp) have been split into a target-and-host
3944 foo.lisp file and a target-only target-foo.lisp file, with their key type
3945 definitions in the target-and-host part, so that the cross-compiler will
3946 know more about target types.
3947 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3948 gone away. In their place are various *BACKEND-FOO* variables
3949 corresponding to the slots of the old structure.
3950 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3951 into the target SBCL has been fixed.
3952 * The chill.lisp system for loading cold code into a running SBCL
3954 * Support for the CMU CL "scavenger hook" extension has been removed.
3955 (It was undocumented and unused in the CMU CL sources that SBCL was
3956 derived from, and stale in sbcl-0.6.1.)
3957 * Various errors in the cross-compiler type system were detected
3958 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3959 (enabling various consistency checks). Many of them were fixed,
3960 but some hard problems remain, so the compiler is back to
3961 running without *TYPE-SYSTEM-INITIALIZED* for now.
3962 * As part of the cross-compiler type system cleanup, I implemented
3963 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3964 * I have started adding UNCROSS calls throughout the type system
3965 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3966 on cross-compiler input files will be able to go away, and various
3968 * CONSTANTP now returns true for quoted forms (as explicitly required
3971 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3973 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3974 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3975 (MAKE-LIST -1). In the longer term, it ought to become true
3976 that declarations are assertions even at SAFETY 1. For now, it's not
3977 quite true even at SAFETY 3, but it's at least more nearly true..
3978 (Note that this change seems to increases the size of the system by
3979 O(5%) and to decrease the speed of the compiler by 20% or more.)
3980 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3981 for the problem of printing dozens of lines of distracting information
3982 about low-level system machinery as part of the top stack frame
3983 on entry to the debugger when an undefined function was called.
3984 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3985 is not reset to COMMON-LISP-USER.
3986 * Compilation of stuff related to dyncount.lisp has been made conditional
3987 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3988 smaller. The various dyncount-related symbols have been moved into
3989 a new "SB-DYNCOUNT" package.
3990 * tty-inspect.lisp has been renamed to inspect.lisp.
3991 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3992 feature has gone away. (When we eventually port to other flavors of
3993 libc and/or Unix, we'll try to make the differences between flavors
3994 invisible at the user level.)
3995 * Various other *FEATURES* tags, and/or their associated conditionals,
3996 have been removed if obsolescent, or given better documentation, or
3997 sometimes given more-mnemonic names.
3999 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4001 * tidied up "make.sh" script
4002 * tidied up system directory structure
4003 * better "clean.sh" behavior
4004 * added doc/FOR-CMUCL-DEVELOPERS
4005 * many many small tweaks to output format, e.g. removing possibly-confusing
4006 trailing #\. character in DESCRIBE-INSTANCE
4007 * (EQUALP #\A 'A) no longer signals an error.
4008 * new hashing code, including EQUALP hashing
4009 * tidied up Lisp initialization and toplevel
4010 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4011 * command line argument processing
4012 * added POSIX-GETENV function to deal with Unix-ish environment variables
4013 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4014 terminating SBCL on EOF
4015 * non-verbose GC by default
4016 * There is no more "sbcl" shell script; the sbcl file is now the C
4017 runtime executable (just like CMU CL).
4018 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4020 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4021 symbol and package information as Lisp native symbol and package objects
4022 instead of strings naming symbols and strings naming packages. This way,
4023 whenever packages are renamed (as in warm init), debug information is
4024 transformed along with everything else.
4025 * tweaked the optimization policy declarations which control the building
4026 of SBCL itself. Now, among other things, the system no longer saves
4027 source location debugging information. (This helps two problems at once
4028 by reducing SBCL size and by keeping SBCL from trying to look for its
4029 sources -- which may not exist -- when reporting errors.)
4030 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4031 debugging and testing purposes
4032 * cleaned up printing, making the printer call PRINT-OBJECT for
4033 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4034 methods, giving nearly-ANSI behavior
4035 * converted almost all special variables to use *FOO* naming convention
4036 * deleted PARSE-TIME functionality, since it can be done portably
4037 * moved some files out of cold init into warm init
4038 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4040 * regularized formatting of source files
4041 * added an install.sh script
4042 * fixed ridiculous memory usage of cross-compiler by making
4043 compiler/alloc.lisp not try to do pooling unless it can hook
4044 itself into the GC of the cross-compilation host. Now the system
4045 builds nicely on my old laptop.
4046 * added :SB-ALLOC in target-features.lisp-expr
4047 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4048 was not implemented)
4049 * re-did condition handling and note reporting in the compiler. Notes
4050 are no longer handled by signalling conditions. Style warnings
4051 and warnings are handled more correctly and reported in such a way
4052 that it's easy to find one or the other in your output (so that you
4053 can e.g. figure out which of many problems caused COMPILE-FILE to
4055 * changed the severity of several compiler warnings from full WARNING
4056 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4057 compiler note reporting so that it doesn't use the condition system
4058 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4059 COMPILE-FILE command)
4060 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4061 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4062 no longer supported, since their implementation was deeply intertwingled
4063 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4064 * removed bogus "support" for compiler macros named (SETF FOO), and
4065 removed the compiler macro for SETF INFO (but only after making a fool
4066 of myself on the cmucl-imp mailing list by posting a bogus patch for
4067 DEFINE-COMPILER-MACRO..)
4068 * Compiled files containing forms which have side effects on the Lisp
4069 reader (such as DEFPACKAGE forms) are now handled more correctly.
4070 (Compiler queuing of top level lambdas has been suppressed by setting
4071 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4072 * deleted various currently-unused source files, e.g. gengc.lisp. They
4073 may be added back at some point e.g. when porting to other architectures,
4074 but until they are it's distracting to distribute them and to try to
4076 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4077 there were so many of them they're just distractions, and UNCROSS is
4078 known to be able to handle the current sources
4079 * moved PROFILE functionality into TRACE, so that it will be clear
4080 how the wrapping and unwrapping of functions when you profile them
4081 interacts with the wrapping and unwrapping of functions when you
4082 trace them. (Actually, the functionality isn't there yet, but at least
4083 the interface specification is there. Hopefully, the functionality will
4084 arrive with some future maintenance release.)
4085 * removed host-oops.lisp
4086 * changed signature of QUIT function to allow UNIX-CODE argument
4087 * fixed READ-SEQUENCE bug
4088 * tweaked verbose GC output so that it looks more like the progress
4089 output that ANSI specifies for functions like LOAD
4090 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4091 * added <http://sbcl.sourceforge.com> to the banner information printed by