1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.1 relative to sbcl-1.0:
3 * improvement: sb-sprof traces call stacks to an arbitrary depth on
4 x86/x86-64, rather than the previous fixed depth of 8
5 * bug fix: non-ascii command-line arguments are processed correctly
6 (thanks to Yaroslav Kavenchuk)
7 * bug fix: TRACE :ENCAPSULATE NIL (and function end breakpoints)
9 * bug fix: non-required arguments were not passed correctly when a method
10 defined using DEFMETHOD was called from a mop-generated method using
11 CALL-NEXT-METHOD (reported by Pascal Costanza)
12 * bug fix: recursion is now permitted in accessors through
13 SLOT-UNBOUND. (reported by Pascal Costanza)
14 * bug fix: an error was signaled at startup if the HOME environment
15 variable was defined, but had an empty value (reported by Peter Van Eynde)
16 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
17 signal a PROGRAM-ERROR, not a BUG.
18 * bug fix: SB-POSIX:READDIR works when built with large file support.
19 * optimization: loading generic functions no longer takes O(n^2) time,
20 proportional to the amount of methods in the generic function
21 (reported by Todd Sabin and Jeremy Brown)
22 * optimization: the FIND and POSITION family of sequence functions
23 are significantly faster on arrays whose element types have been
26 changes in sbcl-1.0 relative to sbcl-0.9.18:
27 * improvement: experimental support for threading on FreeBSD/x86.
28 (thanks to NIIMI Satoshi)
29 * improvement: runtime option --dynamic-space-size can be used
30 to set the size of the dynamic space reserved on startup.
31 * improvement: floating point modes in effect are now saved in
32 core, and restored on startup.
33 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
34 startup, not time since first call to GET-INTERNAL-REAL-TIME.
35 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
36 threads are not running after *SAVE-HOOKS* have run.
37 * improvement: writes to CLOS instance slots are type-checked in code
38 compiled with (SAFETY 3)
39 * improvement: floating-point exception handling on FreeBSD (thanks to
41 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
42 (thanks to Zach Beane)
43 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
45 * improvement: added support for the Shift-JIS external format.
46 (contributed by NIIMI Satoshi)
47 * improvement: callbacks are supported on Linux/PPC. (thanks to
49 * bug fix: compiler bug triggered by a (non-standard) VALUES
50 declaration in a LET* was fixed. (reported by Kaersten Poeck)
51 * bug fix: file compiler no longer confuses validated and already
52 dumped structurres. (reported by Kaersten Poeck)
53 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
54 fill-pointer signals a type-error as required. (thanks to
56 * bug fix: disassemly of funcallable instances works.
57 * bug fix: single stepping on PPC.
58 * bug fix: fix thread-safety problems in the type system (generally
59 manifesting as nonsensical errors like "STRING is a bad type specifier
60 for sequences" or "The value 1 is not of type FIXNUM")
61 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
62 (reported by Josip Gracin)
63 * bug fix: an error is signaled for attempts to displace arrays with
64 incompatible element types (thanks to Mario Mommer)
65 * bug fix: more correct handling of wide characters in debug info
66 (bug reported by Attila Lendvai and fixed by Juho Snellman)
67 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
68 and don't cause extra consing
69 * optimization: MAP and MAP-INTO are significantly faster on vectors
70 whose elements types have been declared.
71 * Improvements to SB-SPROF:
72 ** Support for allocation profiling
73 ** Reduced profiling overhead, especially for long profiling runs
74 * Improvements to the Windows port:
75 ** floating point exceptions are now reported correctly.
76 ** stack exhaustion detection works partially.
77 ** more accurate GET-INTERNAL-REAL-TIME.
78 ** better RUN-PROGRAM behaviour in case of errors in the child
80 ** PROBE-FILE now simplifies pathnames correctly.
81 ** DIRECTORY now works correctly with :WILD-INFERIORS.
82 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
83 1st 1970 (although time timezone data will be unreliable).
84 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
86 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
87 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
88 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
90 * bug fix: two potential GC deadlocks affecting threaded builds.
91 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
92 returns T (reported by Anton Kazennikov)
93 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
94 with non-variable places
95 * bug fix: remove a race condition in the setting of
96 funcallable-instance functions, this should make threaded CLOS
97 code more stable against memory faults.
98 * bug fix: corruption of specials when unbinding is interrupted by an
99 asynchronous unwind (reported by Hannu Koivisto)
100 * improvement: the debugger will now also display local variables that
101 are only used once, for code compiled with a DEBUG optimization quality
104 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
105 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
106 * incompatible change: External-format support for FFI calls. The
107 SB-ALIEN:C-STRING no longer implies an ASCII
108 external-format. Instead the string is subject to external-format
109 conversion. Additionally return values of type SB-ALIEN:C-STRING
110 are by default of type (SIMPLE-ARRAY CHARACTER), not
111 SIMPLE-BASE-STRING. For an alien type that behaves like the old
112 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
113 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
114 * incompatible change: SB-EXT package no longer contains the
115 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
116 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
118 * incompatible change: the single-stepper is no longer available
119 on Alpha, Mips and Sparc platforms.
120 * minor incompatible change: the direct superclasses of
121 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
122 not (STANDARD-OBJECT FUNCTION). This makes the
123 class-precedence-lists of GENERIC-FUNCTION and
124 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
126 * minor incompatible change: the default stream external format on
127 non-unicode SBCL is no longer detected from the locale, but is always
129 * new feature: Add a version of evaluator that uses an interpreter instead
130 of the compiler. EVAL still uses the compiler by default, to switch it
131 to use the interpreter, set the value of the variable
132 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
133 * minor incompatible change: the single-stepper REPL has been merged
134 with the normal debugger (see the "Stepping" heading of the debugger help
135 for more details). The debugger command STEP will no longer switch
136 to the single-stepper REPL.
137 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
138 for a type now works.
139 * bug fix: loading of default sysinit file works. (thanks to Leonid
141 * bug fix: better detection of circularities in the file-compiler.
142 (reported by Marco Monteiro)
143 * bug fix: the CL pathname functions now work with files that have
144 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
145 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
146 systems with Unix98 pty semantics.
147 * bug fix: ASDF-INSTALL will now work with bsd tar.
148 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
150 * bug fix: timers expiring in dead threads no longer cause a
151 type-error (reported by Paul "Nonny Mouse").
152 * bug fix: thanks to more lightweight single-stepper instrumentation,
153 code compiled with (DEBUG 3) will compile and execute significantly faster,
154 and will have more accurate type-inferencing than before
155 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
156 whose bindings are modified
157 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
158 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
159 ** replace dummy implementations of CL:MACHINE-INSTANCE and
160 CL:SOFTWARE-VERSION with proper version
162 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
163 * feature: implemented the READER-METHOD-CLASS and
164 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
165 as specified by AMOP.
166 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
168 * optimization: faster LOGCOUNT implementation on x86 and x86-64
169 (thanks to Lutz Euler)
170 * optimization: hashing of general arrays and vectors has been
171 improved. (reported by Any Fingerhut)
172 * enhancement: SB-INTROSPECT is now able to find definitions of
173 profiled functions. (thanks to Troels Henriksen)
174 * enhancement: compiler-macro expansion applies now to FUNCALL forms
175 as well, allowing compiler-macros for SETF-functions to expand.
176 * enhancement: step-instrumentation no longer wraps known
177 single-value functions with multiple-value context, allowing
178 better type inference.
179 * fixed bug #337: use of MAKE-METHOD in method combination now works
180 even in the presence of user-defined method classes. (reported by
181 Bruno Haible and Pascal Costanza)
182 * fixed bug #339(c): if there are applicable methods not part of any
183 long-form method-combination group, call INVALID-METHOD-ERROR.
184 (reported by Bruno Haible)
185 * fixed bug #361: the :FUNCTION initarg in the protocol for
186 initialization of methods can now be used to override
187 internally-produced optimized functions. (reported by Bruno
189 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
190 system-provided lambda expression no longer cause warnings about
191 unbound #:|pv-table| symbols.
192 * bug fix: improved the handling of type declarations and the
193 detection of violations for keyword arguments with non-constant
195 * bug fix: potentially erronous calls to PATHNAME and
196 MERGE-PATHNAMES were being flushed in some cases.
197 (reported by Richard Kreuter)
198 * bug fix: compiled calls to TYPEP were mishandling obsolete
199 instances. (reported by James Bielman and Attila Lendvai)
200 * bug fix: format strings with ~> without matching ~< no longer
201 trigger an AVER, but signal an understandable error instead.
202 (reported by Antonio Martinez)
203 * bug fix: specifying an output-file without a pathname-type for
204 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
205 of input-file instead of "fasl". (reported by Robert Dodier)
206 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
207 (reported by James Y Knight).
208 * bug fix: compiler-macros-function did not consider the environment
209 argument for shadowing by local functions.
210 * bug fix: compiler-macros expansion was inhibited by local INLINE
212 * bug fix: inline expansions of known functions were subject to
213 step-instrumentation in high debug policies, leading to problems
215 * bug fix: compiler failed to differentiate between different CONS
217 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
218 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
219 * bug fix: #368: incorrect use of expressed vs. upgraded array
221 * bug fix: #306a: more precise unions of array types.
222 * thread-safety improvements:
223 ** CONDITION-WAIT could return early on Linux, if the thread was
224 interrupted and subsequently continued with SIGCONT.
225 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
227 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
228 * added support for the ucs-2 external format. (contributed by Ivan
230 * minor incompatible change: pretty printing of objects of type
231 (cons symbol) is, in the default pprint-dispatch-table, now
232 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
234 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
235 called later in a class's lifetime, possibly as late as when the
236 first instance of the class is created. Previously,
237 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
238 class became finalizeable.
239 * fixed bug: FILE-POSITION sometimes returned inconsistent results
240 for multibyte external-format streams. (thanks to "vbzoli")
241 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
242 with :ALLOCATION :CLASS inherited from superclasses of the
244 * fixed bug: anonymous classes can now be created using the :NAME
245 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
246 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
247 * fixed bug: core-files saved with :EXECUTABLE T can again be
248 executed when SBCL_HOME isn't set. (reported by James Knight)
249 * fixed bug: toplevel LOCALLY forms with declarations could
250 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
251 * fixed bug: printing from several different threads using different
252 values of *print-case* could cause invalid output, due to
253 some internal special variables of the printer not being bound
254 thread-locally (reported by Max Mikhanosha)
255 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
256 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
257 (reported by Pascal Costanza)
258 * fixed bug: FINALIZE-INHERITANCE is called from
259 REINITIALIZE-INSTANCE on classes when the class has previously
260 been finalized, as required by AMOP.
261 * minor code generation optimizations:
262 ** better register allocation in CLOS dispatching functions
263 ** overflow detection when coercing signed bytes to fixnums on x86-64
264 ** is now implemented with one IMUL instruction instead of three shifts
265 ** more efficient bit-vector access on x86 and x86-64
266 ** more efficient access to raw structure slots on x86 and x86-64
267 * fixed some bugs revealed by Paul Dietz' test suite:
268 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
271 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
272 * feature: thread support on Solaris/x86, and experimental thread support
274 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
276 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
277 default method for SLOT-UNBOUND.
278 * minor incompatible change: prevent the user from specializing the
279 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
280 somewhat counter to the intent of the protocol, I (CSR) think, and
281 additionally it just doesn't work in SBCL as currently
282 implemented, thanks to optimizations (that are always valid for
283 the other three SLOT-VALUEish functions, but not for the setter).
284 * minor incompatible change: the :SB-LDB feature is now enabled by
285 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
286 the low-level debugger.
287 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
288 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
289 down to an SB-SYS:FD-STREAM, allowing direct communication in
290 more cases. Temporary files and pipes are still used as a fallback
292 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
293 on the non-reentrant C functions localtime(3) and gmtime(3).
294 * bug fix: global optimization policy was not visible in LOCALLY and
296 * bug fix: class objects can be used as specializers in methods.
297 (reported by Pascal Costanza)
298 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
299 with a type error. (reported by blitz_ on #lisp)
300 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
301 failed with a type error. (reported by Pascal Bourguignon)
302 * bug fix: merging pathnames against defaults with :DIRECTORY
303 starting with '(:RELATIVE :BACK) should preserve the :BACK.
304 (reported by James Y Knight)
305 * bug fix: saving large (>2GB) cores on x86-64 now works
306 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
307 constant index and a (simple-array (signed-byte 32)) array
308 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
309 error (patch by Robert J. Macomber)
310 * feature: TIME now displays the amount of run-time spent in GC
311 * bug fix: The debugger now does a better job of respecting
312 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
313 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
314 workaround for bug 403.)
315 * bug fix: floating point exception handling now works on FreeBSD
316 (thanks to NIIMI Satoshi)
317 * fixed some bugs revealed by Paul Dietz' test suite:
318 ** MISC.641: LET-conversion were not supposed to work in late
320 * improvements to the Win32/x86 port:
321 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
322 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
324 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
325 * new feature: source path information is generated for macro-expansion
326 errors for use in IDE's like Slime (thanks to Helmut Eller)
327 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
328 * bug fix: compiler does not loop forever on an invalid type in
330 * improvement: compilation of most CLOS applications is significantly
332 * optimization: added a limited bytecode compiler for simple toplevel
333 forms, speeding up compilation and FASL loading
334 * bug fix: the statistical profiler now properly distinguishes anonymous
337 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
338 * minor incompatible change: in sbcl-0.9.11 (but not earlier
339 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
340 into subclasses of the finalized class. Now user calls to
341 FINALIZE-INHERITANCE finalize just the one class, and calls by the
342 system return before any subclasses are finalized.
343 * minor incompatible change: The reader no longer ignores errors
344 regarding non-existent packages in #+ and #- feature tests.
345 * new feature: command line options --no-sysinit, --no-userinit to
346 inhibit loading the corresponding init files
347 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
348 for sending data through UDP sockets (thanks to François-René Rideau)
349 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
350 error is signalled instead (thanks to François-René Rideau)
351 * minor incompatible change: Error signalling behaviour of lexical
352 operations violating package locks has changed slightly. Refer to
353 documentation on package locks for details.
354 * enhancement: EVAL can process IF-expressions without resorting to the
356 * bug fix: Some operations involving SIN, COS, or TAN weren't being
357 constant-folded properly due to missing out-of-line functions.
358 (reported by Mika Pihlajamäki)
359 * bug fix: LISTEN sometimes returned T even in cases where no data was
360 immediately available from the stream
361 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
362 were reversed. (reported by Levente Mészáros)
363 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
364 when given a vector displaced to an adjustable array. (reported
366 * bug fix: derivation of float boundaries from numbers outside the
367 appropriate float range (reported by John Wiseman)
368 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
370 * bug fix: precomputing cache entries for generic functions with
371 some subclasses of specializers as yet invalid does not attempt to
372 fill a cache line with a negative offset. (reported by Levente
374 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
375 allow condition class objects as arguments to DOCUMENTATION and
376 (SETF DOCUMENTATION); only find and set documentation for
377 structure names for the STRUCTURE doc-type. (suggested by Gary
379 * improvements to the Win32/x86 port:
380 ** (user-homedir-pathname) and default initialization file
381 locations now know about the user's "Documents and Settings"
382 directory (thanks to Yaroslav Kavenchuk)
383 ** run-program is implemented (thanks to Mike Thomas)
384 ** sockets support (thanks to Timothy Ritchey)
385 ** better backtrace support (thanks to Alastair Bridgewater)
386 ** sb-grovel supported
387 ** asdf-install and sb-posix work somewhat
388 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
389 * improvements to the Solaris/x86 port:
390 ** works on Solaris 11/Solaris Express
391 ** floating-point exception handling support
392 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
393 * fixed some bugs revealed by Paul Dietz' test suite:
394 ** REMOVE-METHOD returns its generic function argument even when
395 the method is not one of the generic functions' methods.
396 ** objects declared as MEMBER types can be admissible arguments to
398 ** printing characters should simply be printed by the FORMAT ~:C
400 ** compiler failure when compiling functions with hairy constant
401 defaults for optional parameters.
402 ** compiler produces wrong code when MAYBE-INLINE-expanding a
403 function, which is already optimized.
405 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
406 * new platform: experimental support for SBCL x86/Darwin, including
407 MacOS X 10.4.5 on Intel.
408 * new feature: Unicode character names are now known to the system
409 (through CHAR-NAME and NAME-CHAR).
410 * new feature: the filesystem location of SBCL's core file is
411 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
412 * minor incompatible change: the contrib modules SB-POSIX and
413 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
414 this change is to make it easier to distribute
415 location-independent binaries.
416 * bug fix: as implied by AMOP, standardized classes no longer have
417 slots named by external symbols of public packages. (reported by
419 * optimization: calling functions via constant symbols -- as in
420 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
421 via the function object as in (FUNCALL #'FOO).
422 * optimization: on x86 and x86-64, the calling convention for the
423 case of dealing with an unknown number of values has been altered
424 to be friendlier to the prediction heuristics implemented,
425 particularly with reference to CALL and RET pairing. (thanks to
426 Alastair Bridgewater)
427 * optimization: on x86, the code for access to thread-local symbol
428 values and binding/unbinding of thread-local symbols is smaller.
429 (thanks to Alastair Bridgewater)
430 * enhancement: CONSTANTP is now able to determine constantness of
431 more complex forms, including calls to constant-foldable standardized
432 functions and some special forms beyond QUOTE.
433 * fixed bug: occasional GC crashes on Solaris/x86
434 * optimization: x86-64 supports stack allocation of results of simple
435 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
436 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
437 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
438 (thanks to James Knight)
439 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
440 by the standard (thanks to Stephen Compall)
442 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
443 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
444 be used for bundling the runtime and the core file into one
445 executable binary. This feature is not currently supported on all SBCL
446 platforms. (thanks to James Bielman and NIIMI Satoshi)
447 * new feature: a generational or ephemeral garbage collector is now
448 the default on the PowerPC platform (both Linux and Darwin). The
449 old Cheney (stop and copy) collector is a build-time option.
450 (thanks to Cyrus Harmon, after Raymond Toy)
451 * minor incompatible change: the method by which SBCL finds its
452 contributed modules has changed; it no longer relies on symbolic
453 links from an $SBCL_HOME/systems directory, but searches directly
454 in the subdirectories of $SBCL_HOME.
455 * enhancement: the dynamic heap size on the Linux/PPC platform is
456 markedly larger, even using the older Cheney garbage collector.
457 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
458 certain complicated string types. (reported by Gary King)
459 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
460 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
461 * fixed bug: DOCUMENTATION on structures no longer signals an
462 error if no structure type of the right name exists. (reported by
464 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
465 applicable non-standard primary or :AROUND method on
466 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
467 no longer cause unbound variable errors. (reported by Kevin Reid)
468 * fixed bug: metacircle resolution in cases where methods have slots
469 added before the slots from STANDARD-METHOD. (reported by Jean
471 * fixed bug: the Power PC assembly code for calling into lisp
472 sometimes computed its return address wrongly (depending on the
473 15th bit of the address, fixed at link time). This used to
474 manifest itself in a segmentation violation while building PCL.
475 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
476 many others over the years)
477 * fixed bug: ROOM no longer reports silly numbers for stack usage if
478 the stack pointer (treated as a signed integer) is negative.
479 (thanks to Peter van Eynde)
481 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
482 * new platform: experimental support for the Windows operating
483 system has been added. (thanks to Alastair Bridgewater)
484 * fixed several bugs in and robustified the PPC FFI (including
485 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
486 * bug fix: allow non-simple string symbol names (reported by Paul
488 * bug fix: interrupt handling on NetBSD (thanks to Richard M
490 * bug fix: saving a core corrupted callbacks on x86/x86-64
491 * bug fix: closed a loophole in metacircularity detection and
492 grounding in the PCL implementation of CLOS.
493 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
494 (reported by tomppa on #lisp)
495 * bug fix: clear the direction flag on Lisp -> C transitions, as
496 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
497 (reported by Andrej Grozin and Hendrik Maryns)
498 * optimization: major improvements to GC efficiency on GENCGC platforms
499 * optimization: faster implementation of EQUAL
500 * optimization: emit more efficient opcodes for some common
501 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
502 * optimization: several other minor code-generation improvements on x86-64
503 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
505 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
506 * minor incompatible change: (SETF CLASS-NAME) and (SETF
507 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
508 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
509 by ANSI as a generic function, and for consistency (SETF
510 GENERIC-FUNCTION-NAME) is treated likewise.
511 * fixed bug #233.b: make constraint propagation notice when a variable
512 value is changed after it is referenced but before it is used
513 * fixed bug #296: no more arbitrary behaviour from filenames with
514 odd characters as --load arguments. (reported by Adam Warner)
515 * fixed bug #390: :CHARACTER-SET pathname components now work as
516 expected. (reported by Tim Daly Jr)
517 * fixed bug #391: complicated :TYPE intersections in slot
518 definitions no longer cause an error in PCL internals.
519 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
520 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
521 returning the number of octets which would be written to the
522 file-stream. (thanks to Robert J. Macomber)
523 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
524 arguments to the internal stream functions. (thanks to David
526 * bug fix: stack exhaustion is now handled on the x86/SunOS
527 platform. (thanks to Daisuke Homma)
528 * bug fix: more accurate ROOM results on GENCGC platforms
529 * optimization: improved type inference for arithmetic-for
530 index variables in LOOP
531 * optimization: faster floating-point SQRT on x86-64
532 * fixed some bugs revealed by Paul Dietz' test suite:
533 ** DOCUMENTATION returns NIL instead of "" for method combinations
534 that don't have a docstring
536 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
537 * minor incompatible change: (SETF CLASS-NAME) and (SETF
538 GENERIC-FUNCTION-NAME) are no longer generic functions, and
539 therefore are not customizeable by user code (as seems to be at
540 least permitted and maybe required by AMOP). As a consolation,
541 however, the SBCL implementation of these functions now calls
542 REINITIALIZE-INSTANCE as specified by AMOP.
543 * bug fix: slot-definition documentation is propagated as per ANSI
544 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
545 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
546 argument, as do eslotd initialization methods. (from Pascal
547 Costanza's "Closer" project)
548 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
549 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
551 * bug fix: it is now possible to have more than one subclass of
552 STANDARD-GENERIC-FUNCTION without causing stack overflow.
553 (reported by Bruno Haible, Pascal Costanza and others)
554 * bug fix: the dependent update protocol now works for generic
555 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
557 * bug fix: condition-class instances corresponding to
558 DEFINE-CONDITION forms are now created eagerly. (reported by
559 Kalle Olavi Niemitalo on comp.lang.lisp)
560 * bug fix: floating point printing is more accurate in some
561 circumstances. (thanks to Simon Alexander)
562 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
563 merged with *DEFAULT-PATHNAME-DEFAULTS*.
564 * bug fix: callbacks on OS X now preserve stack-alignment.
565 (thanks to Cyrus Harmon)
566 * enhancement: the x86-64 disassembler is much better at
567 disassembling SSE instructions. (thanks to Lutz Euler)
568 * enhancement: CHECK-TYPE on a variable now allows type-inference
569 (thanks to Vincent Arkesteijn)
570 * optimization: improved performance of EUC-JP external format.
571 (thanks to NIIMI Satoshi)
572 * optimization: performance improvements to IO on file streams of
573 :ELEMENT-TYPE CHARACTER
574 * optimization: much faster memory allocation on x86-64
575 * optimization: faster garbage collections (latency and throughput)
577 * optimization: faster bignum subtraction and division on x86 and x86-64
578 (thanks to Lutz Euler)
579 * optimization: more accurate type-inference for loops that use a
580 floating point index variable or a negative step.
582 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
583 * bug fix: add a workaround to SBCL looping infinitely at startup on
584 Linux kernels with apparently buggy implementations of personality().
585 (thanks to Svein Ove Aas)
586 * bug fix: Unicode symbols are correctly printed in LDB backtraces
587 (thanks to David Lichteblau)
588 * bug fix: local bindings shadow global symbol macros for
589 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
590 * optimization: non-open coded uses of numeric comparison operators
591 (e.g. >) no longer cons when called with more than one parameter
592 on platforms supporting dynamic-extent allocation.
593 * enhancement: saving cores with foreign code loaded is now supported
594 on MIPS/Linux in addition to the previously supported platforms.
595 * bug fix: FIND-RESTART now tests for activity, not applicability when given
596 a restart object as identifier. (reported by Helmut Eller for CMUCL)
597 * bug fix: division by zero in sb-sprof when no samples were collected
598 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
599 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
600 (thanks to Alexander Kjeldaas)
601 * new feature: ignoring the return values of destructive functions
602 like NREVERSE signals a compile-time style-warning.
603 (thanks to Kevin Reid)
604 * new feature: the alignment of alien structure fields can be
605 explicitly specified. (thanks to Cyrus Harmon)
606 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
607 and dump core on SIGQUIT
609 ** incompatible change: threads do not inherit values of specials
610 from their parents (see manual)
611 ** bug fix: threads stacks belonging to dead threads are freed by the
612 next exiting thread, no need to gc to collect thread stacks anymore
613 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
614 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
615 ** bug fix: dynamic variable and thread start related gc lossage
616 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
618 * fixed some bugs revealed by Paul Dietz' test suite:
619 ** SUBTYPEP is slightly more accurate on heinously complicated
620 CONS types where some of the members have uncertain (in the
621 NIL, NIL sense) type relationships to each other.
623 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
624 no longer purified unless :PURIFY T is explicitly specified.
625 ** Non-purified cores are significantly smaller than before
627 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
628 * new feature: timers based on Zach Beane's excellent timer package
629 * added support for the following external formats: koi8-u,
630 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
631 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
632 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
633 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
634 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
635 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
636 * incompatible change: a threaded SBCL will no longer revert to
637 non-threaded mode on non-NPTL systems, but refuse to start entirely.
638 * bug fix: interrupts are disabled until startup is complete; no
639 more sigsegvs when receiving a signal to soon
640 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
642 * bug fix: add a workaround for the memory randomization features in
643 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
644 workaround will only be in effect on systems with the proc filesystem
646 * bug fix: printing objects of type HASH-TABLE signals a
647 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
649 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
650 have been running for over 50 days. (reported by Gilbert Baumann)
651 * bug fix: the logic for getting names of functions gets less
652 confused when confronded with alternate-metaclass
653 funcallable-instances. (reported by Cyrus Harmon)
654 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
655 consistent, even on internal alternate-metaclass objects.
656 * bug fix: flush closure information collected by physical
657 environment analysis prepass before the main pass. (bug reported
659 * bug fix: compiler pack phase does not modify a hashtable, which is
660 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
661 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
662 may be expanded. (reported by Alexander <malishev> on c.l.l)
663 * bug fix: no more style-warnings when compiling code using LAST
664 under high-SPEED low-SPACE settings. (reported by David Wallin
666 * fixed some incompatibilities between SBCL's MOP and the MOP
668 ** the METAOBJECT class is now implemented;
669 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
670 STANDARD-OBJECT, as required;
671 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
672 now compatible (as required by VALIDATE-SUPERCLASS); there
673 remains a consistency requirement over the presence or absence
674 of the FUNCTION class in the superclass list of finalized
675 classes; see the manual for more details;
676 ** the :AROUND method for COMPUTE-SLOTS on
677 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
678 requested slot ordering.
680 ** bug fix: parent thread now can be gc'ed even with a live
682 ** bug fix: binding a special with PROGV to no value is not
683 the same as that symbol not having been bound (thanks to
685 ** bug fix: binding specials is thread safe (thanks to
687 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
688 and SB-SPROF do not die with 'no handler for signal XX in
689 interrupt_handle_now(..)' anymore
690 ** bug fix: WITH-TIMEOUT works with multiple running threads
691 * fixed some bugs revealed by Paul Dietz' test suite:
692 ** ENSURE-GENERIC-FUNCTION should take a method class object for
693 the :method-class keyword argument.
695 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
696 * new port: the Solaris operating system on x86 processors is now
697 mostly supported, though some rough edges in the environment
698 remain. (thanks to Daisuke Homma)
699 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
700 space for its heap. (thanks to Thiemo Seufer)
701 * optimization: SBCL on MIPS platforms now supports dynamic-extent
702 closures. (thanks to Thiemo Seufer)
703 * minor incompatible change: eof selects abort in the debugger.
704 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
705 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
707 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
708 deprecated, and will go away in a future revision of SBCL.
709 * minor incompatible change: GC-ON and GC-OFF are no longer
710 implemented with a counter, it does not matter how many times gc
711 is switched on or off
712 * bug fix: discriminating functions for generic function classes
713 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
714 make invalid assumptions about method precedence order. (reported
716 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
717 platforms. (thanks to Thiemo Seufer)
718 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
719 * bug fix: correct pathname printing with printer escaping is on.
720 (thanks to Kevin Reid)
721 * bug fix: complex VOP definitions in "user-space" no longer trigger
722 package locks. (reported by Zach Beane)
723 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
724 for funcallable-instance functions, and is no different from
726 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
727 characters in the current readtable. (reported by Nicholas Neuss)
728 * bug fix: SUBTYPEP on various CONS types returns more a more
729 accurate acknowledgment of its certainty.
730 * optimizations: REMOVE-DUPLICATES now runs in linear time on
731 lists in some cases. This partially fixes bug 384.
732 * flush all standard streams before prompting in the REPL and the
734 * bug fix: signal handling and triggering gc do not conflict
735 directly anymore, in particular a high frequency sb-sprof does
736 not prevent gc from running
737 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
738 approximation for timezone and DST information between the
739 universal time epoch and the smallest negative 32-bit time_t.
740 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
741 year 1899 so long as the encoded time is non-negative.
742 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
743 the correct answer for non-constant rotate amounts when performing
744 an inline 32-bit rotation.
746 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
747 there is only one thread in the session
748 ** bug fix: memory leak for streams created in one thread and
749 written to in another
750 ** bug fix: lockup when compiled with gcc4
751 ** bug fix: race that allows the gc to be triggered when gc is
753 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
754 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
756 ** bug fix: finished threads are gc'ed properly
757 * fixed some bugs revealed by Paul Dietz' test suite:
758 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
759 has arguments with a different set of applicable methods from
760 the orignal arguments.
761 ** The type error thrown by MAP now has a correct expected-type
763 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
764 name a compiled function.
765 ** Three MISC tests where a large bignum was improperly coerced to
766 a float that couldn't represent that bignum during type
767 derivation were fixed.
768 ** SUBTYPEP can now handle types involving the negation of a
769 list-form FUNCTION type.
770 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
771 as (COMPLEX (AND RATIO (NOT FIXNUM))).
772 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
774 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
775 * New feature: Experimental support for bivalent streams: streams
776 opened with :element-type :default now allow character and binary
777 (unsigned-byte 8) I/O
778 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
779 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
781 * bug fix: no more highly sporadic "couldn't check whether ~S is
782 readable" when reading a stream and an interrupt hits in the middle
783 of a select system call
784 * compiler better recognizes complex arrays (reported by Eduardo
786 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
787 for variables on SBCL built with linkage-tables. (reported by Luis
789 * various error reporting improvements.
790 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
791 (Thanks to Thiemo Seufer)
792 * bug fix: nested reader invokations work correctly
793 * bug fix: it is possible to have simultaneous references to foreign
794 code and foreign data with the same name.
796 ** added x86-64 support
797 ** incompatible change: the threading api now works with thread
798 objects instead of thread ids
799 ** bug fix: threads are protected from signals and interruption when
800 starting up or going down
801 ** bug fix: a race where an exiting thread could lose its stack to gc
802 ** bug fix: don't halt on infinite error in threads if possible
803 ** fixed numerous gc deadlocks introduced in the pthread merge
804 ** bug fix: fixed thread safety issues in read and print
805 ** bug fix: debugger doesn't hang on session lock if interrupted at
806 an inappropriate moment
807 ** bug fix: run-program is now thread safe(r)
808 ** bug fix: inner with-recursive-lock no longer releases the mutex
809 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
810 * fixed some bugs revealed by Paul Dietz' test suite:
811 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
812 EXPECTED-TYPE slots filled.
813 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
815 ** MAKE-CONDITION accepts classes as type-designators.
816 ** COMPILE may never return NIL.
817 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
818 range before calling Unix time functions
820 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
821 * numerous signal handling fixes to increase stability
822 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
823 * minor incompatible change: we now correctly canonize default
824 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
825 as per the MOP, rather than the historical (INITARG INITFUNCTION
826 INITFORM). (reported by Bruno Haible)
827 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
828 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
829 for more information.
830 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
831 pathname is a directory pathname.
832 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
833 * dynamic space size on PPC has been increased to 768Mb. (thanks to
835 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
836 :METACLASS argument in addition to a class name. (reported by
837 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
838 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
839 (reported by Bruno Haible; patch from Gerd Moellmann)
840 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
842 * bug fix: more cleanups to the floating point exception handling on
843 x86-64 (thanks to James Knight)
844 * bug fix: the compiler does not try to constant fold calls of
845 COERCE and BIT-* functions when they return freshly constructed
846 objects. (reported by Nikodemus Siivola and Paul Dietz)
847 * optimization: improved the MIPS versions of generic (in the
848 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
849 * optimization: direct conversion of (unsigned-byte 32) to floats on
850 the PowerPC platform.
851 * optimization: structure instances with raw slots now use less
852 memory, and probably show better memory locality. (thanks to
854 * optimization: DYNAMIC-EXTENT declarations for lists and closures
855 are treated as requests for stack allocation on the x86-64,
856 Alpha-32, PPC and SPARC platforms.
857 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
858 the wrong file descriptor; implementation of SOCKET-OPEN-P.
859 (thanks to Tony Martinez)
861 ** gcing a dead thread can no longer lead to lockups
862 ** threads block signals until they are set up properly
863 ** errno is no longer shared by threads
864 ** interrupt-thread restores the eflags register on x86
865 ** fixed some lockups due to gc/thread interaction
866 * fixed some bugs revealed by Paul Dietz' test suite:
867 ** invalid dotted lists no longer raise a read error when
869 ** signal an error if a symbol that names a declaration is used
870 as the name of a type, or vice versa
871 ** allow using the (declare (typespec var*)) abbreviation for
872 (declare (type typespec var*)) with all type specifiers
873 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
874 ** always return NIL from PPRINT-POP when OBJECT is NIL
875 ** don't signal errors when pretty-printing malformed LABELS,
876 FLET or MACROLET forms
877 ** declarations in a DOLIST body are also in scope for the
879 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
881 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
884 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
885 * fixed cross-compiler leakages that prevented building a 32-bit
886 target with a 64-bit host compiler.
887 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
888 opened with :IF-EXISTS :OVERWRITE.
889 * fixed bug 281: error for an invalid qualifier in a short-form method
890 combination method is not signalled until the faulty method is called.
891 * bug fix: iteration variable type inferrer failed to deal with open
892 intervals. (reported by Alan Shields)
893 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
894 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
895 or more runtime options were provided to the sbcl binary.
896 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
897 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
898 * optimization: there's now a fast-path for fixnum arguments in the
899 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
900 * optimization: the code generated on x86-64 is more compact thanks
901 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
902 * bug fix: floating-point exception handling now partly works on
903 x86-64. (Thanks to James Knight)
904 * improvement to the MIPS backend from Thiemo Seufer: C-style
905 64-bit long long arguments and return values to alien functions
907 * fixed some bugs revealed by Paul Dietz' test suite:
908 ** the type-error signalled from WARN has a filled-in DATUM slot.
909 ** the type-error required when a stream is not associated with
910 a file has the stream as its datum.
911 ** type-errors on single-floats on x86-64 no longer have
912 :INVALID-OBJECT as the datum
913 ** the type-errors signalled for invalid function names now have
914 a correct expected type
915 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
916 for structures defined with a :TYPE. Documentation strings for
917 typed structures are no longer immediately discarded
918 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
919 broadcast streams and synonym streams. FILE-LENGTH now also works
920 on broadcast streams.
922 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
923 * incompatible change: the --noprogrammer option, deprecated since
924 version 0.7.5, has been removed. Please use the equivalent
925 --disable-debugger option instead.
926 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
928 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
929 inoperational for a while now) has been completely removed.
930 * null lexical environments are now printed as #<NULL-LEXENV>,
931 significantly reducing the amount of clutter in typical backtraces.
932 * documentation on weak pointers, finalization, and after GC hooks
933 has been added to the manual.
934 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
935 as other specialized array types, is much faster. SUBSEQ and
936 COPY-SEQ on such arrays have also been sped up.
937 * optimization: EQL is now more efficient when at least other argument
938 is known to be of type (OR FIXNUM (NOT NUMBER)).
939 * fixed bug: the runtime is now less vulnerable to changes in the
940 size of the SBCL object on OS X, and virtual memory is reserved for
941 all spaces that need to be at a fixed address.
942 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
943 * fixed bug: finalizers and after GC hooks that cause consing are now
945 * fixed bug: compiler error messages and summaries are now printed to
946 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
947 * fixed inference of the upper bound of an iteration variable.
948 (reported by Rajat Datta).
949 * fixed bug 211e: calling local functions with duplicated constant
950 keyword argument no longer causes a bogus style warning about an
952 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
953 ftype declaration to be disregarded. (reported by Dave Roberts)
954 * fixed bug 373: caused by erronous compilation of references to alien
955 variables in the runtime on ppc/darwin.
956 * fixed bug 376: CONJUGATE type deriver.
957 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
958 (reported by Baughn on #lisp)
959 * a cross-compiler bug on non-x86 platforms has been identified and
960 fixed. (thanks to Bruno Haible)
961 * improvements to the MIPS runtime code for increased stability.
962 (thanks to Thiemo Seufer)
963 * increased the maximimum compact environment size to allow
964 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
965 * partial workaround for bug 135: don't name the function we're
966 calling for hairy cases of EVAL, so as not to accumulate one environment
967 entry per call to EVAL. (reported by Kevin Reid)
968 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
969 * optimization: type testing for non-vector arrays should be faster.
970 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
972 * bug fix: setting 31st element of a bit vector to zero did not work
974 * bug fix: redefining a class definition which failed due to a
975 previous accessor / function clash now works (but see BUGS entry
976 #380 for more problems in this area). (thanks to Zach Beane)
977 * the long form of DEFINE-METHOD-COMBINATION disables method group
978 checking when given a single method group with pattern *. (thanks
980 * on x86 compiler supports stack allocation of results of simple
981 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
982 * support for building and running on Mac OS X 10.4 "Tiger" has been added
983 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
984 * fixed some bugs related to Unicode integration:
985 ** the restarts for recovering from input and output encoding
986 errors only appear when there is in fact such an error to
988 * fixed some bugs revealed by Paul Dietz' test suite:
989 ** MISC.549 and similar: late transformation of unsafe type
990 assertions into derived types caused unexpected code
992 ** SCALE-FLOAT type deriver is less wrong.
993 ** type derivers for EXP, LOG and similar functions compute result
994 types for complex arguments better.
995 ** (MISC.563) CONJUGATE type deriver works for very restricted
997 ** out-of-line type testers for character strings are available.
998 ** EQUAL compiler transform understands specialness of objects
1000 ** accessing double-floats stored on the stack now works on x86-64.
1001 ** debugger internals could sometimes create invalid lispobjs,
1002 resulting in GC crashes.
1003 ** MISC.548: type check weakening can convert required type into
1005 ** initialization forms for bindings are not in scope of free special
1008 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1009 * incompatible change: thread support for non-NPTL systems has
1010 been removed - locking is buggy and unreliable. A threaded
1011 SBCL build will now warn at startup and refuse to create
1012 new threads, unless futex support is detected in the kernel
1013 * incompatible change: the top level REPL now has only an ABORT
1014 restart associated with it, not TOPLEVEL and ABORT as it used to.
1015 TOP and TOPLEVEL are now available as debugger commands for
1016 returning to the top level.
1017 * incompatible change: forms evaluated in the REPL now use the
1018 global optimization policy.
1019 * incompatible change: user- and system-initialization files are
1020 no longer processed with LOAD, but by READ and EVAL; hence the
1021 global optimization policy, startup package, readtable, etc,
1023 * The .fasl file format number has been incremented because of
1024 various incompatible changes.
1025 * internal entry point details and argument counts no longer appear
1026 in backtraces unless explicitly requested by setting
1027 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1028 * built-in and standard functions no longer have names like "top
1029 level local call to FOO".
1030 * fixed bug 32: functions defined in non-null lexical environments
1031 now have more legible printed representation
1032 * fixed bug 33: functions defined in non-null lexical environemnts
1033 are now more amenable to inspection by INSPECT.
1034 * workaround for bug 354: XEPs no longer appear in backtraces unless
1035 explicitly requested.
1036 * fixed bug: receiving the signal which results from attempting to
1037 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1038 SIGBUS on other *BSDs) on architectures where the C stack is also
1039 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1040 to debugger confusion or wild instability. (reported by Cheuksan
1042 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1043 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1044 (reported by Lutz Euler)
1045 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1046 compiled code. (reported by Paul Dietz)
1047 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1048 specialized parameter to a method to something that is not TYPEP
1049 the specializer is now possible.
1050 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1051 face of package deletion.
1052 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1053 pathname. (reported by David Tolpin on c.l.l)
1054 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1055 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1056 STRING-STREAM and FILE-STREAM, respectively.
1057 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1058 than previously; in addition, its test suite is now run on build.
1059 (thanks to Luke Gorrie)
1060 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1061 * optimization: added a immediate representation for single-floats
1063 * fixed some bugs revealed by Paul Dietz' test suite:
1064 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1065 ** attempting to create a package with a colliding nickname causes
1066 correctable errors to be signalled.
1067 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1068 ** several x86-64 backend bugs related to sign-extension of immediate
1071 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1072 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1073 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1075 * fixed loading of multiply forward-referenced layouts.
1076 (thanks to Cheuksan Wang)
1077 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1078 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1079 (see COMPILE-FILE documentation for details of :PRINT option)
1080 * fixed bugs 19 and 317: fixed-format floating point printing is
1081 more accurate. This also fixes a bug reported by Adam Warner
1082 related to the ~@F format directive.
1083 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1085 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1086 dispatch table if the from-char is a dispatch macro character.
1087 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1088 by the wordsize no longer ignore the last word. (reported by Lutz
1090 * fixed bug in type checking of dynamic-extent variables. (reported
1092 * optimization: sequence traversal functions use their freedom to
1093 coerce function designators to functions.
1094 * optimization: code with many calls to CLOS methods specialized on
1095 CLOS classes has had redundant type checks removed.
1096 * optimization: added declarations to speed up operations that access
1097 the internal character database (for example STRING-UPCASE)
1098 * optimization: comparison operations between floats and sufficiently small
1099 fixnums no longer create extra rationals
1100 * fixed some bugs related to Unicode integration:
1101 ** portions of multibyte characters at the end of buffers for
1102 character-based file input are correctly transferred to the
1103 start of the buffer at the next read.
1104 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1105 passing it through to OPEN.
1106 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1107 argument given to internal calls to OPEN.
1108 ** the built-in comment readers (introduced by character sequences
1109 ";" and "#|") are more forgiving to encoding errors; they will
1110 STYLE-WARN and then attempt to resync the stream at a character
1111 boundary. (thanks to Teemu Kalvas)
1112 * fixed some bugs in the x86-64 port:
1113 ** Negative short int return values from c-calls are sign-extended
1115 ** The stack is aligned to 16-bytes for c-calls, as required by
1116 the ABI. (thanks to Cheuksan Wang)
1117 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1118 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1119 for enums. (thanks to Vincent Arkesteijn)
1120 ** Multiple small optimizations and bugfixes for floating point
1122 * fixed some bugs revealed by Paul Dietz' test suite:
1123 ** Space, Tab, Linefeed, Return and Page have the invalid
1124 secondary constituent character trait.
1125 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1127 ** WITH-INPUT-FROM-STRING should only update the index place on
1129 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1130 works more reliably.
1131 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1132 with one-dimensional arrays with fill pointers.
1133 ** TYPE-OF failed on a complex with an integer realpart and a
1135 ** compiler failure during type inference for the code of form
1136 (IF (EQL X (THE ...)) ...) (MISC.535).
1138 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1139 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1140 platform. The port supports 61-bit fixnums, large memory spaces
1141 and reloading shared object files.
1142 * enhancement: saving cores with foreign code loaded is now
1143 supported on ppc/Darwin in addition to the previously supported
1145 * enhancement: the statistical profiler now walks deeper into the
1146 call stack for more meaningful call-graphs and accrued time
1147 reports (x86/x86-64 only). It also now reports time spent in
1149 * enhancement: it is now possible to trace most individual methods
1150 of a generic function in addition to tracing the generic function
1152 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1154 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1155 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1156 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1157 produces an error. (thanks to Vincent Arkesteijn)
1158 * bug fix: NAMESTRING on pathnames with :WILD components in their
1159 directories works correctly. (thanks to Artem V. Andreev)
1160 * fixed bug 125: compiler preserves identity of closures. (reported
1162 * bug fixed: functions with &REST arguments sometimes failed with
1163 "Undefined function" when compiled with (DEBUG 3). (reported by
1165 * bug fix: overflow during compiling of setting element of a bit
1166 vector with constant index and value. (reported by Timmy Douglas)
1167 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1168 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1169 * on x86 compiler partially supports stack allocation of dynamic-extent
1171 * GO and RETURN-FROM do not check the extent of their exit points
1172 when compiled with SAFETY 0.
1173 * fixed some bugs related to Unicode integration:
1174 ** encoding and decoding errors are now much more robustly
1175 handled; it should now be possible to recover even from invalid
1176 input or output to the terminal. (thanks to Teemu Kalvas)
1177 ** provided a first cut at implementing STRING-TO-OCTETS and
1178 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1179 ** altered the SB-MD5 contributed module slightly, changing the
1180 interface just enough for it to be supportable for builds where
1181 lisp characters are not eight bits.
1182 * fixed some bugs revealed by Paul Dietz' test suite:
1183 ** the FORMATTER-generated functions for ~V[ conditionals require
1184 the correct number of arguments.
1185 ** READ-FROM-STRING returns the mandated second value when applied
1186 to displaced strings.
1187 ** the #\Rubout and #\Backspace characters are treated as invalid
1188 constituent characters by the tokenizer.
1190 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1191 * new feature: reloading changed shared object files with
1192 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1193 * new feature: references to foreign variables and functions
1194 can now be compiled and loaded before the corresponding shared
1195 object file is loaded, as long as the foreign definitions are
1196 available at runtime.
1197 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1198 supported. (thanks to Dan Debertin)
1199 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1200 just keywords. (thanks to Vincent Arkesteijn)
1201 * fixed bug #331: structure-class instances corresponding to
1202 DEFSTRUCT forms are now created eagerly.
1203 * fixed bug #345: backtraces from calls to undefined functions work
1204 on x86 as well. Related bug #61 is now also partially fixed on x86
1205 (backtraces from throws to unknown catch tags.)
1206 * bug fix: lambda-list parsing is now stricter vrt. order and number
1207 of lambda-list keywords.
1208 * bug fix: as specified by AMOP, an error is signalled if a
1209 class-option appears multiple times in a DEFCLASS form. (reported
1211 * bug fix: RANDOM can be compiled when the compiler derives the type
1212 of its numeric argument as a disjoint set of small integers.
1213 (reported by Paul Dietz)
1214 * bug fix: starting a core saved with shared objects loaded when
1215 those objects are not available no longer causes threaded SBCL to
1216 hang. (reported by Sean Ross)
1217 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1219 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1220 argument, which must always be NIL. (reported by Kalle Niemitalo)
1221 * bug fix: printing 1.0d+23 no longer results in an error.
1222 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1223 * bug fix: structure slot setters preserve evaluation order. (thanks
1225 * fixed some bugs related to Unicode integration:
1226 ** RUN-PROGRAM can allow its child to take input from a Lisp
1227 stream. (reported by Stefan Scholl)
1228 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1229 locales. (reported by Ken Causey)
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** INCF, DECF and REMF evaluate their place form as specified in
1233 ** FORMATTER expands ~{ ~} iteration directives with V or #
1234 parameters correctly.
1235 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1236 consequent uses no arguments correctly.
1237 ** the system has a partial understanding of the (COMPLEX RATIO)
1239 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1242 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1243 * new feature: a build-time option (controlled by the :SB-UNICODE
1244 keyword feature, enabled by default) for building the system with
1245 support for the entire 21-bit character space defined by the
1247 * new feature: the system now has rudimentary external-format
1248 support; the primary user-visible change at this time is that
1249 characters with the high bit set (such as lower-case-e-acute) will
1250 print correctly to a terminal in a UTF-8 environment.
1251 * minor incompatible change: BASE-CHAR no longer names a class;
1252 however, CHARACTER continues to do so, as required by ANSI.
1253 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1254 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1255 has been moved to the SB-EXT package (temporarily re-exported from
1257 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1258 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1260 * fixed bug #351: better error-handlind and reporting for malformed
1262 * fixed bug #350: bignum-printing is now more memory-efficient,
1263 allowing printing of very large bignums, eg. (expt 2 10000000).
1264 (reported by Bruno Haible)
1265 * fixed bug #302: better primitive-type selection for intersection
1267 * fixed bug #308: non-graphic characters now all have names, as
1268 required. (reported by Bruno Haible)
1269 * bug fix: redefining a class with different superclasses now correctly
1270 removes it from the direct-subclasses of its previous superclasses.
1271 (reported by David Morse)
1272 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1273 new value now works. (reported by Bruno Haible)
1274 * bug fix: correct canonicalization of multiple non-standard slot
1275 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1276 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1277 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1278 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1279 now exists, an signals an error.
1280 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1281 during macroexpansion for non-iteration variables as well. (reported
1282 by Bruno Haible for CMUCL)
1283 * bug fix: Cyclic structures and unprintable objects in compiler
1284 messages no longer cause errors. (reported by Bruno Haible)
1285 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1286 and READ-FROM-STRING all now return a primary value of NIL if
1287 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1288 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1289 (reported by Bruno Haible for CMUCL)
1290 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1291 do not designate a condition. (reported by Bruno Haible for
1293 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1294 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1295 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1296 permitting resolution in favour of any of the conflicting symbols.
1297 (reported by Bruno Haible for CMUCL)
1298 * bug fix: EQUAL compiler optimizations is less aggressive on
1299 strings which can potentially compare true despite having distinct
1300 specialized array element types.
1301 * bug fix: unit enumerations can be defined without dividing by
1302 zero. (thanks to Vincent Arkesteijn)
1303 * FORMAT compile-time argument count checking has been enhanced.
1304 (report from Bruno Haible for CMUCL)
1305 * a partial workaround for the bug 262: the compiler does not try to
1306 inline-expand a local function doing RETURN-FROM from a deleted
1307 BLOCK. (thanks to Peter Denno for the bug report and to David
1308 Wragg for the simple test case)
1309 * fixed some bugs revealed by Paul Dietz' test suite:
1310 ** RENAME-PACKAGE allows all package designators as new package
1312 ** constraint propagation and lambda variable substitution are
1313 more cautious in dealing with partially deleted code.
1314 ** compiler sometimes lost reoptimization passes.
1315 ** CERROR, given a condition as condition designator, passes its
1316 remaining arguments to the continue format control without
1318 ** Case-altering FORMAT directives work correctly on non-ASCII
1320 ** The REAL type specifier handles bounds outside the floating
1321 point ranges without signalling FLOATING-POINT-OVERFLOW.
1322 ** Functions with IR1-transformations can create intercomponent
1323 references to global functions.
1324 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1326 ** FORMAT ~R treats a nil value for its first parameter correctly.
1328 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1329 * enhancement: saving cores with foreign code loaded is now
1330 supported on x86/NetBSD and sparc/Linux in addition to the previously
1331 supported platforms.
1332 * bug fix: on some platforms repeated installations caused multiple
1333 copies of HTML documentation to be installed -- should not happen
1334 any more. (reported by Stefan Scholl)
1335 * bug fix: parsing self-recursive alien record types multiple times
1336 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1337 original patch by Helmut Eller for CMUCL)
1338 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1339 (thanks to Richard Kreuter)
1340 * bug fix: defining classes whose accessors are methods on existing
1341 generic functions in other (locked) packages no longer signals
1342 bogus package lock violations. (reported by François-René Rideau)
1343 * bug fix: special variables as DEFMETHOD parameters no longer have
1344 associated bogus type declarations. (reported by David Wragg and
1346 * bug fix: read-write consistency on streams of element-type
1347 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1348 * bug fix: redefiniton of the only method of a generic function with
1349 no DEFGENERIC no longer emits a full WARNING. In addition,
1350 redefinition of generic functions with no DEFGENERIC to an
1351 incompatible lambda list now signals an error. (thanks to Zach
1353 * bug fix: DEFGENERIC now works even when there's a function of the
1354 same name in an enclosing lexical environment. (thanks to Zach
1356 * fixed compiler failure, caused by instrumenting code during
1357 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1358 * optimization: added loop analysis and improved register allocation
1359 * fixed some bugs revealed by Paul Dietz' test suite:
1360 ** POSITION on displaced vectors with non-zero displacement
1361 returns the right answer.
1362 ** (SIMPLE-STRING) is a valid type specifier for sequence
1364 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1366 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1367 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1369 ** The justification version of the FORMAT ~< directive treats
1370 non-zero minpad parameter correctly.
1372 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1373 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1374 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1375 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1376 the supported interface.
1377 * new feature: Single-stepping of code compiled with DEBUG 2 or
1378 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1379 * new feature: saving cores with foreign code loaded is now
1380 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1381 Timothy Moore's work for CMUCL)
1382 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1383 parameters to * instead of NIL if no initform is supplied.
1384 (reported by Johan Bockgaard)
1385 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1386 correctly when FUNCALL appears as the car of the form. Note:
1387 despite this FUNCALL forms are not currently subject to
1388 compiler-macro expansion. (port of Raymond Toy's fix for the
1389 same from CMUCL, reported by Johan Bockgaard)
1390 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1391 (thanks for Teemu Kalvas)
1392 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1393 case instead of terminating the iteration (thanks for Julian
1394 Squires, Sean Champ and Raymond Toy)
1395 * bug fix: incorrect expansion of defgeneric that caused a style
1396 warning. (thanks for Zach Beane)
1397 * bug fix: slot accessor effective method computation works properly
1398 for classes with multiple non-standard applicable methods on
1399 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1400 * on x86 compiler supports stack allocation of results of LIST and
1401 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1402 CMUCL implementation by Gerd Moellmann)
1403 * fixed some bugs revealed by Paul Dietz' test suite:
1404 ** FORMAT strings with both the ~<~:;~> form of the justification
1405 directive and pretty-printing directives cause an error.
1407 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1408 * incompatible change: the internal functions
1409 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1410 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1411 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1412 instead of the old functions.
1413 * new feature: on platforms where "dladdr" is available foreign
1414 function names now appear in backtraces. (based on Helmut Eller's
1416 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1417 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1419 * documentation: ASDF manual is now distributed with SBCL.
1420 * bug fix: SBCL can now load its contributed modules with REQUIRE
1421 even if the system-provided entries have been removed from
1422 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1424 * bug fix: backtraces involving undefined functions or assembly
1425 routines are more informative. (thanks to Brian Downing)
1426 * bug fix: mutually referent alien structures now work correctly.
1427 (reported by Rick Taube)
1428 * bug fix: structures defined by WITH-ALIEN can be referred to
1429 within other definitions in the same WITH-ALIEN.
1430 * bug fix: division operators (MOD, TRUNCATE and the like) with
1431 constant zero divisors and integer dividends no longer generate
1433 * bug fix: provide default methods for INPUT-STREAM-P and
1434 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1435 * bug fix: improve the detection and resolution of MOP metacycles
1436 (where effective method computation depends on the generic function
1437 having its effective method computed). (reported by Bruno Haible)
1438 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1439 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1440 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1441 * bug fix: pathnames with relative directory components can be
1442 represented relative to default pathnames.
1443 * optimization: in taking the GCD of bignums, reduce the two bignums
1444 to approximately the same size (using Euclid's algorithm) before
1445 applying the more sophisticated binary GCD. (thanks to Juho
1447 * optimization: COUNT on bitvectors now operates word-at-a-time.
1448 * optimization: ASH with a positive, but not necessarily constant,
1449 (leftwards) shift, when in a modular context, is compiled to a
1451 * fixed some bugs revealed by Paul Dietz' test suite:
1452 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1453 if the corresponding argument is NIL.
1454 ** FORMAT directives accepting a minpad parameter treat negative
1456 ** literal commas as character parameters to FORMAT directives are
1458 ** literal spaces directly after ~<Newline> directives within a
1459 format-logical-block (~:< ~@:>) do not induce :FILL-style
1460 conditional newlines.
1461 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1462 as specified: it no longer includes conditional newlines).
1463 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1465 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1466 :PER-LINE-PREFIX argument does not evaluate to a string.
1468 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1469 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1470 the manual for details; for now, package locks can be disabled by
1471 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1472 if no major problems are found then it is likely that they will be
1473 compiled in unconditionally.
1474 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1475 unsupported operators on all platforms. To load a shared library
1476 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1477 object file, link it into a shared library outside of SBCL and
1478 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1480 * bug workaround: ROOM T and the GC were not working together
1481 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1482 reporting facility aren't preserved. That reporting has been
1483 disabled, so now until and unless someone figures out how to make it
1484 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1485 * minor incompatible change: as threatened around sbcl-0.8.0, the
1486 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1487 The symbols are also exported from SB-PCL for backwards
1488 compatibility, but more so than before SB-PCL should be treated as
1489 an implementation-internal package.
1490 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1492 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1493 It is known as of this release not to work on the Alpha, however.
1494 * fixed bug #167: errors signalled due to illegal syntax in method
1495 bodies are now more legible.
1496 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1497 designators and can hence be used with TYPEP.
1498 * fixed bug #333: CHECK-TYPE now ensures that the type error
1499 signalled, if any, has the right object to be accessed by
1500 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1501 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1502 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1504 * fixed bug #334: programmatic addition of slots using specialized
1505 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1506 and :ALLOCATION :CLASS slots.
1507 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1508 float exponents. (rereported by Peter Seibel)
1509 * fixed a bug: #\Space (and other whitespace characters) are no
1510 longer considered to be macro characters in standard syntax by
1511 GET-MACRO-CHARACTER.
1512 * fixed bug: initialization of condition class metaobjects no longer
1513 causes an instance of the condition to be created. (reported by
1515 * fixed bug: it is now possible to have slots such that
1516 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1517 description is neither :INSTANCE nor :CLASS.
1518 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1519 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1520 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1521 non-standard SLOT-DEFINITION-ALLOCATION values.
1522 * fixed bugs: various race conditions handling exiting threads.
1523 CL-PPCRE's thread torture test now passes.
1524 * fixed arguably-a-bug: GC time proportional to number of threads in
1525 system even when most of them are idle
1526 * optimization: improved performance of BIT and SBIT on bit-vectors.
1527 * .fasl file incompatibility: The fasl file version number has
1528 been incremented because of changes associated with package locks.
1530 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1531 * minor incompatible change: the system no longer provides
1532 optimization hints (conditions which are TYPEP
1533 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1534 mode; these hints are still emitted when higher SPEED optimization
1536 * new contrib module: a sampling profiler (profiling by statistical
1537 sampling, rather than by instrumenting functions) is available as
1538 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1539 * the behaviour of the standard function ED is now customizeable by
1540 third parties through a hook variable: see ED's documentation
1541 string for information on the protocol.
1542 * the compiler no longer emits efficiency notes for (FUNCALL X)
1543 when the type of X is uncertain under default optimization
1545 * fixed bug 276: mutating a binding of a specialized parameter to a
1546 method to something that is not TYPEP the specializer is now
1548 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1549 exhibit the required behaviour on the x86 platform. (thanks to
1550 Peter van Eynde, Eric Marsden and Bruno Haible)
1551 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1552 even for difficult arguments. (reported by Peter Graves)
1553 * fixed bug 141a: the backquote printer now descends quoted
1555 * fixed another bug in backquote printing: no more destructive
1556 modification of the form's list structure. (reported by Brian
1558 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1559 move between its address being taken and the call to
1560 interrupt_thread, fixing a crashing race condition.
1561 * the SB-POSIX contrib implementation has been adjusted so that it
1562 no longer exhibits ridiculously poor performance when constructing
1563 instances corresponding to C structs.
1565 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1566 * minor incompatible change: the sb-grovel contrib now treats C
1567 structures as alien (in the sense of SB-ALIEN) objects rather than
1568 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1569 has implications for memory management of client code
1570 (sb-grovel-returned objects must now be manually managed) and for
1571 type safety (alien objects now have full types).
1572 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1573 used to control emission of compiler diagnostics, rather than the
1574 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1575 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1576 quality should be considered deprecated.
1577 * install.sh now installs the user manual as well
1578 * (not quite a new documentable feature, but worth considering in
1579 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1580 beginnings of a semantically meaningful condition hierarchy is
1581 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1583 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1584 designator as the defaults argument.
1585 * fixed bug: Displaced arrays whose displaced-to array has become
1586 too small now cause ARRAY-DIMENSION to signal an error, providing
1587 for safer bounds-checking. (reported by Bruno Haible)
1588 * fixed bug: DEFCLASS slot definitions with identical :READER and
1589 :WRITER names now signal a reasonable error. (reported by Thomas
1591 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1593 * fixed bug: Invalid slot specification errors now print correctly.
1594 (thanks to Zach Beane)
1595 * fixed bug 320: Shared to local slot value transfers in class
1596 redefinitions now happen corrently from superclasses as
1597 well. (reported by Bruno Haible)
1598 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1600 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1601 improper lists correctly. (reported by Bruno Haible)
1602 * fixed bug 313: source-transform for <fun-name> was erroneously
1603 applied to a call of a value of a variable with name <fun-name>.
1604 (reported by Antonio Menezes Leitao)
1605 * fixed bug 307: The obsolete instance protocol ensures that
1606 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1607 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1608 remove dead unknown-values globs from the middle of the stack.
1609 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1610 easier to resume long computations after using *BREAK-ON-SIGNALS*
1611 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1612 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1613 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1614 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1615 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1616 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1618 * fixed a (fairly theoretical) bug in string printing: if
1619 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1620 does not have array-element-type equal to the most general string
1622 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1623 function name. (thanks to Nikodemus Siivola)
1624 * fixed bug: compile-time format string checker failed on
1625 non-closed ~{. (reported by Thomas F Burdick)
1626 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1627 don't warn on legal loop code involving more than one aggregate
1628 boolean. (thanks to Nikodemus Siivola)
1629 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1630 in keyword position to MAKE-PACKAGE was not being reported as
1632 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1633 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1635 * fixed bugs 280 and 312: the checking for multiple definitions in a
1636 file is less likely to become confused by uses of inline
1638 * fixed bug: the #S reader macro performs the keyword coercion
1639 specified for slot names. (reported by Kalle Niemitalo)
1640 * fixed bug: lambda lists may contain symbols whose names start with
1641 & but are not lambda-list-keywords; their occurrence triggers a
1643 * fixed bug 321: define-method-combination argument lambda lists do
1644 not cause invalid code to be generated when &optional or &aux
1645 variables have default values. (reported by Bruno Haible)
1646 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1647 have CLOS classes; this appears also to have fixed reported
1648 instability in stack exhaustion detection.
1649 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1650 MUFFLE-WARNING when no associated restart is present is now
1652 * optimization: rearranged the expansion of various defining macros
1653 so that each expands into only one top-level form in a
1654 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1656 * optimization: used a previously little-used slot in symbols to
1657 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1659 * fixed some bugs revealed by Paul Dietz' test suite:
1660 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1662 ** Fixed an optimization bug related to inheritance of initforms
1663 from local to shared slots.
1664 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1665 ** CAST optimizer forgot to flush argument derived type.
1666 ** print/read consistency on floats is now orders of magnitude
1667 more likely. (thanks also to Bruno Haible for a similar report
1669 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1670 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1671 for objects of type REAL. Make it so.
1672 ** SXHASH is specified (infelicitously) to respect similarity,
1673 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1674 it so. (thanks to Markus Ziegler)
1675 ** on the Alpha, the compiler succeeds in compiling functions
1676 returning a known number of arguments greater than 63.
1677 ** fixed handling of invalid NIL arguments in keyword position
1679 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1681 ** provided a readably-printable representation for RANDOM-STATE
1683 ** ensured that pathnames, where they have a namestring, always
1684 print using #P"..." syntax.
1686 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1687 * Support for the forthcoming 2.0 version of the NetBSD kernel
1688 running on x86 hardware has been added. (thanks to Perry
1689 E. Metzger most immediately, and others for their past work)
1690 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1691 current release version; SBCL's OpenBSD support had been broken
1692 since about the time of OpenBSD's switch to ELF binary format.)
1693 * [placeholder for DX summary]
1694 ** user code with &REST lists declared dynamic-extent, under high
1695 speed or space and low safety and debug optimization policy.
1696 * The manual has been converted to Texinfo format and the debugger
1697 chapter from the cmucl manual has been added.
1698 * A facility has been added to extract documentation strings from
1699 sbcl and store them as Texinfo-formatted snippets for inclusion in
1700 the manual (via Texinfo's @include directive)
1701 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1702 the test case to Dave Roberts)
1703 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1704 pointers. (reported by Sean Ross)
1705 * bug fix: PROFILE output is printed nicely even for large numerical
1706 values. (thanks to Zach Beane)
1707 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1708 greater than 32 handle EOF correctly.
1709 * bug fix: on X86 an immediate argument of the IMUL instruction is
1710 correctly printed in disassembly. (thanks to Lutz Euler)
1711 * bug fix: class slots in redefined classes preserve their old
1712 values. (thanks to Bruno Haible and Nikodemus Siivola)
1713 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1714 succeeds. (reported by Marco Baringer)
1715 * fixed some bugs revealed by Paul Dietz' test suite:
1716 ** READ-SEQUENCE now works on ECHO-STREAMs.
1717 ** RATIONALIZE works more according to its specification. (thanks
1720 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1721 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1722 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1723 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1724 should matter to you only if you rebind the printer control
1725 variables and then find you want different bindings in the
1726 debugger than in the ordinary execution of your program.)
1727 * The runtime build system has been tweaked to support building
1728 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1729 assembler and linker. (thanks to Nikodemus Siivola)
1730 * Unbound, undefined, undeclared variables now trigger full
1731 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1732 more often programmer error than deliberate exploitation of undefined
1734 * optimization: the hash algorithm for strings has changed to one
1735 that is less vulnerable to spurious collisions. (thanks to Juho
1737 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1738 needless bounds checking. (thanks to Juho Snellman)
1739 * optimization: implemented multiplication as a modular
1740 (UNSIGNED-BYTE 32) operation on the PPC backend.
1741 * fixed some bugs revealed by Paul Dietz' test suite:
1742 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1744 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1745 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1746 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1748 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1750 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1752 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1753 their output stream on EOF from read.
1754 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1755 have been read to end-of-file.
1756 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1758 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1759 description of determination of which consecutive characters
1761 ** Printing the "Space" character with escaping on now yields "#\\ ",
1762 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1763 ** Reading floating-point numbers with *READ-BASE* set to a number
1764 less than 10 works correctly.
1765 ** Reading floating-point numbers with *READ-BASE* set to a number
1766 more than 10 works correctly.
1767 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1768 the readtable currently in effect.
1770 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1771 * minor incompatible change: parsing of namestrings on a physical
1772 (Unix) host has changed; numbers after the final #\. in a
1773 namestring are no longer interpreted as a version field. This is
1774 intented to be largely invisible to the user, except that the
1775 meaning of the namestring "*.*.*" has changed: it now refers to a
1776 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1777 should usually be replaced by
1778 (make-pathname :name :wild :type :wild :version :wild)
1779 with the added benefit that this is more likely to be portable.
1780 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1781 option to OPEN now signals an error if the file being opened
1782 exists; this may have an impact on existing code.
1783 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1784 signals received, on the PowerPC platforms (both Linux and
1785 Darwin). (thanks to Pierre Mai for pointing out the location of
1787 * several fixes on OS X: The system now builds and runs cleanly on
1788 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1789 (thanks to Brian Mastenbrook)
1790 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1791 second-resolution: integer multiples of 1/3600 between -24 and 24.
1792 (thanks to Vincent Arkesteijn)
1793 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1794 type when called with 1 argument; PEEK-CHAR checked type of
1795 PEEK-TYPE only after having read first character from a
1796 stream. (reported by Peter Graves)
1797 * bug fix: the garbage collector now has much better locality
1798 behaviour, and in particular no longer treats all memory as being
1799 exhausted when many small objects point to each other in a deeply
1801 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1802 recognized as being TYPEP their class.
1803 * bug fix: the PUSHNEW documentation string has been corrected.
1804 (thanks to Vincent Arkesteijn)
1805 * bug fix: defaulting of the value for the last of an atypically
1806 large number of multiple values being bound was not being
1807 performed correctly on the Alpha or PPC platforms
1808 * optimization: implemented multiplication as a modular
1809 (UNSIGNED-BYTE 32) operation on the x86 backend.
1810 * optimization: SEARCH on simple-base-strings can now be open-coded.
1811 (see also contrib/compiler-extras.lisp for inspiration for
1812 teaching the compiler about the Boyer-Moore algorithm).
1813 * value, returned by MAX (and MIN) called with several EQUALP, but
1814 not EQL, arguments now does not depend on compiler settings.
1815 * fixed some bugs revealed by Paul Dietz' test suite:
1816 ** in stack analysis liveness information is propagated from
1817 non-local entry points.
1818 ** pathwise CAST removing failed when the CAST node did not start
1820 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1821 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1823 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1824 host is already defined.
1825 ** RENAME-FILE works on streams instead of signalling an internal
1827 ** PEEK-CHAR uses the current readtable when determining whether
1828 or not a character is whitespace.
1829 ** MERGE-PATHNAMES handles the case when the pathname does not
1830 specify a name while the default-pathname specifies a version.
1831 ** Pathnames now stand a better chance of respecting print/read
1833 ** Attempting to use standardized file system operators with a
1834 pathname with invalid :DIRECTORY components signals a
1836 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1838 ** DIRECTORY on logical pathnames is more correct.
1839 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1840 signal a TYPE-ERROR if their argument is not a stream.
1841 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1842 designator argument does not designate a stream.
1843 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1844 examining the synonym.
1845 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1847 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1848 element-type larger than ([UN]SIGNED-BYTE 32).
1850 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1851 * When built with the :SB-FUTEX feature, threaded builds now take
1852 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1853 for faster/more reliable mutex and condition variable support.
1854 * Incompatible change (but one you probably shouldn't have been using
1855 anyway): the interface and code for arbitrating between multiple
1856 threads in the same user session has been redesigned.
1857 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1858 error when called without an explicit environment argument.
1859 (thanks to Nikodemus Siivola)
1860 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1861 confused about their position. (thanks to Adam Warner and Gerd
1863 * bug fix: Pretty printing backquoted forms with unquotations in the
1864 argument list position of various code constructs such as LAMBDA
1865 now works correctly. (reported by Paul Dietz)
1866 * bug fix: Pretty printing unquotations no longer loses all
1867 stream position information.
1868 * optimization: performance of string output streams is now less
1869 poor for multiple small sequence writes.
1870 * optimization: performance of CSUBTYPEP in the presence of complex
1871 expressions involving CONS and NOT many times has been improved.
1872 (reported by Paul Dietz)
1873 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1875 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1876 streams for the REPL, for the debugger, and so on. (thanks to
1878 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1881 is ignored if it is too small, rather than propagating through
1882 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1883 ** extremely complex negations of CONS types were not being
1884 sufficiently canonicalized, leading to inconsistencies in
1886 ** VALUES tranformer lost derived type.
1888 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1889 * fixed a bootstrapping bug: the build process no longer assumes
1890 that the various BOOLE-related constants have the same value in
1891 host and target lisps. (noted by Paul Dietz' test suite on an
1892 SBCL binary built from CLISP)
1893 * The system can now be dynamically linked on the MIPS platform,
1894 which enables dynamic loading of foreign code from Lisp. (thanks
1895 to Ralf Baechle for discussions on the MIPS ABI)
1896 * The system now records debugging information for its own source
1897 files in a filesystem-position-independent manner, relative to
1898 the "SYS" logical host.
1899 * fixed a compiler bug: MV-LET convertion did not check references
1900 to the "max args" entry point. (reported by Brian Downing)
1901 * tweaked disassembly notes to be less confident about proclaiming
1902 some instruction as an LRA. (thanks to Brian Downing)
1903 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1904 now each have their own history, command character, and other
1905 characteristics. (thanks to David Lichteblau)
1906 * fixed some bugs revealed by Paul Dietz' test suite:
1907 ** compiler failure in compiling LOGAND expressions including a
1909 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1910 shift greater than 32.
1911 ** FLUSH-DEST did not mark blocks for type check regeneration.
1912 ** HANDLER-CASE failed to accept declarations in handler clauses
1913 in some circumstances.
1915 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1916 * New code in contrib/sb-introspect (still probably not entirely
1917 stable yet) provides some support for smart Lisp development
1918 environments like SLIME.
1919 * The conditions signalled for errors occurring when loading .fasl
1920 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1921 in a way which should help ASDF recover gracefully.
1922 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1923 stuff has been cleaned up. If you code contrib/ stuff, this might
1924 affect you, and you can look at contrib/README, contrib/STANDARDS,
1925 and/or the 0.8.4.27 diff to check.
1926 * In full calls the compiler now does not generate checks for declared
1927 argument types for all arguments.
1928 * various threading fixes
1929 ** and some experimental patches which didn't make it into
1930 the main tree for this release, but which are shipped in
1931 contrib/experimental-thread.patch as a possible fix for some
1932 failures (deadlock, spinning...) in GC-intensive multithreaded
1934 * fixed PPC build problem (source code incompatibility of different
1935 library versions): added offsetof() hackery which attempts to divine
1936 where glibc maintainers put uc_mcontext today
1937 * fixed bug 282: compiler does not trust type assertions while passing
1938 arguments to a full call.
1939 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1940 &OPTIONAL VALUES type parameter.
1941 * fix bug 214: algorithm for noting rejected templates is now more
1942 similar to that of template seletion. (also reported by rydis on
1944 * fixed bug 141b: printing backquoted information readably and prettily
1945 inserts a space where necessary.
1946 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1947 (CHARACTER 10) are now reported as errors, rather than propagated
1948 as unknown types. (reported by piso on #lisp)
1949 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1950 with values NIL and :ERROR. (thanks to Milan Zamazal)
1951 * fixed bug 191c: CLOS now does proper keyword argument checking as
1952 described in CLHS 7.6.5 and 7.6.5.1.
1953 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1954 longer raise an error; further, using a list as a for-as-arithmetic
1955 counter now raises a meaningful error.
1956 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1957 now understood by sequence creation functions such as MAKE-SEQUENCE
1959 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1960 a STRING-INPUT-STREAM.
1961 * compiler enhancement: SIGNUM is now better able to derive the type
1963 * type declarations inside WITH-SLOTS are checked. (reported by
1965 * fixed some bugs revealed by Paul Dietz' test suite:
1966 ** incorrect optimization of TRUNCATE for a positive first
1967 argument and negative second.
1968 ** compiler failure in let-convertion during flushing dead code.
1969 ** compiler failure while deriving type of TRUNCATE on an
1970 interval, containing 0.
1971 ** ASH of a negative bignum by a negative bignum count now returns
1973 ** intersection of CONS types now canonicalizes properly, fixing
1974 inconsistencies in SUBTYPEP.
1976 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1977 * incompatible change: The --disable-debugger command line
1978 option now clobbers the debugger at a more fundamental
1979 level, by redefining #'INVOKE-DEBUGGER instead of by
1980 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1981 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1982 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1983 enter the debugger REPL and then suffer recursive errors
1984 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1985 while under the new system, BREAK in --disable-debugger mode
1986 terminates the system just as an unhandled error would.
1987 * fixed compiler performance when processing loops with a step >1;
1988 * bug fix: DOCUMENTATION now retrieves generic function
1989 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1990 support has been systematized, and now supports the methods
1991 specified by ANSI, along with a default method and a method for
1992 slot documentation. (reported by Nathan Froyd)
1993 * bug fix: effective methods associated with a generic function are
1994 no longer cached over a change of that generic function's method
1995 combination. (reported by Andreas Fuchs)
1996 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1997 names a non-existent directory. (thanks to Andreas Fuchs)
1998 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1999 the PPC platform, lead to stack corruption; this has been fixed.
2000 (reported by Rainer Joswig)
2001 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2002 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2003 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2004 arithmetic in general, is now much more reliable on the PPC
2006 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2007 platform now returns the right answer.
2008 * optimization: restored some effective method precomputation in
2009 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2010 precomputation is now tunable.
2011 * optimization: compiler-internal data structure use has been
2012 reviewed, and changes have been made that should improve the
2013 performance of the compiler by about 20%.
2014 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2015 simple-base-strings and simple-bit-vectors is improved.
2016 * optimization: the optimization of 32-bit logical and arithmetic
2017 functions introduced in version 0.8.3 on the x86 has been
2018 implemented on the mips, ppc and sparc platforms; an
2019 implementation of the same facility, but for 64-bit arithmetic,
2020 has been added for the alpha.
2021 * microoptimization: the compiler is better able to make use of the
2022 x86 LEA instruction for multiplication by constants.
2023 * bug fix: in some situations compiler did not report usage of
2024 generic arithmetic in (SPEED 3) policy.
2025 * bug 145b fix: compiler used wrong type specifier while converting
2026 MEMBER-types to numeric.
2027 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2029 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2030 index argument now works properly on the Alpha platform.
2031 * bug fix: floating point exception treatment on the Alpha platform
2033 * bug fix: FILE-POSITION works much better on string input and
2034 output streams. (thanks to Nikodemus Siivola)
2035 * bug fix: many threading/garbage collection symptoms sorted.
2036 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2037 might be pseudo-atomic.
2038 * internal change: Stopping for GC is now done with signals not
2039 ptrace. GC is now done in whichever thread wanted it, instead of
2041 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2043 * bug fix: result form in DO is not contained in the implicit
2045 * incompatible change: ICR structure is changed; the value part of
2046 CONTINUATION is now called LVAR; corresponding functions are
2047 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2048 * added type deriver for ISQRT (thanks to Robert E. Brown).
2049 * bug fix: better support for loading from the command line when an
2050 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2052 * fixed some bugs revealed by Paul Dietz' test suite:
2053 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2054 ** ROUND and FROUND now give the right answer when given very
2055 small float arguments.
2056 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2058 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2059 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2060 ** LCM should return a non-negative integer.
2061 ** PARSE-INTEGER returned the index of a terminator instead of the
2062 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2063 ** PARSE-INTEGER returned an incorrect index being applied to a
2065 ** LCM with two arguments of 0 returns 0 rather than signalling
2067 ** unsigned addition of a 32-bit constant with the high bit set no
2068 longer causes an internal compiler error on the x86.
2069 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2070 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2071 with negative last argument.
2072 ** byte specifiers with zero size and position no longer cause
2073 an error during type derivation.
2074 ** bignum multiplication on the Alpha platform now returns the
2076 * porting: The system now builds on SuSE AMD64, although it still
2077 generates a 32-bit binary.
2078 * .fasl file incompatibility: The fasl file version number has
2079 been incremented (because of the changes to internal compiler
2080 data structures referred to above).
2082 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2083 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2084 more accurately, on the Darwin kernel running on PowerPC hardware.
2085 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2086 * Compiler code deletion notes now signal a condition of type
2087 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2088 an associated MUFFLE-WARNING restart.
2089 * The compiler now performs limited argument count validation of
2090 constant format strings in FORMAT, and where appropriate in ERROR,
2091 CERROR and WARN. (thanks to Gerd Moellmann)
2092 * New ASDF-INSTALL contrib can be used for automatic download and
2093 installation of third-party Lisp code from CCLAN or other sites
2095 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2096 which forces another thread to execute a function supplied by the
2098 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2099 now accept and act upon their :ELEMENT-TYPE keyword argument.
2100 (reported by Martin Atzmueller, Edi Weitz)
2101 * bug fix: FILE-POSITION now accepts position designators up to
2102 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2103 is the greater. (thanks to Patrik Nordebo)
2104 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2105 Kalle Olavi Niemitalo)
2106 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2107 printable. (reported by Eric Marsden)
2108 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2109 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2110 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2111 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2112 (thanks to Stig Sandoe)
2113 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2114 UNREAD-CHAR now correctly includes the unread character in the
2115 target sequence. (thanks to Gerd Moellmann)
2116 * bug fix in threaded builds: the system can now be suspended and
2117 resumed by shell job control with minimal disruption.
2118 * bug fixes in times and timezones >2038AD
2119 * better handling of "where is GNU make?" problem in build scripts
2120 (thanks to Nikodemus Siivola)
2121 * new optimization: inside a named function any reference to a
2122 function with the same name is considered to be a self-reference;
2123 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2124 optimization quality.
2125 * new optimization on x86: logical functions and + now have
2126 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2127 used when the result is truncated to 32 bits.
2128 * VALUES declaration is partially enabled.
2129 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2130 * fixed some bugs revealed by Paul Dietz' test suite:
2131 ** The system now obeys the constraint imposed by
2132 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2133 types form a lattice under type intersection.
2134 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2135 ** ASSOC now ignores NIL elements in an alist.
2136 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2137 and (1+ MOST-POSITIVE-FIXNUM) answers.
2138 ** The addition of a method with invalid qualifiers to a generic
2139 function does not cause an error to be signalled immediately;
2140 a warning is signalled, and the error is generated only on
2141 calling the generic function.
2142 * changed .fasl file version number, in part to add type codes for
2143 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2144 obscure ANSI requirements
2146 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2147 * fixed bug 148: failure to inline-expand a local function left
2148 garbage, confusing the compiler.
2149 * fixed bugs 3cd: structure slot readers perform type check if the
2150 slot can have an invalid value (i.e. it is either not initialized
2151 or can be written with a less specific slot writer).
2152 * bug fix: the compiler now traps array references to elements off
2153 the end of an array; previously, the bounds checking in some
2154 circumstances could go off-by-one.
2155 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2157 * type declarations for array element types now obey the description
2158 on the CLHS page "Declaration TYPE", as per discussions on
2159 sbcl-help around 2003-05-08. This means that a declaration
2160 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2161 declaration, all references to BAR will be asserted or assumed
2162 (with THE, so dependent on compiler policy) to involve objects of
2163 type FOO. Note that no such declaration is implied in
2164 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2165 * declared types of functions from the "Conditions"
2166 chapter. (reported by Paul Dietz)
2167 * bug fix: CERROR accepts a function as its first argument.
2168 * bug fix: NTH an NTHCDR accept a bignum as index
2169 arguments. (reported by Adam Warner)
2170 * optimization: character compare routines now optimize comparing
2171 against a constant character. (reported by Gilbert Baumann)
2172 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2173 argument now works properly on the MIPS platform.
2174 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2175 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2176 declared to have a functional type, does not cause an error
2178 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2179 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2180 * fixed bug: if last continuation of a deleted block has a
2181 destination, this destination should be deleted too. (reported by
2183 * fixed a bug in the bootstrap process: the host compiler's values
2184 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2185 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2186 test case from Patrik Nordebo)
2187 * improved the ability of the disassembler on the PPC platform to
2188 provide helpful disassembly notes.
2189 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2190 the class in more cases than previously.
2191 * bug fix: FILE-POSITION now understands :START and :END for
2192 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2193 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2194 8) data. (thanks to David Lichteblau)
2195 * bug fix: it is possible to add a method to a generic function
2196 without lambda list.
2197 * bug fix: reader failed to signal END-OF-FILE inside an
2198 object representation. (reported by Nikodemus Siivola)
2199 * fixed some bugs revealed by Paul Dietz' test suite:
2200 ** LAST and [N]BUTLAST should accept a bignum.
2201 ** condition slot accessors are methods.
2202 ** (VECTOR NIL) is a subtype of STRING.
2204 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2205 * minor incompatible change: some nonsensical specialized lambda
2206 lists (used in DEFMETHOD) which were previously ignored now signal
2208 * minor incompatible change: the system is now aware of the types of
2209 variables in the COMMON-LISP package, and will signal errors for
2210 most violations of these type constraints (where previously they
2211 were silently accepted).
2212 * minor incompatible change: COMPILE-FILE now uses the freedom
2213 afforded (ANSI 3.2.2.3) to use derived function types for
2214 functions defined in the same file. This also permits the system
2215 to warn on static type mismatches and function
2216 redefinition. (Currently it does not work with high DEBUG level.)
2217 * minor incompatible change: VALUES declaration is disabled.
2218 * When issuing notes, the compiler now signals a condition of type
2219 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2220 restart for use in user handlers. It is expected that the
2221 COMPILER-NOTE condition will eventually become a condition
2222 supertype to a hierarchy of note types, which will then be
2223 handleable in a similar fashion. However, at the moment, no such
2224 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2225 but it's an implementation detail, not a classification for the
2227 * Changes in type checking closed the following bugs:
2228 ** type checking of unused values (192b, 194d, 203);
2229 ** template selection based on unsafe type assertions (192c, 236);
2230 ** type checking in branches (194bc).
2231 * A short form of VALUES type specifier has ANSI meaning (it has
2232 increased the number of situations when SBCL cannot perform type
2234 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2235 keywords or constants is permissible.
2236 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2237 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2238 classes. (thanks to Antonio Martinez)
2239 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2240 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2241 argument to a pretty-print stream. Instead, it leaves any such
2242 operations to DESCRIBE-OBJECT methods.
2243 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2245 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2246 lambda lists. (thanks to David Lichteblau)
2247 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2248 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2249 (thanks to Gerd Moellmann)
2250 * fixed bug 246: increased compilation speed of long
2251 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2253 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2255 * DEFCONSTANT now throws a condition of type
2256 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2257 constant to a non-EQL value; CONTINUE and ABORT restarts
2258 respectively change and preserve the value.
2259 * fixed bug 63: The code walker, part of the implementation of CLOS,
2260 is now better at handling symbol macros.
2261 * bug fix: There is no longer an internal implementation type named
2262 CL:LENGTH. (reported by Raymond Toy)
2263 * bug fix: In macro-like defining macros/special operators the
2264 implicit block does not enclose the lambda list.
2265 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2266 enclosed in parentheses) are not suitable as type specifiers, and
2267 their use properly signals an error now.
2268 * bug fix: An argument count mismatch for a type specifier in code
2269 being compiled no longer causes an unhandled error at compile
2270 time, but signals a compile-time warning.
2271 * fixed simple vector readable printing
2272 * bug fix: DESCRIBE takes more care over whether the class
2273 precedence list slot of a class is bound before accessing it.
2274 (reported by Markus Krummenacker)
2275 * bug fix: FORMATTER can successfully compile pretty-printer format
2276 strings which use variants of the ~* directive inside.
2277 * bug fix: SEARCH now applies its TEST predicate to the elements of
2278 the arguments in the correct order. (thanks to Wolfhard Buss)
2279 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2280 in type inference. (thanks to Robert E. Brown)
2281 * bug fix: Reading in symbols with an explicit package name of ""
2282 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2283 * fixed some bugs revealed by Paul Dietz' test suite:
2284 ** NIL is now allowed as a structure slot name.
2285 ** Arbitrary numbers, not just REALs, are allowed in certain
2286 circumstances in LOOP for-as-arithmetic clauses.
2287 ** Multiple class redefinitions before slot access no longer
2288 causes a type error.
2289 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2290 association between the name and a class.
2291 ** Generic functions with non-standard method-combination and over
2292 five methods all of which return constants no longer return NIL
2293 after the first few invocations. (thanks to Gerd Moellmann)
2294 ** CALL-NEXT-METHOD with no arguments now passes the original
2295 values of the arguments, even in the presence of assignment.
2296 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2298 ** DISASSEMBLE works with closures and funcallable instances.
2299 ** ADD-METHOD now returns the generic function, not the new method.
2300 ** FIND-METHOD signals an error if the lengths of the specializers
2301 is incompatible with the generic function, even if the ERRORP
2303 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2304 which its argument is a member.
2305 ** DEFCLASS only redefines the class named by its class-name
2306 argument if that name is the proper name of the class;
2307 otherwise, it creates a new class.
2308 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2309 of the UNBOUND-SLOT condition to the name of the slot.
2310 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2311 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2312 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2313 SLOT-MAKUNBOUND in the specified fashion.
2315 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2316 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2317 cross-compilation host. As a consequence, we can now bootstrap our
2318 way up to SBCL starting with a bare gcc toolchain and human-readable
2319 source code (first the source to CLISP, then the source to SBCL).
2320 * A contributed module containing a partial implementation of the
2321 simple-streams interface has been included. (thanks to Rudi
2323 * A contributed module implementing the RFC1321 Message Digest
2324 Algorithm, known as MD5, has been included.
2325 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2326 no longer has any effect, as the code controlled by this feature
2327 has been deleted. (As far as we know, no-one has ever built using
2328 this feature, and its semantics were confused in any case).
2329 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2330 work on conditions (as required by the ANSI specification),
2331 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2332 expected behaviour on conditions. Users should note, however,
2333 that such behaviour is not required by the ANSI specification,
2334 and so use of this behaviour may render their code unportable.
2335 * fixed some bugs revealed by Paul Dietz' test suite:
2336 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2338 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2339 ** SXHASH on condition objects no longer returns NIL.
2340 ** :ALLOCATION :CLASS slots are better treated; their values are
2341 updated on class redefinition, and initforms inherited from
2342 superclasses are applied.
2343 ** REMOVE-METHOD returns its generic function argument even when
2344 no method was removed.
2345 ** SHARED-INITIALIZE now initializes the values of the requested
2346 slots, including those with :ALLOCATION :CLASS.
2347 ** ALLOCATE-INSTANCE now works on structure classes defined via
2348 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2350 ** SLOT-EXISTS-P now works on conditions, as well as structures
2352 ** MAKE-LOAD-FORM now has the required methods on
2353 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2354 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2355 keyword, and now implements the SLOT-NAMES argument.
2356 ** methods with &OPTIONAL arguments no longer allow too many
2357 arguments to be passed in the call without error.
2358 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2359 option is consistent with the required arguments of the generic
2360 function lambda list.
2361 * bug fix: REQUIRE accepts a string designator. (Thanks to
2363 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2364 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2365 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2366 * bug fix: The compiler now checks for duplicated variables in macro
2368 * bug fix: SETQ on globals returns the correct value.
2369 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2370 not a COMPILER-ERROR (followed by some other strange error on
2371 choosing the CONTINUE restart).
2372 * bug fix: make.sh and friends are now more consistent in the way that
2373 they look for GNU "make".
2375 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2376 * experimental native threads support (on x86 Linux >=2.4 only).
2377 This is not compiled in by default: you need to add :SB-THREAD to
2378 the target features. See the "Beyond ANSI" chapter of the manual
2380 * fix for longstanding nonANSIism: The old distinction between
2381 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2382 The return value from CL:FIND-CLASS is now a CLOS class, and
2383 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2384 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2385 * An interface to the MetaObject Protocol, as described in Kiczales,
2386 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2387 MIT Press, 1991, is available from the SB-MOP package.
2388 * incompatible change: the SB-PCL package should now be considered
2389 a private implementation detail, and no longer a semi-private MOP
2391 * minor incompatible change: due to rearrangement for threads, the
2392 control stack and binding stack are now allocated at arbitrary
2393 addresses instead of being hardcoded per-port. Users affected by
2394 this probably have to be doing advanced things with shared
2395 libraries, and will know who they are.
2396 * minor incompatible change: Previously, all --eval forms used were
2397 processed with READ before any of them were processed with EVAL.
2398 Now each --eval form is processed with both READ and EVAL before
2399 the next --eval form is processed. (Thus package operations like
2400 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2401 work as the user might reasonably expect.)
2402 * minor incompatible change: *STANDARD-INPUT* is now only an
2403 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2405 * minor incompatible change: Y-OR-N-P is now character-oriented,
2406 not line oriented. Also, YES-OR-NO-P now works without errors.
2407 (thanks to Antonio Martinez)
2408 * sb-aclrepl module improvements: an integrated inspector, added
2409 repl features, and a bug fix to :trace command.
2410 * Known functions, which cannot be open coded by the backend, are
2411 considered to be able to check types of their arguments. (fixing
2412 a bug report by Nathan J. Froyd)
2413 * fixed a bug in computing method discriminating functions: It is
2414 now possible to define methods specialized on classes which have
2415 forward-referenced superclasses. (thanks to Gerd Moellmann)
2416 * fixed evaluation order in optional entries (reported by Gilbert
2418 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2419 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2420 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2421 required-by-AMOP NAME argument, as well as CLASS and
2422 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2423 * fixed bug 20: DEFMETHOD can define methods using names that are
2424 not the proper names of classes to designate class specializers.
2425 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2426 Unix file descriptors, instead of blowing up. (thanks to Antonio
2428 * Garbage collection refactoring: user-visible change is that a
2429 call to the GC function during WITHOUT-GCING will not do garbage
2430 collection until the end of the WITHOUT-GCING. If you were doing
2431 this you were probably losing anyway.
2432 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2433 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2434 TYPEP the latter but not the former.
2435 * The compiler issues a full WARNING for calls to undefined functions
2436 with names from the CL package.
2437 * MAP-INTO for a vector destination is open coded. (reported by
2438 Brian Downing on c.l.l)
2439 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2440 documentation string.
2441 * fixed some bugs revealed by Paul Dietz' test suite:
2442 ** COPY-ALIST now signals an error if its argument is a dotted
2444 ** Condition slots are now accessed more correctly in the presence
2445 of multiple initargs for a given slot.
2446 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2447 exclude restarts of the same name associated with a different
2449 ** DEFCLASS of forward-referenced classes with another
2450 forward-referenced class in the superclasses list no longer
2452 ** Condition slots are now initialized once each, not multiple
2453 times. (thanks to Gerd Moellmann)
2454 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2456 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2457 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2458 arguments contain duplicated elements.
2459 ** RESTART-CASE understands local macros.
2460 ** RESTART-CASE associates exactly its own restarts with a condition.
2461 ** ENDP in safe mode checks its argument to be of type LIST.
2462 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2463 in question is unbound.
2464 ** Optimized MAKE-INSTANCE functions no longer cause internal
2465 assertion failures in the presence of duplicate initargs.
2466 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2467 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2468 propagated even in the ctor optimized implementation of
2470 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2472 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2473 * a better implementation of SXHASH on (simple) bit vectors,
2474 measured both in execution speed and in distribution of results
2475 over the positive fixnums, has been installed. Likewise, a better
2476 implementation of EQUAL for simple bit vectors is now available.
2477 * fixed CEILING optimization for a divisor of form 2^k.
2478 * fixed bug 240 (emitting extra style warnings "using the lexical
2479 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2480 by Antonio Martinez)
2481 * fixed SXHASH, giving different results for NIL depending on type
2482 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2483 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2484 variable. (found by Rolf Wester)
2485 * SBCL does not ignore type declarations for special
2486 variables. (reported by rif on c.l.l 2003-03-05)
2487 * some bug fixes in contrib/sb-aclrepl/
2488 * fixed some bugs revealed by Paul Dietz' test suite:
2489 ** a bug in the CONS type specifier, whereby the CAR and CDR
2490 types got intertwined, has been fixed;
2491 ** the type system is now able to reason about the interaction
2492 between INTEGER and RATIO types more completely;
2493 ** APPEND, [N]REVERSE and NRECONC check that those their
2494 arguments, which must be proper lists, are really so;
2495 ** An array specialized to be unable to hold elements has been
2496 implemented, as required -- yes, really -- by ANSI;
2497 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2498 on malformed property lists;
2500 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2501 * incompatible packaging change: in line with Unix convention,
2502 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2503 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2504 * REQUIRE and PROVIDE are now optionally capable of doing something
2505 useful. See the documentation string for REQUIRE.
2506 * infrastructure for a managed SBCL contrib system: contributed
2507 modules in this release include:
2508 ** the ASDF system definition facility;
2509 ** an interface to the BSD Sockets API;
2510 ** an ACL-like convenience interface to the repl;
2511 (thanks to Kevin Rosenberg)
2512 ** an implementation of ROTATE-BYTE, with efficient implementation
2514 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2515 gives a closer approximation to the right answer than previously.
2516 (thanks to Raymond Toy)
2517 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2518 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2519 optional environment arguments, as required by ANSI.
2520 * fixed bugs in other functions taking environment objects, allowing
2521 calls with an explicit NIL environment argument to be compiled
2523 * fixed bug 228: primary return values from
2524 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2525 COMPILE or FUNCTION.
2526 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2527 :TYPE will no longer signal errors on innocuous objects.
2528 * fixed bug 231b: SETQ is better at respecting type declarations in
2529 the lexical environment.
2530 * fixed a bug in DEFCLASS: classes named by symbols with no or
2531 unprintable packages can now be defined.
2532 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2533 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2534 * fixed compiler failure related to checking types of functions.
2535 (reported by Robert E. Brown)
2536 * the compiler is now much more consistent in its error-checking
2537 treatment of bounding index arguments to sequence functions: in
2538 (SAFETY 3) code, errors will be signalled in almost all cases if
2539 invalid sequence bounding indices are passed to functions defined
2540 by ANSI to operate on sequences.
2541 * fixed a bug in the build procedure: documentation of SBCL-specific
2542 packages is now preserved and available in the final Lisp image.
2543 * lifted FDEFINITION lookup out of loops in the implementation of
2544 many list operations. (thanks to Robert E. Brown)
2545 * fixed a bug in the reader: the #n# reader macro now works for
2546 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2547 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2548 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2549 E. Brown and Tony Martinez respectively)
2550 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2551 SLOT-BOUNDP: in particular, it is now able to optimize them much
2552 better, and is now not vulnerable to having packages renamed.
2553 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2554 always called when a slot is not present in an instance. (thanks
2556 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2557 longer signal an error on unreading a character following EOF on
2558 the previous constituent stream. (thanks to Tony Martinez)
2559 * fixed some bugs revealed by Paul Dietz' test suite:
2560 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2561 not just nonnegative fixnums;
2562 ** the logical bit-array operators such as BIT-AND now accept an
2563 explicit NIL for their "opt-arg" argument (to indicate a
2564 freshly-consed result bit-array);
2565 ** ELT now signals an error on an invalid sequence index in safe
2567 ** the type system is now cleverer about negations of numeric
2568 types, and consequently understands the BIGNUM and RATIO types
2570 ** the type system is now cleverer about the interaction between
2571 INTEGER and RATIO types: while bugs still remain, many more
2572 cases are accurately computed;
2573 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2574 if it is in the last clause;
2575 ** CONSTANTLY now correctly returns a side-effect-free function in
2577 ** DECLARE is no longer treated as a special-operator; in
2578 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2579 * incremented fasl file version number due to the change in the
2580 DEFSTRUCT-SLOT-DESCRIPTION structure.
2582 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2583 * minor incompatible change: code processed by the "interpreter" or
2584 EVAL now has a compilation optimization policy of (DEBUG 2)
2585 (changed from (DEBUG 1)) to improve debuggability of interactive
2586 development, and to allow the use of the debug RETURN command in
2588 * an experimental implementation of the RETURN command for the
2589 debugger has been included. (thanks to Frederik Kuivinen)
2590 * fixed bug 62: constraints were not propagated into a loop.
2591 * fixed bug in embedded calls of SORT (reported and investigated by
2593 * fixed some bugs revealed by Paul Dietz' test suite:
2594 ** printing and reading of arrays with some dimensions having
2595 length 0 (thanks to Gerd Moellmann);
2596 ** BOA constructor with &AUX argument without a default value does
2597 not cause a type error;
2598 ** CONSTANTP now returns true for all self-evaluating objects.
2600 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2601 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2602 accessors that are related by inheritance, as specified in the
2603 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2604 Valtteri Vuorikoski)
2605 * The compiler is now able to inline functions that were defined in
2606 a complex lexical environment (e.g. inside a MACROLET).
2607 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2609 * Support for the upcoming FreeBSD-5.0 release has been included.
2610 (thanks to Dag-Erling Smorgrav)
2611 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2612 effect when it is not in a toplevel context.
2613 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2614 stand a better chance of being correct. (thanks to Gerd
2616 * fixed bug in COERCE, which now signals an error on coercing a
2617 rational to a bounded real type which excludes the expected
2619 * The compiler is now able to derive types more accurately from the
2620 COERCE and COMPILE functions.
2621 * fixed bug 223: functional binding is considered to be constant
2622 only for symbols in the CL package.
2623 * fixed bug 231: SETQ did not check the type of a variable being set
2624 (reported by Robert E. Brown)
2625 * A new optimization for MAKE-INSTANCE has been included, fixing
2626 various bugs (including relating to :ALLOCATION :CLASS slots and
2627 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2629 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2630 ** As required by ANSI, LOOP now disallows anonymous collection
2631 clauses such as COLLECT I in conjunction with aggregate boolean
2632 clauses such as THEREIS (= I 1);
2633 ** LOOP now signals an error when any variable is reused in the
2634 same loop (including the potentially useful construct analogous
2635 to WITH A = 1 WITH A = (1+ A);
2636 ** IT is only a special loop symbol within the first clause of a
2637 conditional loop clause;
2638 ** LOOP with a typed iteration variable over a hashtable now
2639 signals a type error iff it should.
2640 * fixed some other bugs revealed by Paul Dietz' test suite:
2641 ** FILE-STREAM now names the class previously known as FD-STREAM;
2642 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2643 argument) no longer signals an error;
2644 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2645 of the slot symbol, rather than using the current package
2646 ((:CONC-NAME "") continues to intern the slot's name in the
2648 * incremented fasl file version number, because of the incompatible
2649 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2650 of the new implementation of DEFINE-COMPILER-MACRO.
2652 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2653 * Support for building SBCL for MIPS platforms running in
2654 little-endian mode has now been checked in, and basic
2655 functionality on said platforms verified.
2656 * minor incompatible change: PCL now records the pathname of a file
2657 in which methods and the like are defined, rather than its
2659 * minor incompatible change: TRUENAME now considers the truename of
2660 a file naming a directory to be the pathname with :DIRECTORY
2661 component indicating that directory.
2662 * minor incompatible change: a NAMED clause in the extended form of
2663 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2664 reason for the previous behaviour is unclear.
2665 * more systematization and improvement of CLOS and MOP conformance
2666 in PCL (thanks to Gerd Moellman and Pierre Mai):
2667 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2669 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2670 primary methods with no specializers;
2671 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2673 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2674 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2675 FORWARD-REFERENCED-CLASSes; error reporting on
2676 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2678 ** SXHASH on CLOS instances now uses a slot internal to the
2679 instance to return different numbers on distinct instances,
2680 while preserving the same return value through invocations of
2682 ** DEFMETHOD signals errors when methods with longer incongruent
2683 lambda lists are added to generic functions;
2684 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2685 CLASS, as specified in AMOP;
2686 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2687 on the order returned by the primary method for classes of
2688 class STANDARD-CLASS;
2689 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2690 * fixed some bugs shown by Paul Dietz' test suite:
2691 ** DOLIST puts its body in TAGBODY;
2692 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2694 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2695 value producing form;
2696 ** if more variables are given to PROGV than values, extra
2697 variables are bound and made to have no value;
2698 ** NSUBSTITUTE on list arguments gets the right answer with
2700 ** ELT signals an error of type TYPE-ERROR when the index argument
2701 is not a valid sequence index;
2702 ** LOOP signals (at macroexpansion time) an error of type
2703 PROGRAM-ERROR when duplicate variable names are found;
2704 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2705 ** FUNCALL of special-operators now cause an error of type
2707 ** PSETQ now works as required in the presence of side-effecting
2708 symbol-macro places;
2709 ** NCONC accepts any object as its last argument;
2710 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2712 ** loop-for-as-package does not require a package to be explicitely
2714 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2716 * fixed bug 166: compiler preserves "there is a way to go"
2717 invariant when deleting code.
2718 * fixed bug 172: macro lambda lists with required arguments after
2719 &REST arguments now cause an error to be signalled. (thanks to
2721 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2722 bug. (thanks to Matthew Danish)
2723 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2725 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2726 arguments are lexically rebound. (thanks to Gerd Moellmann and
2728 * fixed bug 194: error messages are now more informative when there
2729 is no primary method applicable in a call to a generic
2730 function. (thanks to Gerd Moellmann)
2731 * fixed bug in command line argument checking (thanks to Julian
2733 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2734 (thanks to Matthew Danish)
2735 * incremented fasl file version number, because of the
2736 SXHASH-related changes in the layout of CLOS data structures
2738 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2739 * minor incompatible change: The runtime (the Unix executable named
2740 "sbcl") is now much pickier about the .core files it will load.
2741 Essentially it now requires .core files to descend from the same
2742 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2743 as the runtime does. (The intent is to prevent the crashes which
2744 can occur, and which can even be reported as mysterious failures,
2745 when people patch the sources or change the build parameters
2746 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2747 sbcl and .core files.)
2748 * fixed bug: VALUES-LIST is no longer optimized away.
2749 * fixed bug 142: The FFI conversion of C string values to Lisp
2750 string values no longer conses excessively. (thanks to Nathan
2751 Froyd porting Raymond Toy's fix to CMU CL)
2752 * began to systematize and improve MOP conformance in PCL (thanks to
2753 Nathan Froyd, Gerd Moellman and Pierre Mai):
2754 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2756 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2757 ** FINALIZE-INHERITANCE is now called on class finalization;
2758 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2759 argument precedence order.
2760 * fixed bug 202: The compiler no longer fails on functions whose
2761 derived types contradict their declared type.
2762 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2763 so it can be non-toplevel.
2764 * The fasl file version number has changed (because of the new
2765 implementation of DEFMACRO).
2766 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2767 safe code, that any length requirement by their type-specifier
2768 argument is valid. The exceptions to this are described in bug
2770 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2771 functions now check the types of their inputs as required by ANSI.
2772 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2773 introduced symbol is DECLAREd to be SPECIAL.
2774 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2775 * fixed bug: :COUNT argument to sequence functions may be negative
2776 * fixed bug: body of DO-SYMBOLS may contain declarations
2777 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2778 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2779 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2780 symbol macro only once
2781 * fixed printing of call frame when argument list is unavailable
2782 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2783 * compiler no longer signals WARNING on unknown keyword
2786 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2787 * A beta-quality port to the mips architecture running Linux,
2788 based on the old CMUCL backend, has been made. It has been tested
2789 on a big-endian kernel, and works sufficiently well to be able to
2790 rebuild itself; it has not been tested in little-endian mode.
2791 * fixed an inconsistency between gencgc.c and purify.c which made
2792 dumping/loading .core files unreliable
2793 * fixed bug 120a: The compiler now deals correctly with IFs where
2794 the consequent is the same as the alternative, instead of
2795 misderiving the return type. (thanks to Alexey Dejneka)
2796 * fixed bug 113: Logical pathnames are now dumpable (the logical
2797 host is resolved at load-time, throwing an error if it is not
2799 * fixed bug 174: FORMAT's error message is slightly clearer when a
2800 non-printing character is used in a format directive.
2801 * fixed several bugs in compiler checking of type declarations, i.e.
2802 violations of the Python "declarations are assertions" principle
2803 (thanks to Alexey Dejneka)
2804 * fixed several bugs in PCL's error checking (thanks to Gerd
2806 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2808 * fixed bug in compilation of functions as first class values
2809 (thanks to Antonio Martinez-Shotton)
2810 * The compiler's handling TYPE-ERRORs which it can prove will
2811 inevitably happen at runtime has been cleaned up and corrected
2812 in several ways. (thanks to Alexey Dejneka)
2813 * improved argument type checking for various basic arithmetic
2814 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2815 lavished on them in the past that they can be compiled in many
2816 ways in different special cases
2817 * fixed bug 181: compiler checks validity of user supplied type
2819 * cleaned up code flushing in optimization: Function calls which
2820 should signal errors for safety purposes (e.g. which ANSI says
2821 should signal errors when their arguments are of incorrect type)
2822 are no longer optimized away.
2823 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2824 * incremented fasl file version number, because changes in the
2825 implementation of sequence functions like COERCE caused
2826 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2827 in old inline expansions) to become undefined. (Actually these
2828 changes were later undone, so we might very well be binary
2829 compatible with 0.7.7 after all, but leaving the version number
2830 incremented seemed like the simplest and most conservative
2833 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2834 * An alpha-quality port to the parisc architecture running Linux,
2835 based on the old CMUCL backend, has been made. This, even more so
2836 than the other backends, should be considered still a work in
2837 progress; known problems include that the Linux kernel in 64-bit
2838 mode does not propagate the correct sigcontext structure to
2839 userspace, and consequently SBCL on a parisc64 kernel will not
2841 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2842 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2843 suppressed.) Also now that the compiler is looking at declarations
2844 in the environment, it checks optimization declarations as well,
2845 and suppresses inlining when (> DEBUG SPEED).
2846 * More fixes have been made to treatment of floating point exception
2847 treatment and other Unix signals. In particular, floating point
2848 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2849 * The detection and handling of control stack exhaustion (infinite
2850 or very deeply nested recursion) has changed. Stack exhaustion
2851 detection is now done by write-protecting pages at the OS level
2852 and applies at all optimization settings; when found, a
2853 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2854 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2855 be caught using IGNORE-ERRORS.
2856 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2857 correctly when transforming recursive LABELS functions to LETs.
2858 (thanks to Alexey Dejneka)
2859 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2860 that are names of constants or global variables.
2861 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2862 alien routines with docstrings.
2863 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2864 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2866 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2867 (thanks to Eric Marsden)
2868 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2869 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2870 * Bug fix: The SPARC backend can now compile functions involving
2871 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2872 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2873 to a functions expecting a PATHNAME-DESIGNATOR.
2874 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2875 lambda lists. (thanks to Alexey Dejneka)
2876 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2877 functions so that EXPORTing the name of the class doesn't cause
2878 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2879 to Antonio Martinez for reporting this)
2880 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2881 constant-folded without failing an assertion. (thanks to Einar
2882 Floystad Dorum for reporting this)
2883 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2884 types are now dealt with more correctly by the compiler.
2885 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2886 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2887 behaviour analogous to RENAME-FILE. This puts its behaviour more
2888 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2890 * The fasl file version number has changed again. (because of the
2891 bug fix involving the names of PCL MAKE-INSTANCE functions)
2893 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2894 * bug fix: Floating point exceptions are treated much more
2895 consistently on the x86/Linux and PPC/Linux platforms.
2896 * Array initialization with :INITIAL-ELEMENT is now much faster for
2897 cases when the compiler cannot open code the array creation, but
2898 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2899 array accesses have also seen a speed increase.
2900 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2901 specified. (This was at the root of some bad interactions between
2902 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2904 * bug fix: Internal error arguments for undefined functions are now
2905 computed correctly on the PPC/Linux platform.
2906 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2907 Raymond Toy's patch for CMU CL)
2908 * Support for the Solaris 9 operating environment has been included
2909 (thanks to Daniel Merritt)
2910 * A very ugly but hopefully complete draft of the missing FFI chapter
2911 of the manual has been created by reformatting the corresponding
2912 CMU CL manual chapter into (currently very ugly and incoherent)
2913 DocBook and bringing it up to date for SBCL behavior. Thus, the
2914 manual is now essentially complete, at least by my extreme
2915 once-and-only-once standards, whereby it's acceptable to refer to
2916 the doc strings of SB-EXT functions as the primary documentation.
2917 * The fasl file version number has changed again, due to cleanup of
2918 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2919 is no longer a static symbol.)
2921 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2922 * SBCL now builds with OpenMCL (version 0.12) as the
2923 cross-compilation host; also, more progress has been made toward
2924 bootstrapping under CLISP.
2925 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2927 * bug 158 fixed: The compiler can now deal with integer loop
2928 increments different from 1; fixing this turned out also to fix
2930 * bug 169 fixed: no more bogus warnings about using lexical bindings
2931 despite the presence of perfectly good SPECIAL declarations (thanks
2932 to David Lichteblau)
2933 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2934 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2935 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2937 * bug 180 fixed: Method combination specifications no longer ignore
2938 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2939 * bug fix: Structure type predicate functions now check their argument
2940 count as they should.
2941 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2942 correctly. (thanks to Pierre Mai)
2943 * minor incompatible change: The --noprogrammer option is deprecated
2944 in favor of the new --disable-debugger option, which is very similar.
2945 (The major difference is that it takes effect at a slightly different
2946 time at startup, causing handling of errors in --sysinit and
2947 --userinit files will be affected differently.) The
2948 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2949 been added to allow this functionality to be controlled from ordinary
2950 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2951 maintainers, who might want to run non-interactive scripts to
2952 build SBCL cores which will later be used interactively.)
2953 * minor incompatible change: The LOAD function no longer, when given
2954 a wild pathname to load, loads all files matching that pathname.
2955 Instead, an error of type FILE-ERROR is signalled.
2957 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2958 * bug 147 fixed: The compiler preserves its block link/count
2959 invariants more correctly now so that it doesn't crash. (thanks
2961 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2963 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2964 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2965 no-value-for-this-character as NIL (as specified by ANSI).
2966 * HOST-NAMESTRING on physical pathnames now returns a string that is
2967 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2968 (thanks to Christophe Rhodes)
2969 * The Alpha port handles icache flushing more correctly. (thanks to
2971 * More progress has been made toward bootstrapping under CLISP. (thanks
2972 to Christophe Rhodes)
2973 * The fasl file format has changed again, because dynamic loading
2974 on OpenBSD (which has non-ELF object files) motivated some cleanups
2975 in the way that foreign symbols are transformed and passed around.
2976 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2977 is no longer treated as whitespace by the reader, but instead as
2978 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2979 |A
\7fB|, instead of A as it used to.
2981 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2982 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2983 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2984 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2985 this as of 0.7.1.45, but was left out of the previous news section
2986 (thanks to Dan Barlow)
2987 * SBCL now runs on the Solaris operating system on SPARC architectures
2988 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2989 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2990 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2991 * SPARC backend cleanups, allowing builds of cores optimized for V8
2992 and V9 SPARCS, and also emission of code targeted to a particular
2993 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2995 * SBCL is closer to bootstrapping under CLISP, thanks to various
2996 fixes by Christophe Rhodes.
2997 * The fasl file format has changed again, to allow the compiler's
2998 INFO database to support symbol macros.
2999 * The user manual (in doc/) is formatted into HTML more nicely.
3000 (thanks to coreythomas)
3001 * The system is smarter about SUBTYPEP relationships, especially
3002 those involving NOT types (including types such as ATOM which are
3003 represented internally using NOT types). Thus SUBTYPEP is less
3004 likely to return (VALUES NIL NIL) in general, and in particular
3005 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3007 * The fasl file format has changed again, because the internal
3008 representation of types now includes a new slot to support the new
3009 SUBTYPEP-of-NOT-types logic.
3010 * (not a change in the main branch of SBCL, but a related prototype
3011 which can hopefully be merged into the main branch of SBCL in the
3012 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3013 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3014 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3015 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3016 dispatch macro character combinations. (thanks to Alexey Dejneka)
3017 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3018 that superficially look like logical namestrings correctly.
3019 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3021 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3022 default supplied-p) for &optional and &key arguments. (thanks to
3025 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3026 * incompatible change: The compiler is now less aggressive about
3027 tail call optimization, doing it only when (> SPACE DEBUG) or
3028 (> SPEED DEBUG). (This is an incompatible change because there are
3029 programs which relied on the old CMU-CL-style behavior to optimize
3030 away their unbounded recursion which will now die of stack overflow.)
3031 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3032 for non-GENCGC systems has been increased to 20M (since that
3033 seems much closer to the likely performance optimum for modern
3034 systems than the old 4M value was)
3035 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3036 and *DEBUG-PRINT-LEVEL*
3037 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3038 of CMU CL's support for SPARC, and various endianness and other
3039 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3040 * new syntactic sugar for the Unix command line: --load foo.bar is now
3041 an alternate notation for --eval '(load "foo.bar")'.
3043 ** The system now detects stack overflow and handles it gracefully,
3044 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3045 optimization settings. (This is a good thing in general, and
3046 its introduction in this version should be particularly timely
3047 for anyone whose code fails because of suppression of tail
3049 ** The system now hunts for the C variable "environ" in a more
3050 devious way, to avoid segfaults when the C library version differs
3051 between compile time and run time. (thanks to Christophe Rhodes)
3052 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3053 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3054 ** The compiler no longer issues bogus style warnings for undefined
3055 classes in the same source file as the DEFCLASSes which defined
3056 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3058 ** fixes in CONDITION class precedence list for undefined function
3059 errors (thanks to Alexei Dejneka)
3060 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3061 correctly. (thanks to Dan Barlow)
3062 ** portability fixes aiming at bootstrapping under CLISP (thanks
3063 to Dave McDonald and Christophe Rhodes)
3064 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3065 ** fixes in type translation and and type inference (thanks to
3067 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3068 ** various fixes in the new ports (thanks to Dan Barlow)
3069 * several changes related to debugging:
3070 ** suppression of tail recursion, as noted above
3071 ** stack overflow detection, as noted above
3072 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3073 is now the default. (For some time encapsulation has been more
3074 reliable than the breakpoint-based :ENCAPSULATE NIL
3075 implementation, at least on X86 systems; and I just noticed that
3076 encapsulation also seems closer to the spirit of the ANSI
3079 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3081 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3082 up properly again. (There was a packaging bug in 0.7.0 which
3083 left their definitions in SB-SYS::LOAD-FOREIGN and
3084 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3085 vital for most things which interface to C-level interfaces,
3086 like extensions working with sockets or databases or
3087 Perl-compatible regexes or whatever, and the need to fix
3088 this bug is the main reason that 0.7.1 was released so
3090 ** DEFGENERIC is now choosier about the methods it redefines, so that
3091 reLOADing a previously-LOADed file containing DEFGENERICs does
3092 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3093 cycle now works as it should. (thanks to Alexey Dejneka)
3094 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3095 fixed by Christophe Rhodes. (He actually submitted this patch
3096 months ago, and I delayed until after 0.7.0.)
3097 ** Bug 111 (internal compiler confusion about runtime checks on
3098 FUNCTION types) was fixed by Alexey Dejneka.
3099 * Some internal cleanups (getting rid of variables which aren't
3100 needed now that the byte interpreter is gone) caused the fasl
3101 file format number to change again.
3103 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3104 * major incompatible change: The default fasl file extension, i.e. the
3105 default extension for files produced by COMPILE-FILE, has changed
3106 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3108 ** There are many changes in the implementation of the compiler.
3109 SBCL is now essentially a compiler-only implementation of ANSI
3110 Common Lisp. EVAL still "interprets" a few special cases, but
3111 almost all the interesting cases are handled by creating
3112 a LAMBDA expression, calling COMPILE on it, then calling
3113 FUNCALL on the result.
3114 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3115 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3116 Since the code is newer, there might still be some new bugs
3117 (though not as many as before Martin Atzmueller's fixes:-). But
3118 the new code is substantially simpler and clearer, and hopefully
3119 any remaining bugs will be simpler, less fundamental, and more
3120 fixable then the bugs in the old code.
3121 ** The revised compiler is still a little unsteady on its feet.
3123 *** The debugging information it produces (particularly the names
3124 of FUNCTION objects) is sometimes much less useful than what
3125 the old compiler produced.
3126 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3127 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3128 within a MACROLET) has been temporarily weakened.
3129 ** There are new compiler optimizations for various functions:
3130 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3131 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3132 *** the math functions TRUNCATE, FLOOR, and CEILING
3133 *** the function-of-all-trades COERCE
3134 Mostly these should be transparent, but there's one
3135 potentially-annoying problem (bug 117): when the compiler
3136 inline-expands a function and does type analysis on the result,
3137 it can create control paths which have type mismatches, and
3138 when it can't prove that those control paths aren't taken,
3139 it will issue WARNINGs about the type mismatches. This is
3140 a particular problem in practice for the new sequence functions.
3141 It's not clear how this should be fixed, and for now, a
3142 workaround is given in the entry for 117 in the BUGS file.
3143 ** (Because of the interaction between the two previous items --
3144 occasional inlining problems and new inline expansions -- some
3145 of the new sequence function optimizations won't really kick in
3146 completely until debugging information, and then inlining, are
3147 straightened out in some future version.)
3148 * minor incompatible changes:
3149 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3150 DIRECTORY no longer implicitly promotes NIL slots of its
3151 pathname argument to :WILD. In particular, when you ask for the
3152 contents of a directory (which you used to be able to do without
3153 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3154 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3155 ** changes in behavior that ANSI explicitly defines to be
3156 implementation dependent:
3157 *** The new compiler-only implementation still conforms with ANSI,
3158 but acts a little different than before. Besides the obvious
3159 changes in performance tradeoffs (that the cost per form passed
3160 to EVAL has gone up, and the cost per form executed by EVAL
3161 has gone down), the behavior of the system changes a little
3162 because there are no longer any interpreted function objects.
3163 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3164 e.g. doing COMPILE on the output of interactive DEFUN is
3166 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3168 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3169 has changed from (:CL) to NIL.
3170 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3171 e.g. #\Newline and #\Tab, don't have names specified in the
3172 ANSI Common Lisp standard, is now based on their ASCII symbolic
3173 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3174 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3175 are no longer used for output.
3176 ** changes in internal implementation constants:
3177 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3178 4 million. (If your application spends a lot of time GCing and
3179 you have a lot of RAM, you might want to experiment with
3180 increasing it even more.)
3181 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3182 However, almost all old code should still continue to work without
3183 immediate update, as SB-C-CALL is now a (deprecated) nickname
3185 ** Old operator names in the style DEF-FOO are now deprecated in
3186 favor of new corresponding names DEFINE-FOO, for consistency with
3187 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3188 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3189 internal symbols, but a few supported extensions like
3190 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3191 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3192 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3193 etc. as you get deeper into recursive calls to the debugger
3194 command loop, instead of the old "5]", "5]]", "5]]]"
3195 sequence. (I was motivated to do this when squabbles between
3196 ILISP and SBCL left me very deeply nested in the debugger. In the
3197 short term, this change will probably provoke more ILISP/SBCL
3198 squabbles, but hopefully it will be an improvement in the long run.)
3199 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3200 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3201 function, since declaiming return types involving aliens is
3202 (1) annoyingly messy to do by hand and (2) vital to efficient
3203 compilation of code which calls such functions.
3204 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3205 longer reexported by the SB-EXT package. They're solely useful
3206 for alien code, so it seems more logical that you should get
3207 them from the SB-ALIEN package, not in SB-EXT.
3208 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3209 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3210 features. Instead, the code that they used to control is always
3211 built into the system.
3212 * many other bug fixes
3213 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3214 advantage of the new EVAL-WHEN stuff and to clean them up in
3215 general, and they are now more ANSI-compliant in a number of
3216 ways. Martin Atzmueller is responsible for a lot of this.
3217 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3219 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3220 *** correct ERROR type for various file operations
3221 *** some fixes for Lisp streams
3222 *** DEFMETHOD syntax checking
3223 *** changing old weird representation of debug information as
3224 strings (which, among their other deficiencies, don't transform
3225 correctly when you rename packages, and don't change their
3226 print representation when you change things like *PACKAGE*
3227 and *PRINT-LENGTH*) to symbols and lists of symbols
3228 He also made several improvements and fixed several bugs in DESCRIBE.
3229 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3231 *** misbehavior of WRITE-STRING/WRITE-LINE
3232 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3233 and several other LOOP problems as well
3234 *** DIRECTORY when similar filenames are present
3235 *** DEFGENERIC with :METHOD options
3236 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3237 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3238 *** argument ordering in FIND with :TEST option
3239 *** mishandled package designator argument in APROPOS-LIST
3240 *** various problems in the backquote readmacro
3241 *** a bug in APROPOS
3242 *** probably some others that I'm not describing very well here,
3243 since the CVS log documents them by reference to sbcl-devel
3244 messages, and the SourceForge archives aren't working well.:-(
3245 ** Dan Barlow improved the Alpha port (and is making progress on the
3246 PPC port, for those of you who think different).
3247 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3248 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3249 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3250 continues to work on the SPARC port (for those of you in a position
3251 to look down upon our little PC-compatible boxes from a great height).
3252 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3253 and uses that copy, rather than the current dynamic value, when
3254 it's trying to decide whether to truncate output. Thus e.g.
3255 (let ((*print-lines* 50))
3256 (pprint-logical-block (stream nil)
3258 (let ((*print-lines* 8))
3259 (print (aref possiblybigthings i) stream)))))
3260 should now truncate the logical block only at 50 lines, instead of
3261 often truncating it at 8 lines, as it did before.
3262 * The doc/cmucl/ directory, containing old CMU CL documentation
3263 from the time of the fork, is no longer part of the base system.
3264 SourceForge has shut down its anonymous FTP service, and with it
3265 my original plan for distributing the old CMU CL documentation
3266 there. For now, if you need these files you can download an old
3267 SBCL source release and extract them from it.
3268 * The fasl file version number changed again, for dozens of reasons,
3269 some of which are apparent above.
3271 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3272 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3273 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3274 patch, so that DISASSEMBLE output is much nicer.
3275 * The code in the SB-PROFILE package now seems reasonably stable.
3276 I still haven't decided what the final interface should look like
3277 (I'd like PROFILE to interact cleanly with TRACE, since both
3278 facilities use function encapsulation) but if you have a need
3279 for profiling now, you can probably use it successfully with
3280 the current CMU-CL-style interface.
3281 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3282 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3283 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3284 link now returns the dangling link itself, and for similar
3285 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3286 link itself. (In these cases the old code signalled an error and
3287 looped endlessly, respectively.) Thus, DIRECTORY now works even
3288 in the presence of dangling and cyclic symbolic links.
3289 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3290 is now a supported extension again, since the consensus on
3291 sbcl-devel was that it can be useful for ordinary development
3292 work, not just for debugging SBCL itself.
3293 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3294 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3295 that currently-defined functions might be redefined later with
3296 different return types.
3297 * Hash tables can be printed readably, as inspired by CMU CL code
3298 of Eric Marsden and SBCL code of Martin Atzmueller.
3299 * better error handling in CLOS method combination, thanks to
3300 Martin Atzmueller porting Pierre Mai's CMU CL patches
3301 * more overflow fixes for >16Mbyte I/O buffers
3302 * A bug in READ has been fixed, so that now a single Ctrl-D
3303 character suffices to cause end-of-file on character streams.
3304 In particular, now you only need one Ctrl-D at the command
3305 line (not two) to exit SBCL.
3306 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3308 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3309 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3310 does the right thing.
3311 * The implementation of some type tests, especially for CONDITION
3312 types, is now tidier and maybe faster, due to CMU CL code
3313 originally by Douglas Crosher, ported by Martin Atzmueller.
3314 * Some math functions have been fixed, and there are new
3315 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3316 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3317 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3318 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3319 likely to be built into sbcl-0.7.0.)
3320 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3321 CL, ported to SBCL by Martin Atzmueller)
3322 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3323 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3324 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3325 * There's a new slam.sh hack to shorten the edit/compile/debug
3326 cycle for low-level changes to SBCL itself, and a new
3327 :SB-AFTER-XC-CORE target feature to control the generation of
3328 the after-xc.core file needed by slam.sh.
3329 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3330 COMPILE-FILE is no longer supported, so that now every function
3331 gets an entry point, so that block compilation looks a little
3332 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3333 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3334 returns the number of bytes consed since the system started,
3335 rather than the number consed since the first time the function
3336 was called. (The new definition parallels ANSI functions like
3337 CL:GET-INTERNAL-RUN-TIME.)
3338 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3339 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3340 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3341 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3342 * Fasl file version numbers are now independent of the target CPU,
3343 since historically most system changes which required version
3344 number changes have affected all CPUs equally. Similarly,
3345 the byte fasl file version is now equal to the ordinary
3348 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3349 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3350 is no longer recognized. I apologize for this, because it was
3351 listed in SB-EXT as a supported extension, but I found that
3352 its existing behavior was poorly specified, as well as incorrectly
3353 specified, and it looked like too much of a mess to straighten it
3354 out. I have enough on my hands trying to get ANSI stuff to work..
3355 * many patches ported from CMU CL by Martin Atzmueller, with
3356 half a dozen bug fixes in pretty-printing and the debugger, and
3357 half a dozen others elsewhere
3358 * fixed bug 13: Floating point infinities are now supported again.
3359 They might still be a little bit flaky, but thanks to bug reports
3360 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3361 as flaky as they were.
3362 * The --noprogrammer command line option is now supported. (Its
3363 behavior is slightly different in detail from what the old man
3364 page claimed it would do, but it's still appropriate under the
3365 same circumstances that the man page talks about.)
3366 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3367 are now supported, and enabled by default. Thus, the compiler can
3368 handle many floating point and complex operations much less
3369 inefficiently. (Thus e.g. you can implement a complex FFT
3371 * The compiler now detects type mismatches between DECLAIM FTYPE
3372 and DEFUN better, and implements CHECK-TYPE more correctly, and
3373 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3374 thanks to patches from Martin Atzmueller.
3375 * various fixes to make the cross-compiler more portable to
3376 ANSI-conforming-but-different cross-compilation hosts (notably
3377 Lispworks for Windows, following bug reports from Arthur Lemmens)
3378 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3379 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3380 fixed thanks to Pierre Mai's CMU CL patches.
3381 * fixes to keep the system from overflowing internal counters when
3382 it tries to use i/o buffers larger than 16M bytes
3383 * fixed bug 45a: Various internal functions required to support
3384 complex special functions have been merged from CMU CL sources.
3385 (When I was first setting up SBCL, I misunderstood a compile-time
3386 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3387 * improved support for type intersection and union, fixing bug 12
3388 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3389 more obscure bugs as well
3390 * some steps toward byte-compiling non-performance-critical
3391 parts of the system, courtesy of patches from Martin Atzmueller
3392 * Christophe Rhodes has made some debian packages of sbcl at
3393 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3394 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3395 stable, but are nonetheless usable. When he's ready, I'd be happy
3396 to add them to the SourceForge "File Releases" section. (And if
3397 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3398 * new fasl file format version number (because of changes in
3399 internal representation of (OR ..) types to accommodate the new
3400 support for (AND ..) types, among other things)
3402 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3403 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3405 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3407 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3408 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3409 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3410 * DESCRIBE now works on structure objects again.
3411 * Most function call argument type mismatches are now handled as
3412 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3413 know whether the function will be redefined before the call is
3414 executed. (The compiler could flag local calls with full WARNINGs,
3415 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3416 it doesn't keep track of enough information to know whether calls
3417 are local in this sense.)
3418 * Compiler output is now more verbose, with messages truncated
3419 later than before. (There should be some supported way for users
3420 to override the default verbosity, but I haven't decided how to
3421 provide it yet, so this behavior is still controlled by the internal
3422 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3423 src/compiler/ir1util.lisp.)
3424 * Fasl file format version numbers have increased again, because
3425 support for the Gray streams extension changes the layout of the
3426 system's STREAM objects.
3427 * The Gray subclassable streams extension now works, thanks to a
3428 patch from Martin Atzmueller.
3429 * The full LOAD-FOREIGN extension (not just the primitive
3430 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3431 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3432 but like most other programs, it defaults to copying the Unix
3433 environment from the original process instead of starting the
3434 new process in an empty environment.
3435 * Extensions which manipulate the Unix environment now support
3436 an :ENVIRONMENT keyword option which doesn't smash case or
3437 do other bad things. The CMU-CL-style :ENV option is retained
3438 for porting convenience.
3439 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3440 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3442 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3444 * A patch from Martin Atzmueller seems to have solved the SIGINT
3445 problem, and as far as we know, signal-handling now works cleanly.
3446 (If you find any new bugs, please report them!)
3447 * The system no longer defaults Lisp source file names to types
3448 ".l", ".cl", or ".lsp", but only to ".lisp".
3449 * The compiler no longer uses special default file extensions for
3450 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3451 seems to expect a single default extension for all compiled code,
3452 and there's no compelling reason to try to stretch the standard
3453 to allow two different extensions.) Instead, byte-compiled files
3454 default to the same extension as native-compiled files.
3455 * Fasl file format version numbers have increased again, because
3456 a rearrangement of internal implementation packages made some
3457 dumped symbols in old fasl files unreadable in new cores.
3458 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3459 many fewer weird special cases.
3460 * Bug #17 (differing COMPILE-FILE behavior between logical and
3461 physical pathnames) has been fixed, and some related misbehavior too,
3462 thanks to a patch from Martin Atzmueller.
3463 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3464 by Tim Moore, ported to SBCL by Martin Atzmueller.
3465 * Martin Atzmueller fixed several filesystem-related problems,
3466 including bug #36, in part by porting CMU CL patches, which were
3467 written in part by Paul Werkowski.
3468 * More compiler warnings in src/runtime/ are gone, thanks to
3469 more patches from Martin Atzmueller.
3470 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3473 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3475 * DESCRIBE now works on CONDITION objects.
3476 * The debugger now handles errors which arise when trying to print
3477 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3479 * The build system now uses an additional file, customize-target-features.lisp,
3480 to allow local modifications to the target *FEATURES* list. (The point of
3481 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3482 debugging features enabled, without having to worry about propagating it
3483 into everyone's system when I do a "cvs update".) When no
3484 customize-target-features.lisp file exists, the target *FEATURES* list
3485 should be constructed the same way as before.
3486 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3487 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3488 warn about behavior which is undefined under the ANSI spec. Note: This
3489 is specified by ANSI, but it's not very popular with programmers.
3490 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3491 macro in the SBCL sources for an example of a workaround which you
3492 might use to make portable ANSI-standard code which does what you want.
3493 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3494 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3495 when it's used as a non-toplevel form. (This is required in order
3496 to implement the DEFCONSTANT-EQX macro.)
3497 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3498 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3499 funny things, due to bugs in the implementation of EVAL-WHEN
3500 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3501 non-toplevel DEFCONSTANTs are uncommon.)
3502 * The core file version number and fasl file version number have been
3503 incremented, because the old noncompliant DEFCONSTANT behavior involved
3504 calling functions which no longer exist, and because I also took the
3505 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3506 * fixed bug 1 (error handling before read-eval-print loop starts), and
3507 redid debugger restarts and related debugger commands somewhat while
3509 ** The QUIT debugger command is gone, since it did something
3510 rather different than the SB-EXT:QUIT command, and since it never
3511 worked properly outside the main toplevel read/eval/print loop.
3512 Invoking the new TOPLEVEL restart provides the same functionality.
3513 ** The GO debugger command is also gone, since you can just invoke
3514 the CONTINUE restart directly instead.
3515 ** The TOP debugger command is also gone, since it's redundant with the
3516 FRAME 0 command, and since it interfered with abbreviations for the
3518 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3520 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3521 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3522 some porting work by Martin Atzmueller)
3523 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3524 with a boatload of patches which he ported to SBCL. Now that those
3526 ** The system tries to make sure that its low-priority messages
3527 are prefixed by semicolons, to help people who like to use
3528 syntax highlighting in their ILISP buffer. (This patch
3529 was originally due to Raymond Toy.)
3530 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3531 patches originally written by Raymond Toy.
3532 ** The compiler understands coercion between single-value and
3533 multiple-VALUES type expressions better, getting rid of some very
3534 weird behavior, thanks to patches originally by Robert MacLachlan
3535 and Douglas Crosher.
3536 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3537 lambda lists, thanks to a patch originally by Pierre Mai.
3538 ** The system no longer bogusly warns about "abbreviated type
3540 ** The compiler gets less confused by inlining and RETURN-FROM,
3541 thanks to some patches originally by Tim Moore.
3542 ** The system no longer hangs when dumping circular lists to fasl
3543 files, thanks to a patch originally from Douglas Crosher.
3544 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3545 undefined function error.
3546 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3547 in structure slot initforms) for now, documented workaround instead:-|
3548 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3549 * fixed bug 5: added stubs for various Gray stream functions called
3550 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3551 installed, at least appropriate type errors are generated
3552 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3553 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3554 consistently in DEFMETHOD forms.
3555 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3556 it doesn't seem to affect SBCL after all
3557 * The C runtime system now builds with better optimization and many
3558 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3560 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3562 * The system is now under CVS at SourceForge (instead of the
3563 CVS repository on my home machine).
3564 * The new signal handling code has been tweaked to treat register
3565 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3566 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3567 downstream, e.g. in debug-int.lisp, has implicit dependencies
3568 on the unsignedness of integer representation of machine words,
3569 and that caused the system to bomb out with infinite regress
3570 when trying to recover from type errors involving signed values,
3571 e.g. (BUTLAST '(1 2 3) -1).)
3572 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3573 (This was one of the bugs Peter Van Eynde reported back in July.)
3574 * The system now uses code inspired by Colin Walters' O(N)
3575 implementation of MAP (from the cmucl-imp@cons.org mailing
3576 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3577 inline the MAP operation, and there is more than one
3578 sequence argument to the MAP call (so that it can't just
3579 do ETYPECASE once and for all based on the type of the
3580 single sequence argument). (The old non-inline implementation
3581 of the general M-argument sequence-of-length-N case required
3582 O(M*N*N) time when any of the sequence arguments were LISTs.)
3583 * The QUIT :UNIX-CODE keyword argument has been renamed to
3584 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3585 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3587 * Raymond Wiker's patches to port dynamic loading from Linux to
3588 FreeBSD have been added.
3589 * The BUGS file is now more nearly up to date, thanks in large part
3590 to Martin Atzmueller's review of it.
3591 * The debugger now flushes standard output streams before it begins
3592 its output ("debugger invoked" and so forth).
3593 * The core version number and fasl file version number have both
3594 been incremented, because of incompatible changes in the layout
3596 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3597 used to not be called for a saved Lisp image.)
3598 * Martin Atzmueller's version of a patch to fix a compiler crash,
3599 as posted on sbcl-devel 13 September 2000, has been installed.
3600 * Instead of installing Martin Atzmueller's patch for the
3601 compiler transform for SUBSEQ, I deleted the compiler transform,
3602 and transforms for some similar consing operations.
3603 * A bug in signal handling which kept TRACE from working on OpenBSD
3605 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3606 away by constant folding
3607 * The system now defines its address space constants in one place
3608 (in the Lisp sources), and propagates them automatically elsewhere
3609 (through GENESIS and the sbcl.h file). Therefore, patching the
3610 address map is less unnecessarily tedious and error-prone. The
3611 Lisp names of address space constants have also been systematized.
3612 * CVS tags like dollar-Header-dollar have been removed from
3613 the sources, because they have never saved me trouble and
3614 they've been source of trouble working with patches and other
3615 diff-related operations.
3616 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3617 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3619 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3621 * The system has been ported to OpenBSD.
3622 * The system now compiles with a simple "sh make.sh" on the systems
3623 that it's supported on. I.e., now you no longer need to tweak
3624 text in the target-features.lisp-expr and symlinks in src/runtime/
3625 by hand, the make.sh takes care of it for you.
3626 * The system is no longer so grossly inefficient when compiling code
3627 involving vectors implemented as general (not simple) vectors (VECTOR T),
3628 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3629 which dares to use the various sequence functions on non-simple
3630 vectors, takes less of a performance hit.
3631 * There is now a primitive type predicate VECTOR-T-P
3632 to test for the (VECTOR T) type, so that e.g.
3633 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3634 can now be compiled with some semblance of efficiency. (The old code
3635 turned the type declaration into a full call to %TYPEP at runtime!)
3636 * AREF on (VECTOR T) is still not fast, since it's still compiled
3637 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3638 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3639 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3640 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3641 without making full calls to SUBTYPEP at runtime.
3642 (Some analogous efficiency issues for non-simple vectors specialized to
3643 element types other than T, or for non-simple multidimensional arrays,
3644 have not been addressed. They could almost certainly be handled the
3645 same way if anyone is motivated to do so.)
3646 * The changes in array handling break binary compatibility, so
3647 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3648 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3649 instead of (VALUES T T).
3650 * By following the instructions that Dan Barlow posted to sbcl-devel
3651 on 2 July 2000, I was able to enable primitive dynamic object
3652 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3653 functionality is not implemented (since it calls ld to resolve
3654 library references automatically, requiring RUN-PROGRAM for its
3655 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3656 try to resolve library references) is now supported.
3657 * The system now flushes the standard output streams when it terminates,
3658 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3659 them at several other probably-convenient times, e.g. in each pass of
3660 the toplevel read-eval-print loop, and after evaluating a form given
3661 as an "--eval" command-line option. (These changes were motivated by a
3662 discussion of stream flushing issues on cmucl-imp in August 2000.)
3663 * The source transform for TYPEP of array types no longer assumes
3664 that an array whose element type is a not-yet-defined type
3665 is implemented as an array of T, but instead punts, so that the
3666 type will be interpreted at runtime.
3667 * There is now some support for cross-compiling in make.sh: each of
3668 the phases of make.sh has its own script. (This should be transparent
3669 to people doing ordinary, non-cross-compile builds.)
3670 * Since my laptop doesn't have hundreds of megabytes of memory like
3671 my desktop machine, I became more motivated to do some items on
3672 my to-do list in order to reduce the size of the system a little:
3673 ** Arrange for various needed-only-at-cold-init things to be
3674 uninterned after cold init. To support this, those things have
3675 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3676 symbols with such names are now uninterned after cold init).
3677 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3678 fasl files for cold load.
3679 ** Remove the old compiler structure pooling code (which used to
3680 be conditional on the target feature :SB-ALLOC) completely.
3681 ** Redo the representation of some data in cold init to be more compact.
3682 (I also looked into supporting byte compiled code at bootstrap time,
3683 which would probably reduce the size of the system a lot, but that
3684 looked too complicated, so I punted for now.)
3685 * The maximum signal nesting depth in the src/runtime/ support code has
3686 been reduced from 4096 to 256. (I don't know any reason for the very
3687 large old value. If the new smaller value turns out to break something,
3688 I'll probably just bump it back up.)
3689 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3691 * Many, many bugs reported by Peter Van Eynde have been added to
3692 the BUGS list; some have even been fixed.
3693 * While enabling dynamic object file loading, I tried to make the
3694 code easier to understand, renaming various functions and variables
3695 with less ambiguous names, and changing some function calling
3696 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3697 * While trying to figure out how to do the OpenBSD port, I tried to
3698 clean up some of the code in src/runtime/. In particular, I dropped
3699 support for non-POSIX signal handling, added various comments,
3700 tweaked the code to reduce the number of compilation warnings, and
3701 renamed some files to increase consistency.
3702 * To support the new automatic configuration functionality in make.sh,
3703 the source file target-features.lisp-expr has been replaced with the
3704 source file base-target-features.lisp-expr and the machine-generated
3705 file local-target-features.lisp-expr.
3706 * fixed a stupid quoting error in make.sh so that using CMU CL
3707 "lisp -batch" as cross-compilation host works again
3709 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3711 * DESCRIBE no longer tries to call itself recursively to describe
3712 bound/fbound values, so that it no longer fails on symbols which are
3713 bound to themselves (like keywords, T, and NIL).
3714 * DESCRIBE now works on generic functions.
3715 * The printer now prints less-screwed-up representations of closures
3716 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3717 * A private symbol is used instead of the :EMPTY keyword previously
3718 used to mark empty slots in hash tables. Thus
3719 (DEFVAR *HT* (MAKE-HASH-TABLE))
3720 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3721 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3722 now does what ANSI says that it should. (You can still get
3723 similar noncompliant behavior if bang on the hash table
3724 implementation with all the symbols you get back from
3725 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3726 This breaks binary compatibility, since tests for equality to
3727 :EMPTY are wired into things like the macroexpansion of
3728 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3730 * There's now a minimal placeholder implementation for CL:STEP,
3731 as required by ANSI.
3732 * An obscure bug in the interaction of the normal compiler, the byte
3733 compiler, inlining, and structure predicates has been patched
3734 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3735 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3737 * Missing ordinary arguments in a macro call are now detected even
3738 when the macro lambda list contains &KEY or &REST.
3739 * The debugger no longer complains about encountering the top of the
3740 stack when you type "FRAME 0" to explicitly instruct it to go to
3741 the top of the stack. And it now prints the frame you request even
3742 if it's the current frame (instead of saying "You are here.").
3743 * As specified by ANSI, the system now always prints keywords
3744 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3746 * The default initial SIZE of HASH-TABLEs is now smaller.
3747 * Type information from CLOS class dispatch is now propagated
3748 into DEFMETHOD bodies, so that e.g.
3749 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3751 is now basically equivalent to
3752 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3753 (DECLARE (TYPE SINGLE-FLOAT X))
3755 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3756 operation, without having to do run-time type dispatch.
3757 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3758 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3759 * Since I don't seem to be making big file reorganizations very often
3760 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3761 is overflowing my ability to conveniently back them up), I've finally
3762 checked the system into CVS. (The CVS repository is on my home system,
3763 not at SourceForge -- putting it on SourceForge might come later.)
3764 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3765 high-level GC-NOTIFY-FOO functions send their output. (There's
3766 still very little control of where low-level verbose GC functions
3767 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3768 less than it used to -- the GC-NOTIFY-FOO functions are now under
3769 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3770 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3771 in only one place in the source code, and propagates it automatically
3772 everywhere that it's needed. Thus e.g. when I bump the version from
3773 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3774 * The C source files now include boilerplate legalese and documentation
3775 at the head of each file (just as the Lisp source files already did).
3776 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3777 to his page will be replaced with a link to his new CLiki service.
3779 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3781 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3782 * The build process now looks for GNU make under the default name "gmake",
3783 instead of "make" as it used to. If GNU make is not available as "gmake"
3784 on your system, you can change this default behavior by setting the
3785 GNUMAKE environment variable.
3786 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3787 can build without error under CMU CL.
3789 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3791 * There is now a partial SBCL user manual (with some new text and some
3792 text cribbed from the CMU CL manual).
3793 * The beginnings of a profiler have been added (starting with the
3794 CMU CL profiler and simplifying and cleaning up). Eventually the
3795 main interface should be through the TRACE macro, but for now,
3796 it's still accessed through vaguely CMU-CL-style functions and macros
3797 exported from the package SB-PROFILE.
3798 * Some problems left over from porting CMU CL to the new
3799 cross-compilation bootstrap process have been cleaned up:
3800 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3801 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3802 part when getting rid of the compiler *BACKEND* stuff.)
3803 ** The value of *NULL-TYPE* was screwed up, because it was
3804 being initialized before the type system knew the final
3805 definition of the 'NULL type. This screwed up several key
3806 optimizations in the compiler, causing inefficiency in all sorts
3807 of places. (I found it because I wanted to understand why
3808 GET-INTERNAL-RUN-TIME was consing.)
3809 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3810 PROCLAIM FTYPE information. Unfortunately this broke binary
3811 compatibility again, since now the forms output by DEFGENERIC
3812 to refer to functions which didn't exist in 0.6.3.
3813 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3814 can use the new (as of 0.6.3) transform for SOME into MAP into
3816 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3817 values to (INTEGER 0 1000000), so that the time code will no longer
3818 occasionally get blown up by Linux returning 1000000 microseconds
3819 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3820 its output conform to the ANSI spec. (Alas, this makes its output
3821 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3822 * A full call to MAP NIL with a single sequence argument no longer conses.
3823 * fixes to problems pointed out by Martin Atzmueller:
3824 * The manual page no longer talks about multiprocessing as though
3825 it were currently supported.
3826 * The ILISP support patches have been removed from the distribution,
3827 because as of version 5.10.1, ILISP now supports SBCL without us
3828 having to maintain patches.
3829 * added a modified version of Raymond Toy's recent CMU CL patch for
3830 EQUALP comparison of HASH-TABLE
3832 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3834 * The system still can't cross-compile itself with
3835 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3836 entails), but at least it can compile more of itself that way
3837 than it used to be able to, and various buglets which were uncovered
3838 by trying to cross-compile itself that way have now been fixed.
3839 * This release breaks binary compatibility again. This time
3840 at least I've incremented the FASL file format version to 2, so that the
3841 problem can be detected reliably instead of just causing weird errors.
3842 * various new style warnings:
3843 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3844 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3845 ** using the lexical binding of a variable named in the *FOO* style
3846 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3847 as specified by ANSI.
3848 * *RANDOM-STATE* is no longer automatically initialized from
3849 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3850 default behavior of the system is to repeat its behavior every time
3851 it's run. If you'd like to change this behavior, you can always
3852 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3853 old convention there was no comparably easy way to get the system to
3854 repeat its behavior every time it was run.
3855 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3856 FTYPE declarations has been removed, in favor of their ANSI
3857 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3858 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3859 the types of their sequence arguments can be determined at compile time.
3860 This is done through a new open code expansion for MAP which eliminates
3861 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3862 argument types can be determined at compile time.
3863 * The optimizer now transforms COERCE into an identity operation when it
3864 can prove that the coerced object is already of the correct type. (This
3865 can be a win for machine generated code, including the output of other
3866 optimization transforms, such as the MAP transform above.)
3867 * Credit information has been moved from source file headers into CREDITS.
3868 * Source file headers have been made more standard.
3869 * The CASE macro now compiles without complaining even when it has
3872 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3874 * (Note that the way that the PCL macroexpansions were rewritten
3875 to accommodate the change in DEFGENERIC below breaks binary
3876 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3877 not run under sbcl-0.6.2. Once we get out of alpha releases,
3878 i.e. hit release 1.0.0, we'll probably try to maintain binary
3879 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3880 and sbcl-1.4.4. Until then, however, it might be fairly common
3881 for maintenance releases to break binary compatibility.)
3882 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3884 * The reporting of unhandled signals has been changed to print some
3885 explanatory text as well as the report form. (Previously only
3886 the report form was printed.)
3887 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3888 it defines, so that the compiler no longer issues undefined function
3889 warnings for compiled-but-not-yet-loaded generic functions.
3890 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3891 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3892 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3893 section of the ANSI spec are supported.
3894 * The "" nickname for the "KEYWORD" package has been removed.
3895 The reader still handles symbol tokens which begin with a package marker
3896 as keywords, but it doesn't expose its mechanism for doing so in the
3897 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3898 * The system now issues STYLE-WARNINGs for contradictory TYPE
3899 proclamations. (Warnings for contradictory FTYPE proclamations would
3900 be nice too, but those can't be done usefully unless the type system
3901 is made smarter about FUNCTION types.)
3902 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3903 systematized, so that "*target-*.lisp is supposed to exist only on the
3904 target and imply that there's a related file which exists on the
3905 host, and *host-*.lisp is supposed to exist only on the host and imply
3906 that there's a related file which exists on the target. This involves a
3907 lot of renaming. Hopefully the acute confusion caused by the renaming
3908 will be justified by the reduction in chronic confusion..
3909 ** runtime-type.lisp -> early-target-type.lisp
3910 ** target-type.lisp -> late-target-type.lisp
3911 ** early-host-format.lisp -> early-format.lisp
3912 ** late-host-format.lisp -> late-format.lisp
3913 ** host-error.lisp -> misc-error.lisp
3914 ** early-error.lisp -> early-target-error.lisp
3915 ** late-error.lisp -> late-target-error.lisp
3916 ** host-defboot.lisp -> early-defboot.lisp
3917 ** code/misc.lisp -> code/target-misc.lisp
3918 ** code/host-misc.lisp -> code/misc.lisp
3919 ** code/numbers.lisp -> code/target-numbers.lisp
3920 ** code/early-numbers.lisp -> numbers.lisp
3921 ** early-host-type.lisp -> early-type.lisp
3922 ** late-host-type.lisp -> late-type.lisp
3923 ** host-typep.lisp -> typep.lisp
3924 ** load.lisp -> target-load.lisp
3925 ** host-load.lisp -> load.lisp
3926 ** host-disassem.lisp -> disassem.lisp
3927 ** host-insts.lisp -> insts.lisp
3928 ** byte-comp.lisp -> target-byte-comp.lisp
3929 ** host-byte-comp.lisp -> byte-comp.lisp
3930 ** host-signal.lisp -> signal.lisp
3931 ** host-defstruct.lisp -> defstruct.lisp
3932 ** late-target-type.lisp -> deftypes-for-target.lisp
3933 Furthermore, several other previously target-only files foo.lisp (e.g.
3934 hash-table.lisp and random.lisp) have been split into a target-and-host
3935 foo.lisp file and a target-only target-foo.lisp file, with their key type
3936 definitions in the target-and-host part, so that the cross-compiler will
3937 know more about target types.
3938 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3939 gone away. In their place are various *BACKEND-FOO* variables
3940 corresponding to the slots of the old structure.
3941 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3942 into the target SBCL has been fixed.
3943 * The chill.lisp system for loading cold code into a running SBCL
3945 * Support for the CMU CL "scavenger hook" extension has been removed.
3946 (It was undocumented and unused in the CMU CL sources that SBCL was
3947 derived from, and stale in sbcl-0.6.1.)
3948 * Various errors in the cross-compiler type system were detected
3949 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3950 (enabling various consistency checks). Many of them were fixed,
3951 but some hard problems remain, so the compiler is back to
3952 running without *TYPE-SYSTEM-INITIALIZED* for now.
3953 * As part of the cross-compiler type system cleanup, I implemented
3954 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3955 * I have started adding UNCROSS calls throughout the type system
3956 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3957 on cross-compiler input files will be able to go away, and various
3959 * CONSTANTP now returns true for quoted forms (as explicitly required
3962 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3964 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3965 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3966 (MAKE-LIST -1). In the longer term, it ought to become true
3967 that declarations are assertions even at SAFETY 1. For now, it's not
3968 quite true even at SAFETY 3, but it's at least more nearly true..
3969 (Note that this change seems to increases the size of the system by
3970 O(5%) and to decrease the speed of the compiler by 20% or more.)
3971 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3972 for the problem of printing dozens of lines of distracting information
3973 about low-level system machinery as part of the top stack frame
3974 on entry to the debugger when an undefined function was called.
3975 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3976 is not reset to COMMON-LISP-USER.
3977 * Compilation of stuff related to dyncount.lisp has been made conditional
3978 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3979 smaller. The various dyncount-related symbols have been moved into
3980 a new "SB-DYNCOUNT" package.
3981 * tty-inspect.lisp has been renamed to inspect.lisp.
3982 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3983 feature has gone away. (When we eventually port to other flavors of
3984 libc and/or Unix, we'll try to make the differences between flavors
3985 invisible at the user level.)
3986 * Various other *FEATURES* tags, and/or their associated conditionals,
3987 have been removed if obsolescent, or given better documentation, or
3988 sometimes given more-mnemonic names.
3990 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3992 * tidied up "make.sh" script
3993 * tidied up system directory structure
3994 * better "clean.sh" behavior
3995 * added doc/FOR-CMUCL-DEVELOPERS
3996 * many many small tweaks to output format, e.g. removing possibly-confusing
3997 trailing #\. character in DESCRIBE-INSTANCE
3998 * (EQUALP #\A 'A) no longer signals an error.
3999 * new hashing code, including EQUALP hashing
4000 * tidied up Lisp initialization and toplevel
4001 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4002 * command line argument processing
4003 * added POSIX-GETENV function to deal with Unix-ish environment variables
4004 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4005 terminating SBCL on EOF
4006 * non-verbose GC by default
4007 * There is no more "sbcl" shell script; the sbcl file is now the C
4008 runtime executable (just like CMU CL).
4009 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4011 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4012 symbol and package information as Lisp native symbol and package objects
4013 instead of strings naming symbols and strings naming packages. This way,
4014 whenever packages are renamed (as in warm init), debug information is
4015 transformed along with everything else.
4016 * tweaked the optimization policy declarations which control the building
4017 of SBCL itself. Now, among other things, the system no longer saves
4018 source location debugging information. (This helps two problems at once
4019 by reducing SBCL size and by keeping SBCL from trying to look for its
4020 sources -- which may not exist -- when reporting errors.)
4021 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4022 debugging and testing purposes
4023 * cleaned up printing, making the printer call PRINT-OBJECT for
4024 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4025 methods, giving nearly-ANSI behavior
4026 * converted almost all special variables to use *FOO* naming convention
4027 * deleted PARSE-TIME functionality, since it can be done portably
4028 * moved some files out of cold init into warm init
4029 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4031 * regularized formatting of source files
4032 * added an install.sh script
4033 * fixed ridiculous memory usage of cross-compiler by making
4034 compiler/alloc.lisp not try to do pooling unless it can hook
4035 itself into the GC of the cross-compilation host. Now the system
4036 builds nicely on my old laptop.
4037 * added :SB-ALLOC in target-features.lisp-expr
4038 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4039 was not implemented)
4040 * re-did condition handling and note reporting in the compiler. Notes
4041 are no longer handled by signalling conditions. Style warnings
4042 and warnings are handled more correctly and reported in such a way
4043 that it's easy to find one or the other in your output (so that you
4044 can e.g. figure out which of many problems caused COMPILE-FILE to
4046 * changed the severity of several compiler warnings from full WARNING
4047 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4048 compiler note reporting so that it doesn't use the condition system
4049 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4050 COMPILE-FILE command)
4051 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4052 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4053 no longer supported, since their implementation was deeply intertwingled
4054 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4055 * removed bogus "support" for compiler macros named (SETF FOO), and
4056 removed the compiler macro for SETF INFO (but only after making a fool
4057 of myself on the cmucl-imp mailing list by posting a bogus patch for
4058 DEFINE-COMPILER-MACRO..)
4059 * Compiled files containing forms which have side effects on the Lisp
4060 reader (such as DEFPACKAGE forms) are now handled more correctly.
4061 (Compiler queuing of top level lambdas has been suppressed by setting
4062 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4063 * deleted various currently-unused source files, e.g. gengc.lisp. They
4064 may be added back at some point e.g. when porting to other architectures,
4065 but until they are it's distracting to distribute them and to try to
4067 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4068 there were so many of them they're just distractions, and UNCROSS is
4069 known to be able to handle the current sources
4070 * moved PROFILE functionality into TRACE, so that it will be clear
4071 how the wrapping and unwrapping of functions when you profile them
4072 interacts with the wrapping and unwrapping of functions when you
4073 trace them. (Actually, the functionality isn't there yet, but at least
4074 the interface specification is there. Hopefully, the functionality will
4075 arrive with some future maintenance release.)
4076 * removed host-oops.lisp
4077 * changed signature of QUIT function to allow UNIX-CODE argument
4078 * fixed READ-SEQUENCE bug
4079 * tweaked verbose GC output so that it looks more like the progress
4080 output that ANSI specifies for functions like LOAD
4081 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4082 * added <http://sbcl.sourceforge.com> to the banner information printed by