1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.19 (1.0.0?) relative to sbcl-0.9.18:
3 * improvement: runtime option --dynamic-space-size can be used
4 to set the size of the dynamic space reserved on startup.
5 * improvement: floating point modes in effect are now saved in
6 core, and restored on startup.
7 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
8 startup, not time since first call to GET-INTERNAL-REAL-TIME.
9 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
10 threads are not running after *SAVE-HOOKS* have run.
11 * improvement: writes to CLOS instance slots are type-checked in code
12 compiled with (SAFETY 3)
13 * improvement: floating-point exception handling on FreeBSD (thanks to
15 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
16 (thanks to Zach Beane)
17 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
19 * improvement: added support for the Shift-JIS external format.
20 (contributed by NIIMI Satoshi)
21 * improvement: callbacks are supported on Linux/PPC. (thanks to
23 * bug fix: compiler bug triggered by a (non-standard) VALUES
24 declaration in a LET* was fixed. (reported by Kaersten Poeck)
25 * bug fix: file compiler no longer confuses validated and already
26 dumped structurres. (reported by Kaersten Poeck)
27 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
28 fill-pointer signals a type-error as required. (thanks to
30 * bug fix: disassemly of funcallable instances works.
31 * bug fix: single stepping on PPC.
32 * bug fix: fix thread-safety problems in the type system (generally
33 manifesting as nonsensical errors like "STRING is a bad type specifier
34 for sequences" or "The value 1 is not of type FIXNUM")
35 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
36 (reported by Josip Gracin)
37 * bug fix: an error is signaled for attempts to displace arrays with
38 incompatible element types (thanks to Mario Mommer)
39 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
40 and don't cause extra consing
41 * optimization: MAP and MAP-INTO are significantly faster on vectors
42 whose elements types have been declared.
43 * Improvements to SB-SPROF:
44 ** Support for allocation profiling
45 ** Reduced profiling overhead, especially for long profiling runs
46 * Improvements to the Windows port:
47 ** floating point exceptions are now reported correctly.
48 ** stack exhaustion detection works partially.
49 ** more accurate GET-INTERNAL-REAL-TIME.
50 ** better RUN-PROGRAM behaviour in case of errors in the child
52 ** PROBE-FILE now simplifies pathnames correctly.
53 ** DIRECTORY now works correctly with :WILD-INFERIORS.
54 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
55 1st 1970 (although time timezone data will be unreliable).
56 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
58 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
59 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
60 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
62 * bug fix: two potential GC deadlocks affecting threaded builds.
63 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
64 returns T (reported by Anton Kazennikov)
65 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
66 with non-variable places
67 * bug fix: remove a race condition in the setting of
68 funcallable-instance functions, this should make threaded CLOS
69 code more stable against memory faults.
70 * bug fix: corruption of specials when unbinding is interrupted by an
71 asynchronous unwind (reported by Hannu Koivisto)
72 * improvement: the debugger will now also display local variables that
73 are only used once, for code compiled with a DEBUG optimization quality
76 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
77 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
78 * incompatible change: External-format support for FFI calls. The
79 SB-ALIEN:C-STRING no longer implies an ASCII
80 external-format. Instead the string is subject to external-format
81 conversion. Additionally return values of type SB-ALIEN:C-STRING
82 are by default of type (SIMPLE-ARRAY CHARACTER), not
83 SIMPLE-BASE-STRING. For an alien type that behaves like the old
84 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
85 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
86 * incompatible change: SB-EXT package no longer contains the
87 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
88 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
90 * incompatible change: the single-stepper is no longer available
91 on Alpha, Mips and Sparc platforms.
92 * minor incompatible change: the direct superclasses of
93 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
94 not (STANDARD-OBJECT FUNCTION). This makes the
95 class-precedence-lists of GENERIC-FUNCTION and
96 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
98 * minor incompatible change: the default stream external format on
99 non-unicode SBCL is no longer detected from the locale, but is always
101 * new feature: Add a version of evaluator that uses an interpreter instead
102 of the compiler. EVAL still uses the compiler by default, to switch it
103 to use the interpreter, set the value of the variable
104 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
105 * minor incompatible change: the single-stepper REPL has been merged
106 with the normal debugger (see the "Stepping" heading of the debugger help
107 for more details). The debugger command STEP will no longer switch
108 to the single-stepper REPL.
109 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
110 for a type now works.
111 * bug fix: loading of default sysinit file works. (thanks to Leonid
113 * bug fix: better detection of circularities in the file-compiler.
114 (reported by Marco Monteiro)
115 * bug fix: the CL pathname functions now work with files that have
116 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
117 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
118 systems with Unix98 pty semantics.
119 * bug fix: ASDF-INSTALL will now work with bsd tar.
120 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
122 * bug fix: timers expiring in dead threads no longer cause a
123 type-error (reported by Paul "Nonny Mouse").
124 * bug fix: thanks to more lightweight single-stepper instrumentation,
125 code compiled with (DEBUG 3) will compile and execute significantly faster,
126 and will have more accurate type-inferencing than before
127 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
128 whose bindings are modified
129 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
130 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
131 ** replace dummy implementations of CL:MACHINE-INSTANCE and
132 CL:SOFTWARE-VERSION with proper version
134 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
135 * feature: implemented the READER-METHOD-CLASS and
136 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
137 as specified by AMOP.
138 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
140 * optimization: faster LOGCOUNT implementation on x86 and x86-64
141 (thanks to Lutz Euler)
142 * optimization: hashing of general arrays and vectors has been
143 improved. (reported by Any Fingerhut)
144 * enhancement: SB-INTROSPECT is now able to find definitions of
145 profiled functions. (thanks to Troels Henriksen)
146 * enhancement: compiler-macro expansion applies now to FUNCALL forms
147 as well, allowing compiler-macros for SETF-functions to expand.
148 * enhancement: step-instrumentation no longer wraps known
149 single-value functions with multiple-value context, allowing
150 better type inference.
151 * fixed bug #337: use of MAKE-METHOD in method combination now works
152 even in the presence of user-defined method classes. (reported by
153 Bruno Haible and Pascal Costanza)
154 * fixed bug #339(c): if there are applicable methods not part of any
155 long-form method-combination group, call INVALID-METHOD-ERROR.
156 (reported by Bruno Haible)
157 * fixed bug #361: the :FUNCTION initarg in the protocol for
158 initialization of methods can now be used to override
159 internally-produced optimized functions. (reported by Bruno
161 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
162 system-provided lambda expression no longer cause warnings about
163 unbound #:|pv-table| symbols.
164 * bug fix: improved the handling of type declarations and the
165 detection of violations for keyword arguments with non-constant
167 * bug fix: potentially erronous calls to PATHNAME and
168 MERGE-PATHNAMES were being flushed in some cases.
169 (reported by Richard Kreuter)
170 * bug fix: compiled calls to TYPEP were mishandling obsolete
171 instances. (reported by James Bielman and Attila Lendvai)
172 * bug fix: format strings with ~> without matching ~< no longer
173 trigger an AVER, but signal an understandable error instead.
174 (reported by Antonio Martinez)
175 * bug fix: specifying an output-file without a pathname-type for
176 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
177 of input-file instead of "fasl". (reported by Robert Dodier)
178 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
179 (reported by James Y Knight).
180 * bug fix: compiler-macros-function did not consider the environment
181 argument for shadowing by local functions.
182 * bug fix: compiler-macros expansion was inhibited by local INLINE
184 * bug fix: inline expansions of known functions were subject to
185 step-instrumentation in high debug policies, leading to problems
187 * bug fix: compiler failed to differentiate between different CONS
189 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
190 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
191 * bug fix: #368: incorrect use of expressed vs. upgraded array
193 * bug fix: #306a: more precise unions of array types.
194 * thread-safety improvements:
195 ** CONDITION-WAIT could return early on Linux, if the thread was
196 interrupted and subsequently continued with SIGCONT.
197 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
199 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
200 * added support for the ucs-2 external format. (contributed by Ivan
202 * minor incompatible change: pretty printing of objects of type
203 (cons symbol) is, in the default pprint-dispatch-table, now
204 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
206 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
207 called later in a class's lifetime, possibly as late as when the
208 first instance of the class is created. Previously,
209 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
210 class became finalizeable.
211 * fixed bug: FILE-POSITION sometimes returned inconsistent results
212 for multibyte external-format streams. (thanks to "vbzoli")
213 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
214 with :ALLOCATION :CLASS inherited from superclasses of the
216 * fixed bug: anonymous classes can now be created using the :NAME
217 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
218 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
219 * fixed bug: core-files saved with :EXECUTABLE T can again be
220 executed when SBCL_HOME isn't set. (reported by James Knight)
221 * fixed bug: toplevel LOCALLY forms with declarations could
222 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
223 * fixed bug: printing from several different threads using different
224 values of *print-case* could cause invalid output, due to
225 some internal special variables of the printer not being bound
226 thread-locally (reported by Max Mikhanosha)
227 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
228 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
229 (reported by Pascal Costanza)
230 * fixed bug: FINALIZE-INHERITANCE is called from
231 REINITIALIZE-INSTANCE on classes when the class has previously
232 been finalized, as required by AMOP.
233 * minor code generation optimizations:
234 ** better register allocation in CLOS dispatching functions
235 ** overflow detection when coercing signed bytes to fixnums on x86-64
236 ** is now implemented with one IMUL instruction instead of three shifts
237 ** more efficient bit-vector access on x86 and x86-64
238 ** more efficient access to raw structure slots on x86 and x86-64
239 * fixed some bugs revealed by Paul Dietz' test suite:
240 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
243 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
244 * feature: thread support on Solaris/x86, and experimental thread support
246 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
248 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
249 default method for SLOT-UNBOUND.
250 * minor incompatible change: prevent the user from specializing the
251 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
252 somewhat counter to the intent of the protocol, I (CSR) think, and
253 additionally it just doesn't work in SBCL as currently
254 implemented, thanks to optimizations (that are always valid for
255 the other three SLOT-VALUEish functions, but not for the setter).
256 * minor incompatible change: the :SB-LDB feature is now enabled by
257 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
258 the low-level debugger.
259 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
260 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
261 down to an SB-SYS:FD-STREAM, allowing direct communication in
262 more cases. Temporary files and pipes are still used as a fallback
264 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
265 on the non-reentrant C functions localtime(3) and gmtime(3).
266 * bug fix: global optimization policy was not visible in LOCALLY and
268 * bug fix: class objects can be used as specializers in methods.
269 (reported by Pascal Costanza)
270 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
271 with a type error. (reported by blitz_ on #lisp)
272 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
273 failed with a type error. (reported by Pascal Bourguignon)
274 * bug fix: merging pathnames against defaults with :DIRECTORY
275 starting with '(:RELATIVE :BACK) should preserve the :BACK.
276 (reported by James Y Knight)
277 * bug fix: saving large (>2GB) cores on x86-64 now works
278 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
279 constant index and a (simple-array (signed-byte 32)) array
280 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
281 error (patch by Robert J. Macomber)
282 * feature: TIME now displays the amount of run-time spent in GC
283 * bug fix: The debugger now does a better job of respecting
284 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
285 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
286 workaround for bug 403.)
287 * bug fix: floating point exception handling now works on FreeBSD
288 (thanks to NIIMI Satoshi)
289 * fixed some bugs revealed by Paul Dietz' test suite:
290 ** MISC.641: LET-conversion were not supposed to work in late
292 * improvements to the Win32/x86 port:
293 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
294 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
296 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
297 * new feature: source path information is generated for macro-expansion
298 errors for use in IDE's like Slime (thanks to Helmut Eller)
299 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
300 * bug fix: compiler does not loop forever on an invalid type in
302 * improvement: compilation of most CLOS applications is significantly
304 * optimization: added a limited bytecode compiler for simple toplevel
305 forms, speeding up compilation and FASL loading
306 * bug fix: the statistical profiler now properly distinguishes anonymous
309 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
310 * minor incompatible change: in sbcl-0.9.11 (but not earlier
311 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
312 into subclasses of the finalized class. Now user calls to
313 FINALIZE-INHERITANCE finalize just the one class, and calls by the
314 system return before any subclasses are finalized.
315 * minor incompatible change: The reader no longer ignores errors
316 regarding non-existent packages in #+ and #- feature tests.
317 * new feature: command line options --no-sysinit, --no-userinit to
318 inhibit loading the corresponding init files
319 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
320 for sending data through UDP sockets (thanks to François-René Rideau)
321 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
322 error is signalled instead (thanks to François-René Rideau)
323 * minor incompatible change: Error signalling behaviour of lexical
324 operations violating package locks has changed slightly. Refer to
325 documentation on package locks for details.
326 * enhancement: EVAL can process IF-expressions without resorting to the
328 * bug fix: Some operations involving SIN, COS, or TAN weren't being
329 constant-folded properly due to missing out-of-line functions.
330 (reported by Mika Pihlajamäki)
331 * bug fix: LISTEN sometimes returned T even in cases where no data was
332 immediately available from the stream
333 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
334 were reversed. (reported by Levente Mészáros)
335 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
336 when given a vector displaced to an adjustable array. (reported
338 * bug fix: derivation of float boundaries from numbers outside the
339 appropriate float range (reported by John Wiseman)
340 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
342 * bug fix: precomputing cache entries for generic functions with
343 some subclasses of specializers as yet invalid does not attempt to
344 fill a cache line with a negative offset. (reported by Levente
346 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
347 allow condition class objects as arguments to DOCUMENTATION and
348 (SETF DOCUMENTATION); only find and set documentation for
349 structure names for the STRUCTURE doc-type. (suggested by Gary
351 * improvements to the Win32/x86 port:
352 ** (user-homedir-pathname) and default initialization file
353 locations now know about the user's "Documents and Settings"
354 directory (thanks to Yaroslav Kavenchuk)
355 ** run-program is implemented (thanks to Mike Thomas)
356 ** sockets support (thanks to Timothy Ritchey)
357 ** better backtrace support (thanks to Alastair Bridgewater)
358 ** sb-grovel supported
359 ** asdf-install and sb-posix work somewhat
360 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
361 * improvements to the Solaris/x86 port:
362 ** works on Solaris 11/Solaris Express
363 ** floating-point exception handling support
364 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
365 * fixed some bugs revealed by Paul Dietz' test suite:
366 ** REMOVE-METHOD returns its generic function argument even when
367 the method is not one of the generic functions' methods.
368 ** objects declared as MEMBER types can be admissible arguments to
370 ** printing characters should simply be printed by the FORMAT ~:C
372 ** compiler failure when compiling functions with hairy constant
373 defaults for optional parameters.
374 ** compiler produces wrong code when MAYBE-INLINE-expanding a
375 function, which is already optimized.
377 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
378 * new platform: experimental support for SBCL x86/Darwin, including
379 MacOS X 10.4.5 on Intel.
380 * new feature: Unicode character names are now known to the system
381 (through CHAR-NAME and NAME-CHAR).
382 * new feature: the filesystem location of SBCL's core file is
383 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
384 * minor incompatible change: the contrib modules SB-POSIX and
385 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
386 this change is to make it easier to distribute
387 location-independent binaries.
388 * bug fix: as implied by AMOP, standardized classes no longer have
389 slots named by external symbols of public packages. (reported by
391 * optimization: calling functions via constant symbols -- as in
392 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
393 via the function object as in (FUNCALL #'FOO).
394 * optimization: on x86 and x86-64, the calling convention for the
395 case of dealing with an unknown number of values has been altered
396 to be friendlier to the prediction heuristics implemented,
397 particularly with reference to CALL and RET pairing. (thanks to
398 Alastair Bridgewater)
399 * optimization: on x86, the code for access to thread-local symbol
400 values and binding/unbinding of thread-local symbols is smaller.
401 (thanks to Alastair Bridgewater)
402 * enhancement: CONSTANTP is now able to determine constantness of
403 more complex forms, including calls to constant-foldable standardized
404 functions and some special forms beyond QUOTE.
405 * fixed bug: occasional GC crashes on Solaris/x86
406 * optimization: x86-64 supports stack allocation of results of simple
407 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
408 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
409 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
410 (thanks to James Knight)
411 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
412 by the standard (thanks to Stephen Compall)
414 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
415 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
416 be used for bundling the runtime and the core file into one
417 executable binary. This feature is not currently supported on all SBCL
418 platforms. (thanks to James Bielman and NIIMI Satoshi)
419 * new feature: a generational or ephemeral garbage collector is now
420 the default on the PowerPC platform (both Linux and Darwin). The
421 old Cheney (stop and copy) collector is a build-time option.
422 (thanks to Cyrus Harmon, after Raymond Toy)
423 * minor incompatible change: the method by which SBCL finds its
424 contributed modules has changed; it no longer relies on symbolic
425 links from an $SBCL_HOME/systems directory, but searches directly
426 in the subdirectories of $SBCL_HOME.
427 * enhancement: the dynamic heap size on the Linux/PPC platform is
428 markedly larger, even using the older Cheney garbage collector.
429 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
430 certain complicated string types. (reported by Gary King)
431 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
432 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
433 * fixed bug: DOCUMENTATION on structures no longer signals an
434 error if no structure type of the right name exists. (reported by
436 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
437 applicable non-standard primary or :AROUND method on
438 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
439 no longer cause unbound variable errors. (reported by Kevin Reid)
440 * fixed bug: metacircle resolution in cases where methods have slots
441 added before the slots from STANDARD-METHOD. (reported by Jean
443 * fixed bug: the Power PC assembly code for calling into lisp
444 sometimes computed its return address wrongly (depending on the
445 15th bit of the address, fixed at link time). This used to
446 manifest itself in a segmentation violation while building PCL.
447 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
448 many others over the years)
449 * fixed bug: ROOM no longer reports silly numbers for stack usage if
450 the stack pointer (treated as a signed integer) is negative.
451 (thanks to Peter van Eynde)
453 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
454 * new platform: experimental support for the Windows operating
455 system has been added. (thanks to Alastair Bridgewater)
456 * fixed several bugs in and robustified the PPC FFI (including
457 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
458 * bug fix: allow non-simple string symbol names (reported by Paul
460 * bug fix: interrupt handling on NetBSD (thanks to Richard M
462 * bug fix: saving a core corrupted callbacks on x86/x86-64
463 * bug fix: closed a loophole in metacircularity detection and
464 grounding in the PCL implementation of CLOS.
465 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
466 (reported by tomppa on #lisp)
467 * bug fix: clear the direction flag on Lisp -> C transitions, as
468 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
469 (reported by Andrej Grozin and Hendrik Maryns)
470 * optimization: major improvements to GC efficiency on GENCGC platforms
471 * optimization: faster implementation of EQUAL
472 * optimization: emit more efficient opcodes for some common
473 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
474 * optimization: several other minor code-generation improvements on x86-64
475 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
477 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
478 * minor incompatible change: (SETF CLASS-NAME) and (SETF
479 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
480 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
481 by ANSI as a generic function, and for consistency (SETF
482 GENERIC-FUNCTION-NAME) is treated likewise.
483 * fixed bug #233.b: make constraint propagation notice when a variable
484 value is changed after it is referenced but before it is used
485 * fixed bug #296: no more arbitrary behaviour from filenames with
486 odd characters as --load arguments. (reported by Adam Warner)
487 * fixed bug #390: :CHARACTER-SET pathname components now work as
488 expected. (reported by Tim Daly Jr)
489 * fixed bug #391: complicated :TYPE intersections in slot
490 definitions no longer cause an error in PCL internals.
491 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
492 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
493 returning the number of octets which would be written to the
494 file-stream. (thanks to Robert J. Macomber)
495 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
496 arguments to the internal stream functions. (thanks to David
498 * bug fix: stack exhaustion is now handled on the x86/SunOS
499 platform. (thanks to Daisuke Homma)
500 * bug fix: more accurate ROOM results on GENCGC platforms
501 * optimization: improved type inference for arithmetic-for
502 index variables in LOOP
503 * optimization: faster floating-point SQRT on x86-64
504 * fixed some bugs revealed by Paul Dietz' test suite:
505 ** DOCUMENTATION returns NIL instead of "" for method combinations
506 that don't have a docstring
508 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
509 * minor incompatible change: (SETF CLASS-NAME) and (SETF
510 GENERIC-FUNCTION-NAME) are no longer generic functions, and
511 therefore are not customizeable by user code (as seems to be at
512 least permitted and maybe required by AMOP). As a consolation,
513 however, the SBCL implementation of these functions now calls
514 REINITIALIZE-INSTANCE as specified by AMOP.
515 * bug fix: slot-definition documentation is propagated as per ANSI
516 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
517 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
518 argument, as do eslotd initialization methods. (from Pascal
519 Costanza's "Closer" project)
520 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
521 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
523 * bug fix: it is now possible to have more than one subclass of
524 STANDARD-GENERIC-FUNCTION without causing stack overflow.
525 (reported by Bruno Haible, Pascal Costanza and others)
526 * bug fix: the dependent update protocol now works for generic
527 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
529 * bug fix: condition-class instances corresponding to
530 DEFINE-CONDITION forms are now created eagerly. (reported by
531 Kalle Olavi Niemitalo on comp.lang.lisp)
532 * bug fix: floating point printing is more accurate in some
533 circumstances. (thanks to Simon Alexander)
534 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
535 merged with *DEFAULT-PATHNAME-DEFAULTS*.
536 * bug fix: callbacks on OS X now preserve stack-alignment.
537 (thanks to Cyrus Harmon)
538 * enhancement: the x86-64 disassembler is much better at
539 disassembling SSE instructions. (thanks to Lutz Euler)
540 * enhancement: CHECK-TYPE on a variable now allows type-inference
541 (thanks to Vincent Arkesteijn)
542 * optimization: improved performance of EUC-JP external format.
543 (thanks to NIIMI Satoshi)
544 * optimization: performance improvements to IO on file streams of
545 :ELEMENT-TYPE CHARACTER
546 * optimization: much faster memory allocation on x86-64
547 * optimization: faster garbage collections (latency and throughput)
549 * optimization: faster bignum subtraction and division on x86 and x86-64
550 (thanks to Lutz Euler)
551 * optimization: more accurate type-inference for loops that use a
552 floating point index variable or a negative step.
554 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
555 * bug fix: add a workaround to SBCL looping infinitely at startup on
556 Linux kernels with apparently buggy implementations of personality().
557 (thanks to Svein Ove Aas)
558 * bug fix: Unicode symbols are correctly printed in LDB backtraces
559 (thanks to David Lichteblau)
560 * bug fix: local bindings shadow global symbol macros for
561 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
562 * optimization: non-open coded uses of numeric comparison operators
563 (e.g. >) no longer cons when called with more than one parameter
564 on platforms supporting dynamic-extent allocation.
565 * enhancement: saving cores with foreign code loaded is now supported
566 on MIPS/Linux in addition to the previously supported platforms.
567 * bug fix: FIND-RESTART now tests for activity, not applicability when given
568 a restart object as identifier. (reported by Helmut Eller for CMUCL)
569 * bug fix: division by zero in sb-sprof when no samples were collected
570 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
571 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
572 (thanks to Alexander Kjeldaas)
573 * new feature: ignoring the return values of destructive functions
574 like NREVERSE signals a compile-time style-warning.
575 (thanks to Kevin Reid)
576 * new feature: the alignment of alien structure fields can be
577 explicitly specified. (thanks to Cyrus Harmon)
578 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
579 and dump core on SIGQUIT
581 ** incompatible change: threads do not inherit values of specials
582 from their parents (see manual)
583 ** bug fix: threads stacks belonging to dead threads are freed by the
584 next exiting thread, no need to gc to collect thread stacks anymore
585 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
586 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
587 ** bug fix: dynamic variable and thread start related gc lossage
588 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
590 * fixed some bugs revealed by Paul Dietz' test suite:
591 ** SUBTYPEP is slightly more accurate on heinously complicated
592 CONS types where some of the members have uncertain (in the
593 NIL, NIL sense) type relationships to each other.
595 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
596 no longer purified unless :PURIFY T is explicitly specified.
597 ** Non-purified cores are significantly smaller than before
599 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
600 * new feature: timers based on Zach Beane's excellent timer package
601 * added support for the following external formats: koi8-u,
602 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
603 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
604 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
605 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
606 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
607 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
608 * incompatible change: a threaded SBCL will no longer revert to
609 non-threaded mode on non-NPTL systems, but refuse to start entirely.
610 * bug fix: interrupts are disabled until startup is complete; no
611 more sigsegvs when receiving a signal to soon
612 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
614 * bug fix: add a workaround for the memory randomization features in
615 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
616 workaround will only be in effect on systems with the proc filesystem
618 * bug fix: printing objects of type HASH-TABLE signals a
619 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
621 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
622 have been running for over 50 days. (reported by Gilbert Baumann)
623 * bug fix: the logic for getting names of functions gets less
624 confused when confronded with alternate-metaclass
625 funcallable-instances. (reported by Cyrus Harmon)
626 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
627 consistent, even on internal alternate-metaclass objects.
628 * bug fix: flush closure information collected by physical
629 environment analysis prepass before the main pass. (bug reported
631 * bug fix: compiler pack phase does not modify a hashtable, which is
632 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
633 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
634 may be expanded. (reported by Alexander <malishev> on c.l.l)
635 * bug fix: no more style-warnings when compiling code using LAST
636 under high-SPEED low-SPACE settings. (reported by David Wallin
638 * fixed some incompatibilities between SBCL's MOP and the MOP
640 ** the METAOBJECT class is now implemented;
641 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
642 STANDARD-OBJECT, as required;
643 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
644 now compatible (as required by VALIDATE-SUPERCLASS); there
645 remains a consistency requirement over the presence or absence
646 of the FUNCTION class in the superclass list of finalized
647 classes; see the manual for more details;
648 ** the :AROUND method for COMPUTE-SLOTS on
649 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
650 requested slot ordering.
652 ** bug fix: parent thread now can be gc'ed even with a live
654 ** bug fix: binding a special with PROGV to no value is not
655 the same as that symbol not having been bound (thanks to
657 ** bug fix: binding specials is thread safe (thanks to
659 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
660 and SB-SPROF do not die with 'no handler for signal XX in
661 interrupt_handle_now(..)' anymore
662 ** bug fix: WITH-TIMEOUT works with multiple running threads
663 * fixed some bugs revealed by Paul Dietz' test suite:
664 ** ENSURE-GENERIC-FUNCTION should take a method class object for
665 the :method-class keyword argument.
667 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
668 * new port: the Solaris operating system on x86 processors is now
669 mostly supported, though some rough edges in the environment
670 remain. (thanks to Daisuke Homma)
671 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
672 space for its heap. (thanks to Thiemo Seufer)
673 * optimization: SBCL on MIPS platforms now supports dynamic-extent
674 closures. (thanks to Thiemo Seufer)
675 * minor incompatible change: eof selects abort in the debugger.
676 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
677 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
679 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
680 deprecated, and will go away in a future revision of SBCL.
681 * minor incompatible change: GC-ON and GC-OFF are no longer
682 implemented with a counter, it does not matter how many times gc
683 is switched on or off
684 * bug fix: discriminating functions for generic function classes
685 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
686 make invalid assumptions about method precedence order. (reported
688 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
689 platforms. (thanks to Thiemo Seufer)
690 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
691 * bug fix: correct pathname printing with printer escaping is on.
692 (thanks to Kevin Reid)
693 * bug fix: complex VOP definitions in "user-space" no longer trigger
694 package locks. (reported by Zach Beane)
695 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
696 for funcallable-instance functions, and is no different from
698 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
699 characters in the current readtable. (reported by Nicholas Neuss)
700 * bug fix: SUBTYPEP on various CONS types returns more a more
701 accurate acknowledgment of its certainty.
702 * optimizations: REMOVE-DUPLICATES now runs in linear time on
703 lists in some cases. This partially fixes bug 384.
704 * flush all standard streams before prompting in the REPL and the
706 * bug fix: signal handling and triggering gc do not conflict
707 directly anymore, in particular a high frequency sb-sprof does
708 not prevent gc from running
709 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
710 approximation for timezone and DST information between the
711 universal time epoch and the smallest negative 32-bit time_t.
712 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
713 year 1899 so long as the encoded time is non-negative.
714 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
715 the correct answer for non-constant rotate amounts when performing
716 an inline 32-bit rotation.
718 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
719 there is only one thread in the session
720 ** bug fix: memory leak for streams created in one thread and
721 written to in another
722 ** bug fix: lockup when compiled with gcc4
723 ** bug fix: race that allows the gc to be triggered when gc is
725 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
726 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
728 ** bug fix: finished threads are gc'ed properly
729 * fixed some bugs revealed by Paul Dietz' test suite:
730 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
731 has arguments with a different set of applicable methods from
732 the orignal arguments.
733 ** The type error thrown by MAP now has a correct expected-type
735 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
736 name a compiled function.
737 ** Three MISC tests where a large bignum was improperly coerced to
738 a float that couldn't represent that bignum during type
739 derivation were fixed.
740 ** SUBTYPEP can now handle types involving the negation of a
741 list-form FUNCTION type.
742 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
743 as (COMPLEX (AND RATIO (NOT FIXNUM))).
744 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
746 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
747 * New feature: Experimental support for bivalent streams: streams
748 opened with :element-type :default now allow character and binary
749 (unsigned-byte 8) I/O
750 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
751 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
753 * bug fix: no more highly sporadic "couldn't check whether ~S is
754 readable" when reading a stream and an interrupt hits in the middle
755 of a select system call
756 * compiler better recognizes complex arrays (reported by Eduardo
758 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
759 for variables on SBCL built with linkage-tables. (reported by Luis
761 * various error reporting improvements.
762 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
763 (Thanks to Thiemo Seufer)
764 * bug fix: nested reader invokations work correctly
765 * bug fix: it is possible to have simultaneous references to foreign
766 code and foreign data with the same name.
768 ** added x86-64 support
769 ** incompatible change: the threading api now works with thread
770 objects instead of thread ids
771 ** bug fix: threads are protected from signals and interruption when
772 starting up or going down
773 ** bug fix: a race where an exiting thread could lose its stack to gc
774 ** bug fix: don't halt on infinite error in threads if possible
775 ** fixed numerous gc deadlocks introduced in the pthread merge
776 ** bug fix: fixed thread safety issues in read and print
777 ** bug fix: debugger doesn't hang on session lock if interrupted at
778 an inappropriate moment
779 ** bug fix: run-program is now thread safe(r)
780 ** bug fix: inner with-recursive-lock no longer releases the mutex
781 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
782 * fixed some bugs revealed by Paul Dietz' test suite:
783 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
784 EXPECTED-TYPE slots filled.
785 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
787 ** MAKE-CONDITION accepts classes as type-designators.
788 ** COMPILE may never return NIL.
789 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
790 range before calling Unix time functions
792 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
793 * numerous signal handling fixes to increase stability
794 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
795 * minor incompatible change: we now correctly canonize default
796 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
797 as per the MOP, rather than the historical (INITARG INITFUNCTION
798 INITFORM). (reported by Bruno Haible)
799 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
800 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
801 for more information.
802 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
803 pathname is a directory pathname.
804 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
805 * dynamic space size on PPC has been increased to 768Mb. (thanks to
807 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
808 :METACLASS argument in addition to a class name. (reported by
809 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
810 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
811 (reported by Bruno Haible; patch from Gerd Moellmann)
812 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
814 * bug fix: more cleanups to the floating point exception handling on
815 x86-64 (thanks to James Knight)
816 * bug fix: the compiler does not try to constant fold calls of
817 COERCE and BIT-* functions when they return freshly constructed
818 objects. (reported by Nikodemus Siivola and Paul Dietz)
819 * optimization: improved the MIPS versions of generic (in the
820 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
821 * optimization: direct conversion of (unsigned-byte 32) to floats on
822 the PowerPC platform.
823 * optimization: structure instances with raw slots now use less
824 memory, and probably show better memory locality. (thanks to
826 * optimization: DYNAMIC-EXTENT declarations for lists and closures
827 are treated as requests for stack allocation on the x86-64,
828 Alpha-32, PPC and SPARC platforms.
829 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
830 the wrong file descriptor; implementation of SOCKET-OPEN-P.
831 (thanks to Tony Martinez)
833 ** gcing a dead thread can no longer lead to lockups
834 ** threads block signals until they are set up properly
835 ** errno is no longer shared by threads
836 ** interrupt-thread restores the eflags register on x86
837 ** fixed some lockups due to gc/thread interaction
838 * fixed some bugs revealed by Paul Dietz' test suite:
839 ** invalid dotted lists no longer raise a read error when
841 ** signal an error if a symbol that names a declaration is used
842 as the name of a type, or vice versa
843 ** allow using the (declare (typespec var*)) abbreviation for
844 (declare (type typespec var*)) with all type specifiers
845 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
846 ** always return NIL from PPRINT-POP when OBJECT is NIL
847 ** don't signal errors when pretty-printing malformed LABELS,
848 FLET or MACROLET forms
849 ** declarations in a DOLIST body are also in scope for the
851 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
853 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
856 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
857 * fixed cross-compiler leakages that prevented building a 32-bit
858 target with a 64-bit host compiler.
859 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
860 opened with :IF-EXISTS :OVERWRITE.
861 * fixed bug 281: error for an invalid qualifier in a short-form method
862 combination method is not signalled until the faulty method is called.
863 * bug fix: iteration variable type inferrer failed to deal with open
864 intervals. (reported by Alan Shields)
865 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
866 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
867 or more runtime options were provided to the sbcl binary.
868 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
869 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
870 * optimization: there's now a fast-path for fixnum arguments in the
871 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
872 * optimization: the code generated on x86-64 is more compact thanks
873 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
874 * bug fix: floating-point exception handling now partly works on
875 x86-64. (Thanks to James Knight)
876 * improvement to the MIPS backend from Thiemo Seufer: C-style
877 64-bit long long arguments and return values to alien functions
879 * fixed some bugs revealed by Paul Dietz' test suite:
880 ** the type-error signalled from WARN has a filled-in DATUM slot.
881 ** the type-error required when a stream is not associated with
882 a file has the stream as its datum.
883 ** type-errors on single-floats on x86-64 no longer have
884 :INVALID-OBJECT as the datum
885 ** the type-errors signalled for invalid function names now have
886 a correct expected type
887 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
888 for structures defined with a :TYPE. Documentation strings for
889 typed structures are no longer immediately discarded
890 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
891 broadcast streams and synonym streams. FILE-LENGTH now also works
892 on broadcast streams.
894 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
895 * incompatible change: the --noprogrammer option, deprecated since
896 version 0.7.5, has been removed. Please use the equivalent
897 --disable-debugger option instead.
898 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
900 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
901 inoperational for a while now) has been completely removed.
902 * null lexical environments are now printed as #<NULL-LEXENV>,
903 significantly reducing the amount of clutter in typical backtraces.
904 * documentation on weak pointers, finalization, and after GC hooks
905 has been added to the manual.
906 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
907 as other specialized array types, is much faster. SUBSEQ and
908 COPY-SEQ on such arrays have also been sped up.
909 * optimization: EQL is now more efficient when at least other argument
910 is known to be of type (OR FIXNUM (NOT NUMBER)).
911 * fixed bug: the runtime is now less vulnerable to changes in the
912 size of the SBCL object on OS X, and virtual memory is reserved for
913 all spaces that need to be at a fixed address.
914 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
915 * fixed bug: finalizers and after GC hooks that cause consing are now
917 * fixed bug: compiler error messages and summaries are now printed to
918 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
919 * fixed inference of the upper bound of an iteration variable.
920 (reported by Rajat Datta).
921 * fixed bug 211e: calling local functions with duplicated constant
922 keyword argument no longer causes a bogus style warning about an
924 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
925 ftype declaration to be disregarded. (reported by Dave Roberts)
926 * fixed bug 373: caused by erronous compilation of references to alien
927 variables in the runtime on ppc/darwin.
928 * fixed bug 376: CONJUGATE type deriver.
929 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
930 (reported by Baughn on #lisp)
931 * a cross-compiler bug on non-x86 platforms has been identified and
932 fixed. (thanks to Bruno Haible)
933 * improvements to the MIPS runtime code for increased stability.
934 (thanks to Thiemo Seufer)
935 * increased the maximimum compact environment size to allow
936 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
937 * partial workaround for bug 135: don't name the function we're
938 calling for hairy cases of EVAL, so as not to accumulate one environment
939 entry per call to EVAL. (reported by Kevin Reid)
940 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
941 * optimization: type testing for non-vector arrays should be faster.
942 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
944 * bug fix: setting 31st element of a bit vector to zero did not work
946 * bug fix: redefining a class definition which failed due to a
947 previous accessor / function clash now works (but see BUGS entry
948 #380 for more problems in this area). (thanks to Zach Beane)
949 * the long form of DEFINE-METHOD-COMBINATION disables method group
950 checking when given a single method group with pattern *. (thanks
952 * on x86 compiler supports stack allocation of results of simple
953 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
954 * support for building and running on Mac OS X 10.4 "Tiger" has been added
955 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
956 * fixed some bugs related to Unicode integration:
957 ** the restarts for recovering from input and output encoding
958 errors only appear when there is in fact such an error to
960 * fixed some bugs revealed by Paul Dietz' test suite:
961 ** MISC.549 and similar: late transformation of unsafe type
962 assertions into derived types caused unexpected code
964 ** SCALE-FLOAT type deriver is less wrong.
965 ** type derivers for EXP, LOG and similar functions compute result
966 types for complex arguments better.
967 ** (MISC.563) CONJUGATE type deriver works for very restricted
969 ** out-of-line type testers for character strings are available.
970 ** EQUAL compiler transform understands specialness of objects
972 ** accessing double-floats stored on the stack now works on x86-64.
973 ** debugger internals could sometimes create invalid lispobjs,
974 resulting in GC crashes.
975 ** MISC.548: type check weakening can convert required type into
977 ** initialization forms for bindings are not in scope of free special
980 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
981 * incompatible change: thread support for non-NPTL systems has
982 been removed - locking is buggy and unreliable. A threaded
983 SBCL build will now warn at startup and refuse to create
984 new threads, unless futex support is detected in the kernel
985 * incompatible change: the top level REPL now has only an ABORT
986 restart associated with it, not TOPLEVEL and ABORT as it used to.
987 TOP and TOPLEVEL are now available as debugger commands for
988 returning to the top level.
989 * incompatible change: forms evaluated in the REPL now use the
990 global optimization policy.
991 * incompatible change: user- and system-initialization files are
992 no longer processed with LOAD, but by READ and EVAL; hence the
993 global optimization policy, startup package, readtable, etc,
995 * The .fasl file format number has been incremented because of
996 various incompatible changes.
997 * internal entry point details and argument counts no longer appear
998 in backtraces unless explicitly requested by setting
999 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1000 * built-in and standard functions no longer have names like "top
1001 level local call to FOO".
1002 * fixed bug 32: functions defined in non-null lexical environments
1003 now have more legible printed representation
1004 * fixed bug 33: functions defined in non-null lexical environemnts
1005 are now more amenable to inspection by INSPECT.
1006 * workaround for bug 354: XEPs no longer appear in backtraces unless
1007 explicitly requested.
1008 * fixed bug: receiving the signal which results from attempting to
1009 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1010 SIGBUS on other *BSDs) on architectures where the C stack is also
1011 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1012 to debugger confusion or wild instability. (reported by Cheuksan
1014 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1015 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1016 (reported by Lutz Euler)
1017 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1018 compiled code. (reported by Paul Dietz)
1019 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1020 specialized parameter to a method to something that is not TYPEP
1021 the specializer is now possible.
1022 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1023 face of package deletion.
1024 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1025 pathname. (reported by David Tolpin on c.l.l)
1026 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1027 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1028 STRING-STREAM and FILE-STREAM, respectively.
1029 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1030 than previously; in addition, its test suite is now run on build.
1031 (thanks to Luke Gorrie)
1032 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1033 * optimization: added a immediate representation for single-floats
1035 * fixed some bugs revealed by Paul Dietz' test suite:
1036 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1037 ** attempting to create a package with a colliding nickname causes
1038 correctable errors to be signalled.
1039 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1040 ** several x86-64 backend bugs related to sign-extension of immediate
1043 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1044 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1045 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1047 * fixed loading of multiply forward-referenced layouts.
1048 (thanks to Cheuksan Wang)
1049 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1050 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1051 (see COMPILE-FILE documentation for details of :PRINT option)
1052 * fixed bugs 19 and 317: fixed-format floating point printing is
1053 more accurate. This also fixes a bug reported by Adam Warner
1054 related to the ~@F format directive.
1055 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1057 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1058 dispatch table if the from-char is a dispatch macro character.
1059 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1060 by the wordsize no longer ignore the last word. (reported by Lutz
1062 * fixed bug in type checking of dynamic-extent variables. (reported
1064 * optimization: sequence traversal functions use their freedom to
1065 coerce function designators to functions.
1066 * optimization: code with many calls to CLOS methods specialized on
1067 CLOS classes has had redundant type checks removed.
1068 * optimization: added declarations to speed up operations that access
1069 the internal character database (for example STRING-UPCASE)
1070 * optimization: comparison operations between floats and sufficiently small
1071 fixnums no longer create extra rationals
1072 * fixed some bugs related to Unicode integration:
1073 ** portions of multibyte characters at the end of buffers for
1074 character-based file input are correctly transferred to the
1075 start of the buffer at the next read.
1076 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1077 passing it through to OPEN.
1078 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1079 argument given to internal calls to OPEN.
1080 ** the built-in comment readers (introduced by character sequences
1081 ";" and "#|") are more forgiving to encoding errors; they will
1082 STYLE-WARN and then attempt to resync the stream at a character
1083 boundary. (thanks to Teemu Kalvas)
1084 * fixed some bugs in the x86-64 port:
1085 ** Negative short int return values from c-calls are sign-extended
1087 ** The stack is aligned to 16-bytes for c-calls, as required by
1088 the ABI. (thanks to Cheuksan Wang)
1089 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1090 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1091 for enums. (thanks to Vincent Arkesteijn)
1092 ** Multiple small optimizations and bugfixes for floating point
1094 * fixed some bugs revealed by Paul Dietz' test suite:
1095 ** Space, Tab, Linefeed, Return and Page have the invalid
1096 secondary constituent character trait.
1097 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1099 ** WITH-INPUT-FROM-STRING should only update the index place on
1101 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1102 works more reliably.
1103 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1104 with one-dimensional arrays with fill pointers.
1105 ** TYPE-OF failed on a complex with an integer realpart and a
1107 ** compiler failure during type inference for the code of form
1108 (IF (EQL X (THE ...)) ...) (MISC.535).
1110 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1111 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1112 platform. The port supports 61-bit fixnums, large memory spaces
1113 and reloading shared object files.
1114 * enhancement: saving cores with foreign code loaded is now
1115 supported on ppc/Darwin in addition to the previously supported
1117 * enhancement: the statistical profiler now walks deeper into the
1118 call stack for more meaningful call-graphs and accrued time
1119 reports (x86/x86-64 only). It also now reports time spent in
1121 * enhancement: it is now possible to trace most individual methods
1122 of a generic function in addition to tracing the generic function
1124 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1126 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1127 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1128 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1129 produces an error. (thanks to Vincent Arkesteijn)
1130 * bug fix: NAMESTRING on pathnames with :WILD components in their
1131 directories works correctly. (thanks to Artem V. Andreev)
1132 * fixed bug 125: compiler preserves identity of closures. (reported
1134 * bug fixed: functions with &REST arguments sometimes failed with
1135 "Undefined function" when compiled with (DEBUG 3). (reported by
1137 * bug fix: overflow during compiling of setting element of a bit
1138 vector with constant index and value. (reported by Timmy Douglas)
1139 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1140 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1141 * on x86 compiler partially supports stack allocation of dynamic-extent
1143 * GO and RETURN-FROM do not check the extent of their exit points
1144 when compiled with SAFETY 0.
1145 * fixed some bugs related to Unicode integration:
1146 ** encoding and decoding errors are now much more robustly
1147 handled; it should now be possible to recover even from invalid
1148 input or output to the terminal. (thanks to Teemu Kalvas)
1149 ** provided a first cut at implementing STRING-TO-OCTETS and
1150 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1151 ** altered the SB-MD5 contributed module slightly, changing the
1152 interface just enough for it to be supportable for builds where
1153 lisp characters are not eight bits.
1154 * fixed some bugs revealed by Paul Dietz' test suite:
1155 ** the FORMATTER-generated functions for ~V[ conditionals require
1156 the correct number of arguments.
1157 ** READ-FROM-STRING returns the mandated second value when applied
1158 to displaced strings.
1159 ** the #\Rubout and #\Backspace characters are treated as invalid
1160 constituent characters by the tokenizer.
1162 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1163 * new feature: reloading changed shared object files with
1164 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1165 * new feature: references to foreign variables and functions
1166 can now be compiled and loaded before the corresponding shared
1167 object file is loaded, as long as the foreign definitions are
1168 available at runtime.
1169 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1170 supported. (thanks to Dan Debertin)
1171 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1172 just keywords. (thanks to Vincent Arkesteijn)
1173 * fixed bug #331: structure-class instances corresponding to
1174 DEFSTRUCT forms are now created eagerly.
1175 * fixed bug #345: backtraces from calls to undefined functions work
1176 on x86 as well. Related bug #61 is now also partially fixed on x86
1177 (backtraces from throws to unknown catch tags.)
1178 * bug fix: lambda-list parsing is now stricter vrt. order and number
1179 of lambda-list keywords.
1180 * bug fix: as specified by AMOP, an error is signalled if a
1181 class-option appears multiple times in a DEFCLASS form. (reported
1183 * bug fix: RANDOM can be compiled when the compiler derives the type
1184 of its numeric argument as a disjoint set of small integers.
1185 (reported by Paul Dietz)
1186 * bug fix: starting a core saved with shared objects loaded when
1187 those objects are not available no longer causes threaded SBCL to
1188 hang. (reported by Sean Ross)
1189 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1191 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1192 argument, which must always be NIL. (reported by Kalle Niemitalo)
1193 * bug fix: printing 1.0d+23 no longer results in an error.
1194 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1195 * bug fix: structure slot setters preserve evaluation order. (thanks
1197 * fixed some bugs related to Unicode integration:
1198 ** RUN-PROGRAM can allow its child to take input from a Lisp
1199 stream. (reported by Stefan Scholl)
1200 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1201 locales. (reported by Ken Causey)
1202 * fixed some bugs revealed by Paul Dietz' test suite:
1203 ** INCF, DECF and REMF evaluate their place form as specified in
1205 ** FORMATTER expands ~{ ~} iteration directives with V or #
1206 parameters correctly.
1207 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1208 consequent uses no arguments correctly.
1209 ** the system has a partial understanding of the (COMPLEX RATIO)
1211 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1214 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1215 * new feature: a build-time option (controlled by the :SB-UNICODE
1216 keyword feature, enabled by default) for building the system with
1217 support for the entire 21-bit character space defined by the
1219 * new feature: the system now has rudimentary external-format
1220 support; the primary user-visible change at this time is that
1221 characters with the high bit set (such as lower-case-e-acute) will
1222 print correctly to a terminal in a UTF-8 environment.
1223 * minor incompatible change: BASE-CHAR no longer names a class;
1224 however, CHARACTER continues to do so, as required by ANSI.
1225 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1226 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1227 has been moved to the SB-EXT package (temporarily re-exported from
1229 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1230 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1232 * fixed bug #351: better error-handlind and reporting for malformed
1234 * fixed bug #350: bignum-printing is now more memory-efficient,
1235 allowing printing of very large bignums, eg. (expt 2 10000000).
1236 (reported by Bruno Haible)
1237 * fixed bug #302: better primitive-type selection for intersection
1239 * fixed bug #308: non-graphic characters now all have names, as
1240 required. (reported by Bruno Haible)
1241 * bug fix: redefining a class with different superclasses now correctly
1242 removes it from the direct-subclasses of its previous superclasses.
1243 (reported by David Morse)
1244 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1245 new value now works. (reported by Bruno Haible)
1246 * bug fix: correct canonicalization of multiple non-standard slot
1247 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1248 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1249 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1250 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1251 now exists, an signals an error.
1252 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1253 during macroexpansion for non-iteration variables as well. (reported
1254 by Bruno Haible for CMUCL)
1255 * bug fix: Cyclic structures and unprintable objects in compiler
1256 messages no longer cause errors. (reported by Bruno Haible)
1257 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1258 and READ-FROM-STRING all now return a primary value of NIL if
1259 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1260 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1261 (reported by Bruno Haible for CMUCL)
1262 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1263 do not designate a condition. (reported by Bruno Haible for
1265 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1266 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1267 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1268 permitting resolution in favour of any of the conflicting symbols.
1269 (reported by Bruno Haible for CMUCL)
1270 * bug fix: EQUAL compiler optimizations is less aggressive on
1271 strings which can potentially compare true despite having distinct
1272 specialized array element types.
1273 * bug fix: unit enumerations can be defined without dividing by
1274 zero. (thanks to Vincent Arkesteijn)
1275 * FORMAT compile-time argument count checking has been enhanced.
1276 (report from Bruno Haible for CMUCL)
1277 * a partial workaround for the bug 262: the compiler does not try to
1278 inline-expand a local function doing RETURN-FROM from a deleted
1279 BLOCK. (thanks to Peter Denno for the bug report and to David
1280 Wragg for the simple test case)
1281 * fixed some bugs revealed by Paul Dietz' test suite:
1282 ** RENAME-PACKAGE allows all package designators as new package
1284 ** constraint propagation and lambda variable substitution are
1285 more cautious in dealing with partially deleted code.
1286 ** compiler sometimes lost reoptimization passes.
1287 ** CERROR, given a condition as condition designator, passes its
1288 remaining arguments to the continue format control without
1290 ** Case-altering FORMAT directives work correctly on non-ASCII
1292 ** The REAL type specifier handles bounds outside the floating
1293 point ranges without signalling FLOATING-POINT-OVERFLOW.
1294 ** Functions with IR1-transformations can create intercomponent
1295 references to global functions.
1296 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1298 ** FORMAT ~R treats a nil value for its first parameter correctly.
1300 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1301 * enhancement: saving cores with foreign code loaded is now
1302 supported on x86/NetBSD and sparc/Linux in addition to the previously
1303 supported platforms.
1304 * bug fix: on some platforms repeated installations caused multiple
1305 copies of HTML documentation to be installed -- should not happen
1306 any more. (reported by Stefan Scholl)
1307 * bug fix: parsing self-recursive alien record types multiple times
1308 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1309 original patch by Helmut Eller for CMUCL)
1310 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1311 (thanks to Richard Kreuter)
1312 * bug fix: defining classes whose accessors are methods on existing
1313 generic functions in other (locked) packages no longer signals
1314 bogus package lock violations. (reported by François-René Rideau)
1315 * bug fix: special variables as DEFMETHOD parameters no longer have
1316 associated bogus type declarations. (reported by David Wragg and
1318 * bug fix: read-write consistency on streams of element-type
1319 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1320 * bug fix: redefiniton of the only method of a generic function with
1321 no DEFGENERIC no longer emits a full WARNING. In addition,
1322 redefinition of generic functions with no DEFGENERIC to an
1323 incompatible lambda list now signals an error. (thanks to Zach
1325 * bug fix: DEFGENERIC now works even when there's a function of the
1326 same name in an enclosing lexical environment. (thanks to Zach
1328 * fixed compiler failure, caused by instrumenting code during
1329 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1330 * optimization: added loop analysis and improved register allocation
1331 * fixed some bugs revealed by Paul Dietz' test suite:
1332 ** POSITION on displaced vectors with non-zero displacement
1333 returns the right answer.
1334 ** (SIMPLE-STRING) is a valid type specifier for sequence
1336 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1338 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1339 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1341 ** The justification version of the FORMAT ~< directive treats
1342 non-zero minpad parameter correctly.
1344 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1345 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1346 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1347 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1348 the supported interface.
1349 * new feature: Single-stepping of code compiled with DEBUG 2 or
1350 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1351 * new feature: saving cores with foreign code loaded is now
1352 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1353 Timothy Moore's work for CMUCL)
1354 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1355 parameters to * instead of NIL if no initform is supplied.
1356 (reported by Johan Bockgaard)
1357 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1358 correctly when FUNCALL appears as the car of the form. Note:
1359 despite this FUNCALL forms are not currently subject to
1360 compiler-macro expansion. (port of Raymond Toy's fix for the
1361 same from CMUCL, reported by Johan Bockgaard)
1362 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1363 (thanks for Teemu Kalvas)
1364 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1365 case instead of terminating the iteration (thanks for Julian
1366 Squires, Sean Champ and Raymond Toy)
1367 * bug fix: incorrect expansion of defgeneric that caused a style
1368 warning. (thanks for Zach Beane)
1369 * bug fix: slot accessor effective method computation works properly
1370 for classes with multiple non-standard applicable methods on
1371 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1372 * on x86 compiler supports stack allocation of results of LIST and
1373 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1374 CMUCL implementation by Gerd Moellmann)
1375 * fixed some bugs revealed by Paul Dietz' test suite:
1376 ** FORMAT strings with both the ~<~:;~> form of the justification
1377 directive and pretty-printing directives cause an error.
1379 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1380 * incompatible change: the internal functions
1381 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1382 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1383 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1384 instead of the old functions.
1385 * new feature: on platforms where "dladdr" is available foreign
1386 function names now appear in backtraces. (based on Helmut Eller's
1388 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1389 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1391 * documentation: ASDF manual is now distributed with SBCL.
1392 * bug fix: SBCL can now load its contributed modules with REQUIRE
1393 even if the system-provided entries have been removed from
1394 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1396 * bug fix: backtraces involving undefined functions or assembly
1397 routines are more informative. (thanks to Brian Downing)
1398 * bug fix: mutually referent alien structures now work correctly.
1399 (reported by Rick Taube)
1400 * bug fix: structures defined by WITH-ALIEN can be referred to
1401 within other definitions in the same WITH-ALIEN.
1402 * bug fix: division operators (MOD, TRUNCATE and the like) with
1403 constant zero divisors and integer dividends no longer generate
1405 * bug fix: provide default methods for INPUT-STREAM-P and
1406 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1407 * bug fix: improve the detection and resolution of MOP metacycles
1408 (where effective method computation depends on the generic function
1409 having its effective method computed). (reported by Bruno Haible)
1410 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1411 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1412 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1413 * bug fix: pathnames with relative directory components can be
1414 represented relative to default pathnames.
1415 * optimization: in taking the GCD of bignums, reduce the two bignums
1416 to approximately the same size (using Euclid's algorithm) before
1417 applying the more sophisticated binary GCD. (thanks to Juho
1419 * optimization: COUNT on bitvectors now operates word-at-a-time.
1420 * optimization: ASH with a positive, but not necessarily constant,
1421 (leftwards) shift, when in a modular context, is compiled to a
1423 * fixed some bugs revealed by Paul Dietz' test suite:
1424 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1425 if the corresponding argument is NIL.
1426 ** FORMAT directives accepting a minpad parameter treat negative
1428 ** literal commas as character parameters to FORMAT directives are
1430 ** literal spaces directly after ~<Newline> directives within a
1431 format-logical-block (~:< ~@:>) do not induce :FILL-style
1432 conditional newlines.
1433 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1434 as specified: it no longer includes conditional newlines).
1435 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1437 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1438 :PER-LINE-PREFIX argument does not evaluate to a string.
1440 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1441 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1442 the manual for details; for now, package locks can be disabled by
1443 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1444 if no major problems are found then it is likely that they will be
1445 compiled in unconditionally.
1446 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1447 unsupported operators on all platforms. To load a shared library
1448 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1449 object file, link it into a shared library outside of SBCL and
1450 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1452 * bug workaround: ROOM T and the GC were not working together
1453 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1454 reporting facility aren't preserved. That reporting has been
1455 disabled, so now until and unless someone figures out how to make it
1456 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1457 * minor incompatible change: as threatened around sbcl-0.8.0, the
1458 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1459 The symbols are also exported from SB-PCL for backwards
1460 compatibility, but more so than before SB-PCL should be treated as
1461 an implementation-internal package.
1462 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1464 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1465 It is known as of this release not to work on the Alpha, however.
1466 * fixed bug #167: errors signalled due to illegal syntax in method
1467 bodies are now more legible.
1468 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1469 designators and can hence be used with TYPEP.
1470 * fixed bug #333: CHECK-TYPE now ensures that the type error
1471 signalled, if any, has the right object to be accessed by
1472 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1473 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1474 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1476 * fixed bug #334: programmatic addition of slots using specialized
1477 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1478 and :ALLOCATION :CLASS slots.
1479 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1480 float exponents. (rereported by Peter Seibel)
1481 * fixed a bug: #\Space (and other whitespace characters) are no
1482 longer considered to be macro characters in standard syntax by
1483 GET-MACRO-CHARACTER.
1484 * fixed bug: initialization of condition class metaobjects no longer
1485 causes an instance of the condition to be created. (reported by
1487 * fixed bug: it is now possible to have slots such that
1488 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1489 description is neither :INSTANCE nor :CLASS.
1490 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1491 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1492 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1493 non-standard SLOT-DEFINITION-ALLOCATION values.
1494 * fixed bugs: various race conditions handling exiting threads.
1495 CL-PPCRE's thread torture test now passes.
1496 * fixed arguably-a-bug: GC time proportional to number of threads in
1497 system even when most of them are idle
1498 * optimization: improved performance of BIT and SBIT on bit-vectors.
1499 * .fasl file incompatibility: The fasl file version number has
1500 been incremented because of changes associated with package locks.
1502 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1503 * minor incompatible change: the system no longer provides
1504 optimization hints (conditions which are TYPEP
1505 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1506 mode; these hints are still emitted when higher SPEED optimization
1508 * new contrib module: a sampling profiler (profiling by statistical
1509 sampling, rather than by instrumenting functions) is available as
1510 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1511 * the behaviour of the standard function ED is now customizeable by
1512 third parties through a hook variable: see ED's documentation
1513 string for information on the protocol.
1514 * the compiler no longer emits efficiency notes for (FUNCALL X)
1515 when the type of X is uncertain under default optimization
1517 * fixed bug 276: mutating a binding of a specialized parameter to a
1518 method to something that is not TYPEP the specializer is now
1520 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1521 exhibit the required behaviour on the x86 platform. (thanks to
1522 Peter van Eynde, Eric Marsden and Bruno Haible)
1523 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1524 even for difficult arguments. (reported by Peter Graves)
1525 * fixed bug 141a: the backquote printer now descends quoted
1527 * fixed another bug in backquote printing: no more destructive
1528 modification of the form's list structure. (reported by Brian
1530 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1531 move between its address being taken and the call to
1532 interrupt_thread, fixing a crashing race condition.
1533 * the SB-POSIX contrib implementation has been adjusted so that it
1534 no longer exhibits ridiculously poor performance when constructing
1535 instances corresponding to C structs.
1537 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1538 * minor incompatible change: the sb-grovel contrib now treats C
1539 structures as alien (in the sense of SB-ALIEN) objects rather than
1540 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1541 has implications for memory management of client code
1542 (sb-grovel-returned objects must now be manually managed) and for
1543 type safety (alien objects now have full types).
1544 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1545 used to control emission of compiler diagnostics, rather than the
1546 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1547 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1548 quality should be considered deprecated.
1549 * install.sh now installs the user manual as well
1550 * (not quite a new documentable feature, but worth considering in
1551 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1552 beginnings of a semantically meaningful condition hierarchy is
1553 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1555 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1556 designator as the defaults argument.
1557 * fixed bug: Displaced arrays whose displaced-to array has become
1558 too small now cause ARRAY-DIMENSION to signal an error, providing
1559 for safer bounds-checking. (reported by Bruno Haible)
1560 * fixed bug: DEFCLASS slot definitions with identical :READER and
1561 :WRITER names now signal a reasonable error. (reported by Thomas
1563 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1565 * fixed bug: Invalid slot specification errors now print correctly.
1566 (thanks to Zach Beane)
1567 * fixed bug 320: Shared to local slot value transfers in class
1568 redefinitions now happen corrently from superclasses as
1569 well. (reported by Bruno Haible)
1570 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1572 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1573 improper lists correctly. (reported by Bruno Haible)
1574 * fixed bug 313: source-transform for <fun-name> was erroneously
1575 applied to a call of a value of a variable with name <fun-name>.
1576 (reported by Antonio Menezes Leitao)
1577 * fixed bug 307: The obsolete instance protocol ensures that
1578 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1579 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1580 remove dead unknown-values globs from the middle of the stack.
1581 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1582 easier to resume long computations after using *BREAK-ON-SIGNALS*
1583 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1584 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1585 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1586 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1587 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1588 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1590 * fixed a (fairly theoretical) bug in string printing: if
1591 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1592 does not have array-element-type equal to the most general string
1594 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1595 function name. (thanks to Nikodemus Siivola)
1596 * fixed bug: compile-time format string checker failed on
1597 non-closed ~{. (reported by Thomas F Burdick)
1598 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1599 don't warn on legal loop code involving more than one aggregate
1600 boolean. (thanks to Nikodemus Siivola)
1601 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1602 in keyword position to MAKE-PACKAGE was not being reported as
1604 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1605 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1607 * fixed bugs 280 and 312: the checking for multiple definitions in a
1608 file is less likely to become confused by uses of inline
1610 * fixed bug: the #S reader macro performs the keyword coercion
1611 specified for slot names. (reported by Kalle Niemitalo)
1612 * fixed bug: lambda lists may contain symbols whose names start with
1613 & but are not lambda-list-keywords; their occurrence triggers a
1615 * fixed bug 321: define-method-combination argument lambda lists do
1616 not cause invalid code to be generated when &optional or &aux
1617 variables have default values. (reported by Bruno Haible)
1618 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1619 have CLOS classes; this appears also to have fixed reported
1620 instability in stack exhaustion detection.
1621 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1622 MUFFLE-WARNING when no associated restart is present is now
1624 * optimization: rearranged the expansion of various defining macros
1625 so that each expands into only one top-level form in a
1626 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1628 * optimization: used a previously little-used slot in symbols to
1629 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1631 * fixed some bugs revealed by Paul Dietz' test suite:
1632 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1634 ** Fixed an optimization bug related to inheritance of initforms
1635 from local to shared slots.
1636 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1637 ** CAST optimizer forgot to flush argument derived type.
1638 ** print/read consistency on floats is now orders of magnitude
1639 more likely. (thanks also to Bruno Haible for a similar report
1641 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1642 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1643 for objects of type REAL. Make it so.
1644 ** SXHASH is specified (infelicitously) to respect similarity,
1645 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1646 it so. (thanks to Markus Ziegler)
1647 ** on the Alpha, the compiler succeeds in compiling functions
1648 returning a known number of arguments greater than 63.
1649 ** fixed handling of invalid NIL arguments in keyword position
1651 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1653 ** provided a readably-printable representation for RANDOM-STATE
1655 ** ensured that pathnames, where they have a namestring, always
1656 print using #P"..." syntax.
1658 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1659 * Support for the forthcoming 2.0 version of the NetBSD kernel
1660 running on x86 hardware has been added. (thanks to Perry
1661 E. Metzger most immediately, and others for their past work)
1662 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1663 current release version; SBCL's OpenBSD support had been broken
1664 since about the time of OpenBSD's switch to ELF binary format.)
1665 * [placeholder for DX summary]
1666 ** user code with &REST lists declared dynamic-extent, under high
1667 speed or space and low safety and debug optimization policy.
1668 * The manual has been converted to Texinfo format and the debugger
1669 chapter from the cmucl manual has been added.
1670 * A facility has been added to extract documentation strings from
1671 sbcl and store them as Texinfo-formatted snippets for inclusion in
1672 the manual (via Texinfo's @include directive)
1673 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1674 the test case to Dave Roberts)
1675 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1676 pointers. (reported by Sean Ross)
1677 * bug fix: PROFILE output is printed nicely even for large numerical
1678 values. (thanks to Zach Beane)
1679 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1680 greater than 32 handle EOF correctly.
1681 * bug fix: on X86 an immediate argument of the IMUL instruction is
1682 correctly printed in disassembly. (thanks to Lutz Euler)
1683 * bug fix: class slots in redefined classes preserve their old
1684 values. (thanks to Bruno Haible and Nikodemus Siivola)
1685 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1686 succeeds. (reported by Marco Baringer)
1687 * fixed some bugs revealed by Paul Dietz' test suite:
1688 ** READ-SEQUENCE now works on ECHO-STREAMs.
1689 ** RATIONALIZE works more according to its specification. (thanks
1692 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1693 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1694 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1695 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1696 should matter to you only if you rebind the printer control
1697 variables and then find you want different bindings in the
1698 debugger than in the ordinary execution of your program.)
1699 * The runtime build system has been tweaked to support building
1700 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1701 assembler and linker. (thanks to Nikodemus Siivola)
1702 * Unbound, undefined, undeclared variables now trigger full
1703 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1704 more often programmer error than deliberate exploitation of undefined
1706 * optimization: the hash algorithm for strings has changed to one
1707 that is less vulnerable to spurious collisions. (thanks to Juho
1709 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1710 needless bounds checking. (thanks to Juho Snellman)
1711 * optimization: implemented multiplication as a modular
1712 (UNSIGNED-BYTE 32) operation on the PPC backend.
1713 * fixed some bugs revealed by Paul Dietz' test suite:
1714 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1716 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1717 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1718 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1720 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1722 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1724 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1725 their output stream on EOF from read.
1726 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1727 have been read to end-of-file.
1728 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1730 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1731 description of determination of which consecutive characters
1733 ** Printing the "Space" character with escaping on now yields "#\\ ",
1734 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1735 ** Reading floating-point numbers with *READ-BASE* set to a number
1736 less than 10 works correctly.
1737 ** Reading floating-point numbers with *READ-BASE* set to a number
1738 more than 10 works correctly.
1739 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1740 the readtable currently in effect.
1742 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1743 * minor incompatible change: parsing of namestrings on a physical
1744 (Unix) host has changed; numbers after the final #\. in a
1745 namestring are no longer interpreted as a version field. This is
1746 intented to be largely invisible to the user, except that the
1747 meaning of the namestring "*.*.*" has changed: it now refers to a
1748 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1749 should usually be replaced by
1750 (make-pathname :name :wild :type :wild :version :wild)
1751 with the added benefit that this is more likely to be portable.
1752 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1753 option to OPEN now signals an error if the file being opened
1754 exists; this may have an impact on existing code.
1755 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1756 signals received, on the PowerPC platforms (both Linux and
1757 Darwin). (thanks to Pierre Mai for pointing out the location of
1759 * several fixes on OS X: The system now builds and runs cleanly on
1760 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1761 (thanks to Brian Mastenbrook)
1762 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1763 second-resolution: integer multiples of 1/3600 between -24 and 24.
1764 (thanks to Vincent Arkesteijn)
1765 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1766 type when called with 1 argument; PEEK-CHAR checked type of
1767 PEEK-TYPE only after having read first character from a
1768 stream. (reported by Peter Graves)
1769 * bug fix: the garbage collector now has much better locality
1770 behaviour, and in particular no longer treats all memory as being
1771 exhausted when many small objects point to each other in a deeply
1773 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1774 recognized as being TYPEP their class.
1775 * bug fix: the PUSHNEW documentation string has been corrected.
1776 (thanks to Vincent Arkesteijn)
1777 * bug fix: defaulting of the value for the last of an atypically
1778 large number of multiple values being bound was not being
1779 performed correctly on the Alpha or PPC platforms
1780 * optimization: implemented multiplication as a modular
1781 (UNSIGNED-BYTE 32) operation on the x86 backend.
1782 * optimization: SEARCH on simple-base-strings can now be open-coded.
1783 (see also contrib/compiler-extras.lisp for inspiration for
1784 teaching the compiler about the Boyer-Moore algorithm).
1785 * value, returned by MAX (and MIN) called with several EQUALP, but
1786 not EQL, arguments now does not depend on compiler settings.
1787 * fixed some bugs revealed by Paul Dietz' test suite:
1788 ** in stack analysis liveness information is propagated from
1789 non-local entry points.
1790 ** pathwise CAST removing failed when the CAST node did not start
1792 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1793 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1795 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1796 host is already defined.
1797 ** RENAME-FILE works on streams instead of signalling an internal
1799 ** PEEK-CHAR uses the current readtable when determining whether
1800 or not a character is whitespace.
1801 ** MERGE-PATHNAMES handles the case when the pathname does not
1802 specify a name while the default-pathname specifies a version.
1803 ** Pathnames now stand a better chance of respecting print/read
1805 ** Attempting to use standardized file system operators with a
1806 pathname with invalid :DIRECTORY components signals a
1808 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1810 ** DIRECTORY on logical pathnames is more correct.
1811 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1812 signal a TYPE-ERROR if their argument is not a stream.
1813 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1814 designator argument does not designate a stream.
1815 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1816 examining the synonym.
1817 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1819 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1820 element-type larger than ([UN]SIGNED-BYTE 32).
1822 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1823 * When built with the :SB-FUTEX feature, threaded builds now take
1824 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1825 for faster/more reliable mutex and condition variable support.
1826 * Incompatible change (but one you probably shouldn't have been using
1827 anyway): the interface and code for arbitrating between multiple
1828 threads in the same user session has been redesigned.
1829 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1830 error when called without an explicit environment argument.
1831 (thanks to Nikodemus Siivola)
1832 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1833 confused about their position. (thanks to Adam Warner and Gerd
1835 * bug fix: Pretty printing backquoted forms with unquotations in the
1836 argument list position of various code constructs such as LAMBDA
1837 now works correctly. (reported by Paul Dietz)
1838 * bug fix: Pretty printing unquotations no longer loses all
1839 stream position information.
1840 * optimization: performance of string output streams is now less
1841 poor for multiple small sequence writes.
1842 * optimization: performance of CSUBTYPEP in the presence of complex
1843 expressions involving CONS and NOT many times has been improved.
1844 (reported by Paul Dietz)
1845 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1847 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1848 streams for the REPL, for the debugger, and so on. (thanks to
1850 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1851 * fixed some bugs revealed by Paul Dietz' test suite:
1852 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1853 is ignored if it is too small, rather than propagating through
1854 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1855 ** extremely complex negations of CONS types were not being
1856 sufficiently canonicalized, leading to inconsistencies in
1858 ** VALUES tranformer lost derived type.
1860 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1861 * fixed a bootstrapping bug: the build process no longer assumes
1862 that the various BOOLE-related constants have the same value in
1863 host and target lisps. (noted by Paul Dietz' test suite on an
1864 SBCL binary built from CLISP)
1865 * The system can now be dynamically linked on the MIPS platform,
1866 which enables dynamic loading of foreign code from Lisp. (thanks
1867 to Ralf Baechle for discussions on the MIPS ABI)
1868 * The system now records debugging information for its own source
1869 files in a filesystem-position-independent manner, relative to
1870 the "SYS" logical host.
1871 * fixed a compiler bug: MV-LET convertion did not check references
1872 to the "max args" entry point. (reported by Brian Downing)
1873 * tweaked disassembly notes to be less confident about proclaiming
1874 some instruction as an LRA. (thanks to Brian Downing)
1875 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1876 now each have their own history, command character, and other
1877 characteristics. (thanks to David Lichteblau)
1878 * fixed some bugs revealed by Paul Dietz' test suite:
1879 ** compiler failure in compiling LOGAND expressions including a
1881 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1882 shift greater than 32.
1883 ** FLUSH-DEST did not mark blocks for type check regeneration.
1884 ** HANDLER-CASE failed to accept declarations in handler clauses
1885 in some circumstances.
1887 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1888 * New code in contrib/sb-introspect (still probably not entirely
1889 stable yet) provides some support for smart Lisp development
1890 environments like SLIME.
1891 * The conditions signalled for errors occurring when loading .fasl
1892 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1893 in a way which should help ASDF recover gracefully.
1894 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1895 stuff has been cleaned up. If you code contrib/ stuff, this might
1896 affect you, and you can look at contrib/README, contrib/STANDARDS,
1897 and/or the 0.8.4.27 diff to check.
1898 * In full calls the compiler now does not generate checks for declared
1899 argument types for all arguments.
1900 * various threading fixes
1901 ** and some experimental patches which didn't make it into
1902 the main tree for this release, but which are shipped in
1903 contrib/experimental-thread.patch as a possible fix for some
1904 failures (deadlock, spinning...) in GC-intensive multithreaded
1906 * fixed PPC build problem (source code incompatibility of different
1907 library versions): added offsetof() hackery which attempts to divine
1908 where glibc maintainers put uc_mcontext today
1909 * fixed bug 282: compiler does not trust type assertions while passing
1910 arguments to a full call.
1911 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1912 &OPTIONAL VALUES type parameter.
1913 * fix bug 214: algorithm for noting rejected templates is now more
1914 similar to that of template seletion. (also reported by rydis on
1916 * fixed bug 141b: printing backquoted information readably and prettily
1917 inserts a space where necessary.
1918 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1919 (CHARACTER 10) are now reported as errors, rather than propagated
1920 as unknown types. (reported by piso on #lisp)
1921 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1922 with values NIL and :ERROR. (thanks to Milan Zamazal)
1923 * fixed bug 191c: CLOS now does proper keyword argument checking as
1924 described in CLHS 7.6.5 and 7.6.5.1.
1925 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1926 longer raise an error; further, using a list as a for-as-arithmetic
1927 counter now raises a meaningful error.
1928 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1929 now understood by sequence creation functions such as MAKE-SEQUENCE
1931 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1932 a STRING-INPUT-STREAM.
1933 * compiler enhancement: SIGNUM is now better able to derive the type
1935 * type declarations inside WITH-SLOTS are checked. (reported by
1937 * fixed some bugs revealed by Paul Dietz' test suite:
1938 ** incorrect optimization of TRUNCATE for a positive first
1939 argument and negative second.
1940 ** compiler failure in let-convertion during flushing dead code.
1941 ** compiler failure while deriving type of TRUNCATE on an
1942 interval, containing 0.
1943 ** ASH of a negative bignum by a negative bignum count now returns
1945 ** intersection of CONS types now canonicalizes properly, fixing
1946 inconsistencies in SUBTYPEP.
1948 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1949 * incompatible change: The --disable-debugger command line
1950 option now clobbers the debugger at a more fundamental
1951 level, by redefining #'INVOKE-DEBUGGER instead of by
1952 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1953 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1954 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1955 enter the debugger REPL and then suffer recursive errors
1956 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1957 while under the new system, BREAK in --disable-debugger mode
1958 terminates the system just as an unhandled error would.
1959 * fixed compiler performance when processing loops with a step >1;
1960 * bug fix: DOCUMENTATION now retrieves generic function
1961 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1962 support has been systematized, and now supports the methods
1963 specified by ANSI, along with a default method and a method for
1964 slot documentation. (reported by Nathan Froyd)
1965 * bug fix: effective methods associated with a generic function are
1966 no longer cached over a change of that generic function's method
1967 combination. (reported by Andreas Fuchs)
1968 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1969 names a non-existent directory. (thanks to Andreas Fuchs)
1970 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1971 the PPC platform, lead to stack corruption; this has been fixed.
1972 (reported by Rainer Joswig)
1973 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1974 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1975 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1976 arithmetic in general, is now much more reliable on the PPC
1978 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1979 platform now returns the right answer.
1980 * optimization: restored some effective method precomputation in
1981 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1982 precomputation is now tunable.
1983 * optimization: compiler-internal data structure use has been
1984 reviewed, and changes have been made that should improve the
1985 performance of the compiler by about 20%.
1986 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1987 simple-base-strings and simple-bit-vectors is improved.
1988 * optimization: the optimization of 32-bit logical and arithmetic
1989 functions introduced in version 0.8.3 on the x86 has been
1990 implemented on the mips, ppc and sparc platforms; an
1991 implementation of the same facility, but for 64-bit arithmetic,
1992 has been added for the alpha.
1993 * microoptimization: the compiler is better able to make use of the
1994 x86 LEA instruction for multiplication by constants.
1995 * bug fix: in some situations compiler did not report usage of
1996 generic arithmetic in (SPEED 3) policy.
1997 * bug 145b fix: compiler used wrong type specifier while converting
1998 MEMBER-types to numeric.
1999 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2001 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2002 index argument now works properly on the Alpha platform.
2003 * bug fix: floating point exception treatment on the Alpha platform
2005 * bug fix: FILE-POSITION works much better on string input and
2006 output streams. (thanks to Nikodemus Siivola)
2007 * bug fix: many threading/garbage collection symptoms sorted.
2008 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2009 might be pseudo-atomic.
2010 * internal change: Stopping for GC is now done with signals not
2011 ptrace. GC is now done in whichever thread wanted it, instead of
2013 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2015 * bug fix: result form in DO is not contained in the implicit
2017 * incompatible change: ICR structure is changed; the value part of
2018 CONTINUATION is now called LVAR; corresponding functions are
2019 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2020 * added type deriver for ISQRT (thanks to Robert E. Brown).
2021 * bug fix: better support for loading from the command line when an
2022 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2024 * fixed some bugs revealed by Paul Dietz' test suite:
2025 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2026 ** ROUND and FROUND now give the right answer when given very
2027 small float arguments.
2028 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2030 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2031 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2032 ** LCM should return a non-negative integer.
2033 ** PARSE-INTEGER returned the index of a terminator instead of the
2034 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2035 ** PARSE-INTEGER returned an incorrect index being applied to a
2037 ** LCM with two arguments of 0 returns 0 rather than signalling
2039 ** unsigned addition of a 32-bit constant with the high bit set no
2040 longer causes an internal compiler error on the x86.
2041 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2042 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2043 with negative last argument.
2044 ** byte specifiers with zero size and position no longer cause
2045 an error during type derivation.
2046 ** bignum multiplication on the Alpha platform now returns the
2048 * porting: The system now builds on SuSE AMD64, although it still
2049 generates a 32-bit binary.
2050 * .fasl file incompatibility: The fasl file version number has
2051 been incremented (because of the changes to internal compiler
2052 data structures referred to above).
2054 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2055 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2056 more accurately, on the Darwin kernel running on PowerPC hardware.
2057 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2058 * Compiler code deletion notes now signal a condition of type
2059 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2060 an associated MUFFLE-WARNING restart.
2061 * The compiler now performs limited argument count validation of
2062 constant format strings in FORMAT, and where appropriate in ERROR,
2063 CERROR and WARN. (thanks to Gerd Moellmann)
2064 * New ASDF-INSTALL contrib can be used for automatic download and
2065 installation of third-party Lisp code from CCLAN or other sites
2067 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2068 which forces another thread to execute a function supplied by the
2070 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2071 now accept and act upon their :ELEMENT-TYPE keyword argument.
2072 (reported by Martin Atzmueller, Edi Weitz)
2073 * bug fix: FILE-POSITION now accepts position designators up to
2074 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2075 is the greater. (thanks to Patrik Nordebo)
2076 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2077 Kalle Olavi Niemitalo)
2078 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2079 printable. (reported by Eric Marsden)
2080 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2081 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2082 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2083 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2084 (thanks to Stig Sandoe)
2085 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2086 UNREAD-CHAR now correctly includes the unread character in the
2087 target sequence. (thanks to Gerd Moellmann)
2088 * bug fix in threaded builds: the system can now be suspended and
2089 resumed by shell job control with minimal disruption.
2090 * bug fixes in times and timezones >2038AD
2091 * better handling of "where is GNU make?" problem in build scripts
2092 (thanks to Nikodemus Siivola)
2093 * new optimization: inside a named function any reference to a
2094 function with the same name is considered to be a self-reference;
2095 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2096 optimization quality.
2097 * new optimization on x86: logical functions and + now have
2098 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2099 used when the result is truncated to 32 bits.
2100 * VALUES declaration is partially enabled.
2101 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2102 * fixed some bugs revealed by Paul Dietz' test suite:
2103 ** The system now obeys the constraint imposed by
2104 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2105 types form a lattice under type intersection.
2106 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2107 ** ASSOC now ignores NIL elements in an alist.
2108 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2109 and (1+ MOST-POSITIVE-FIXNUM) answers.
2110 ** The addition of a method with invalid qualifiers to a generic
2111 function does not cause an error to be signalled immediately;
2112 a warning is signalled, and the error is generated only on
2113 calling the generic function.
2114 * changed .fasl file version number, in part to add type codes for
2115 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2116 obscure ANSI requirements
2118 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2119 * fixed bug 148: failure to inline-expand a local function left
2120 garbage, confusing the compiler.
2121 * fixed bugs 3cd: structure slot readers perform type check if the
2122 slot can have an invalid value (i.e. it is either not initialized
2123 or can be written with a less specific slot writer).
2124 * bug fix: the compiler now traps array references to elements off
2125 the end of an array; previously, the bounds checking in some
2126 circumstances could go off-by-one.
2127 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2129 * type declarations for array element types now obey the description
2130 on the CLHS page "Declaration TYPE", as per discussions on
2131 sbcl-help around 2003-05-08. This means that a declaration
2132 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2133 declaration, all references to BAR will be asserted or assumed
2134 (with THE, so dependent on compiler policy) to involve objects of
2135 type FOO. Note that no such declaration is implied in
2136 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2137 * declared types of functions from the "Conditions"
2138 chapter. (reported by Paul Dietz)
2139 * bug fix: CERROR accepts a function as its first argument.
2140 * bug fix: NTH an NTHCDR accept a bignum as index
2141 arguments. (reported by Adam Warner)
2142 * optimization: character compare routines now optimize comparing
2143 against a constant character. (reported by Gilbert Baumann)
2144 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2145 argument now works properly on the MIPS platform.
2146 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2147 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2148 declared to have a functional type, does not cause an error
2150 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2151 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2152 * fixed bug: if last continuation of a deleted block has a
2153 destination, this destination should be deleted too. (reported by
2155 * fixed a bug in the bootstrap process: the host compiler's values
2156 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2157 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2158 test case from Patrik Nordebo)
2159 * improved the ability of the disassembler on the PPC platform to
2160 provide helpful disassembly notes.
2161 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2162 the class in more cases than previously.
2163 * bug fix: FILE-POSITION now understands :START and :END for
2164 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2165 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2166 8) data. (thanks to David Lichteblau)
2167 * bug fix: it is possible to add a method to a generic function
2168 without lambda list.
2169 * bug fix: reader failed to signal END-OF-FILE inside an
2170 object representation. (reported by Nikodemus Siivola)
2171 * fixed some bugs revealed by Paul Dietz' test suite:
2172 ** LAST and [N]BUTLAST should accept a bignum.
2173 ** condition slot accessors are methods.
2174 ** (VECTOR NIL) is a subtype of STRING.
2176 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2177 * minor incompatible change: some nonsensical specialized lambda
2178 lists (used in DEFMETHOD) which were previously ignored now signal
2180 * minor incompatible change: the system is now aware of the types of
2181 variables in the COMMON-LISP package, and will signal errors for
2182 most violations of these type constraints (where previously they
2183 were silently accepted).
2184 * minor incompatible change: COMPILE-FILE now uses the freedom
2185 afforded (ANSI 3.2.2.3) to use derived function types for
2186 functions defined in the same file. This also permits the system
2187 to warn on static type mismatches and function
2188 redefinition. (Currently it does not work with high DEBUG level.)
2189 * minor incompatible change: VALUES declaration is disabled.
2190 * When issuing notes, the compiler now signals a condition of type
2191 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2192 restart for use in user handlers. It is expected that the
2193 COMPILER-NOTE condition will eventually become a condition
2194 supertype to a hierarchy of note types, which will then be
2195 handleable in a similar fashion. However, at the moment, no such
2196 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2197 but it's an implementation detail, not a classification for the
2199 * Changes in type checking closed the following bugs:
2200 ** type checking of unused values (192b, 194d, 203);
2201 ** template selection based on unsafe type assertions (192c, 236);
2202 ** type checking in branches (194bc).
2203 * A short form of VALUES type specifier has ANSI meaning (it has
2204 increased the number of situations when SBCL cannot perform type
2206 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2207 keywords or constants is permissible.
2208 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2209 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2210 classes. (thanks to Antonio Martinez)
2211 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2212 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2213 argument to a pretty-print stream. Instead, it leaves any such
2214 operations to DESCRIBE-OBJECT methods.
2215 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2217 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2218 lambda lists. (thanks to David Lichteblau)
2219 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2220 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2221 (thanks to Gerd Moellmann)
2222 * fixed bug 246: increased compilation speed of long
2223 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2225 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2227 * DEFCONSTANT now throws a condition of type
2228 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2229 constant to a non-EQL value; CONTINUE and ABORT restarts
2230 respectively change and preserve the value.
2231 * fixed bug 63: The code walker, part of the implementation of CLOS,
2232 is now better at handling symbol macros.
2233 * bug fix: There is no longer an internal implementation type named
2234 CL:LENGTH. (reported by Raymond Toy)
2235 * bug fix: In macro-like defining macros/special operators the
2236 implicit block does not enclose the lambda list.
2237 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2238 enclosed in parentheses) are not suitable as type specifiers, and
2239 their use properly signals an error now.
2240 * bug fix: An argument count mismatch for a type specifier in code
2241 being compiled no longer causes an unhandled error at compile
2242 time, but signals a compile-time warning.
2243 * fixed simple vector readable printing
2244 * bug fix: DESCRIBE takes more care over whether the class
2245 precedence list slot of a class is bound before accessing it.
2246 (reported by Markus Krummenacker)
2247 * bug fix: FORMATTER can successfully compile pretty-printer format
2248 strings which use variants of the ~* directive inside.
2249 * bug fix: SEARCH now applies its TEST predicate to the elements of
2250 the arguments in the correct order. (thanks to Wolfhard Buss)
2251 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2252 in type inference. (thanks to Robert E. Brown)
2253 * bug fix: Reading in symbols with an explicit package name of ""
2254 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2255 * fixed some bugs revealed by Paul Dietz' test suite:
2256 ** NIL is now allowed as a structure slot name.
2257 ** Arbitrary numbers, not just REALs, are allowed in certain
2258 circumstances in LOOP for-as-arithmetic clauses.
2259 ** Multiple class redefinitions before slot access no longer
2260 causes a type error.
2261 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2262 association between the name and a class.
2263 ** Generic functions with non-standard method-combination and over
2264 five methods all of which return constants no longer return NIL
2265 after the first few invocations. (thanks to Gerd Moellmann)
2266 ** CALL-NEXT-METHOD with no arguments now passes the original
2267 values of the arguments, even in the presence of assignment.
2268 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2270 ** DISASSEMBLE works with closures and funcallable instances.
2271 ** ADD-METHOD now returns the generic function, not the new method.
2272 ** FIND-METHOD signals an error if the lengths of the specializers
2273 is incompatible with the generic function, even if the ERRORP
2275 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2276 which its argument is a member.
2277 ** DEFCLASS only redefines the class named by its class-name
2278 argument if that name is the proper name of the class;
2279 otherwise, it creates a new class.
2280 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2281 of the UNBOUND-SLOT condition to the name of the slot.
2282 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2283 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2284 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2285 SLOT-MAKUNBOUND in the specified fashion.
2287 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2288 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2289 cross-compilation host. As a consequence, we can now bootstrap our
2290 way up to SBCL starting with a bare gcc toolchain and human-readable
2291 source code (first the source to CLISP, then the source to SBCL).
2292 * A contributed module containing a partial implementation of the
2293 simple-streams interface has been included. (thanks to Rudi
2295 * A contributed module implementing the RFC1321 Message Digest
2296 Algorithm, known as MD5, has been included.
2297 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2298 no longer has any effect, as the code controlled by this feature
2299 has been deleted. (As far as we know, no-one has ever built using
2300 this feature, and its semantics were confused in any case).
2301 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2302 work on conditions (as required by the ANSI specification),
2303 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2304 expected behaviour on conditions. Users should note, however,
2305 that such behaviour is not required by the ANSI specification,
2306 and so use of this behaviour may render their code unportable.
2307 * fixed some bugs revealed by Paul Dietz' test suite:
2308 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2310 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2311 ** SXHASH on condition objects no longer returns NIL.
2312 ** :ALLOCATION :CLASS slots are better treated; their values are
2313 updated on class redefinition, and initforms inherited from
2314 superclasses are applied.
2315 ** REMOVE-METHOD returns its generic function argument even when
2316 no method was removed.
2317 ** SHARED-INITIALIZE now initializes the values of the requested
2318 slots, including those with :ALLOCATION :CLASS.
2319 ** ALLOCATE-INSTANCE now works on structure classes defined via
2320 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2322 ** SLOT-EXISTS-P now works on conditions, as well as structures
2324 ** MAKE-LOAD-FORM now has the required methods on
2325 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2326 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2327 keyword, and now implements the SLOT-NAMES argument.
2328 ** methods with &OPTIONAL arguments no longer allow too many
2329 arguments to be passed in the call without error.
2330 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2331 option is consistent with the required arguments of the generic
2332 function lambda list.
2333 * bug fix: REQUIRE accepts a string designator. (Thanks to
2335 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2336 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2337 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2338 * bug fix: The compiler now checks for duplicated variables in macro
2340 * bug fix: SETQ on globals returns the correct value.
2341 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2342 not a COMPILER-ERROR (followed by some other strange error on
2343 choosing the CONTINUE restart).
2344 * bug fix: make.sh and friends are now more consistent in the way that
2345 they look for GNU "make".
2347 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2348 * experimental native threads support (on x86 Linux >=2.4 only).
2349 This is not compiled in by default: you need to add :SB-THREAD to
2350 the target features. See the "Beyond ANSI" chapter of the manual
2352 * fix for longstanding nonANSIism: The old distinction between
2353 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2354 The return value from CL:FIND-CLASS is now a CLOS class, and
2355 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2356 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2357 * An interface to the MetaObject Protocol, as described in Kiczales,
2358 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2359 MIT Press, 1991, is available from the SB-MOP package.
2360 * incompatible change: the SB-PCL package should now be considered
2361 a private implementation detail, and no longer a semi-private MOP
2363 * minor incompatible change: due to rearrangement for threads, the
2364 control stack and binding stack are now allocated at arbitrary
2365 addresses instead of being hardcoded per-port. Users affected by
2366 this probably have to be doing advanced things with shared
2367 libraries, and will know who they are.
2368 * minor incompatible change: Previously, all --eval forms used were
2369 processed with READ before any of them were processed with EVAL.
2370 Now each --eval form is processed with both READ and EVAL before
2371 the next --eval form is processed. (Thus package operations like
2372 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2373 work as the user might reasonably expect.)
2374 * minor incompatible change: *STANDARD-INPUT* is now only an
2375 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2377 * minor incompatible change: Y-OR-N-P is now character-oriented,
2378 not line oriented. Also, YES-OR-NO-P now works without errors.
2379 (thanks to Antonio Martinez)
2380 * sb-aclrepl module improvements: an integrated inspector, added
2381 repl features, and a bug fix to :trace command.
2382 * Known functions, which cannot be open coded by the backend, are
2383 considered to be able to check types of their arguments. (fixing
2384 a bug report by Nathan J. Froyd)
2385 * fixed a bug in computing method discriminating functions: It is
2386 now possible to define methods specialized on classes which have
2387 forward-referenced superclasses. (thanks to Gerd Moellmann)
2388 * fixed evaluation order in optional entries (reported by Gilbert
2390 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2391 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2392 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2393 required-by-AMOP NAME argument, as well as CLASS and
2394 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2395 * fixed bug 20: DEFMETHOD can define methods using names that are
2396 not the proper names of classes to designate class specializers.
2397 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2398 Unix file descriptors, instead of blowing up. (thanks to Antonio
2400 * Garbage collection refactoring: user-visible change is that a
2401 call to the GC function during WITHOUT-GCING will not do garbage
2402 collection until the end of the WITHOUT-GCING. If you were doing
2403 this you were probably losing anyway.
2404 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2405 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2406 TYPEP the latter but not the former.
2407 * The compiler issues a full WARNING for calls to undefined functions
2408 with names from the CL package.
2409 * MAP-INTO for a vector destination is open coded. (reported by
2410 Brian Downing on c.l.l)
2411 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2412 documentation string.
2413 * fixed some bugs revealed by Paul Dietz' test suite:
2414 ** COPY-ALIST now signals an error if its argument is a dotted
2416 ** Condition slots are now accessed more correctly in the presence
2417 of multiple initargs for a given slot.
2418 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2419 exclude restarts of the same name associated with a different
2421 ** DEFCLASS of forward-referenced classes with another
2422 forward-referenced class in the superclasses list no longer
2424 ** Condition slots are now initialized once each, not multiple
2425 times. (thanks to Gerd Moellmann)
2426 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2428 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2429 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2430 arguments contain duplicated elements.
2431 ** RESTART-CASE understands local macros.
2432 ** RESTART-CASE associates exactly its own restarts with a condition.
2433 ** ENDP in safe mode checks its argument to be of type LIST.
2434 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2435 in question is unbound.
2436 ** Optimized MAKE-INSTANCE functions no longer cause internal
2437 assertion failures in the presence of duplicate initargs.
2438 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2439 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2440 propagated even in the ctor optimized implementation of
2442 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2444 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2445 * a better implementation of SXHASH on (simple) bit vectors,
2446 measured both in execution speed and in distribution of results
2447 over the positive fixnums, has been installed. Likewise, a better
2448 implementation of EQUAL for simple bit vectors is now available.
2449 * fixed CEILING optimization for a divisor of form 2^k.
2450 * fixed bug 240 (emitting extra style warnings "using the lexical
2451 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2452 by Antonio Martinez)
2453 * fixed SXHASH, giving different results for NIL depending on type
2454 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2455 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2456 variable. (found by Rolf Wester)
2457 * SBCL does not ignore type declarations for special
2458 variables. (reported by rif on c.l.l 2003-03-05)
2459 * some bug fixes in contrib/sb-aclrepl/
2460 * fixed some bugs revealed by Paul Dietz' test suite:
2461 ** a bug in the CONS type specifier, whereby the CAR and CDR
2462 types got intertwined, has been fixed;
2463 ** the type system is now able to reason about the interaction
2464 between INTEGER and RATIO types more completely;
2465 ** APPEND, [N]REVERSE and NRECONC check that those their
2466 arguments, which must be proper lists, are really so;
2467 ** An array specialized to be unable to hold elements has been
2468 implemented, as required -- yes, really -- by ANSI;
2469 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2470 on malformed property lists;
2472 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2473 * incompatible packaging change: in line with Unix convention,
2474 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2475 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2476 * REQUIRE and PROVIDE are now optionally capable of doing something
2477 useful. See the documentation string for REQUIRE.
2478 * infrastructure for a managed SBCL contrib system: contributed
2479 modules in this release include:
2480 ** the ASDF system definition facility;
2481 ** an interface to the BSD Sockets API;
2482 ** an ACL-like convenience interface to the repl;
2483 (thanks to Kevin Rosenberg)
2484 ** an implementation of ROTATE-BYTE, with efficient implementation
2486 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2487 gives a closer approximation to the right answer than previously.
2488 (thanks to Raymond Toy)
2489 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2490 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2491 optional environment arguments, as required by ANSI.
2492 * fixed bugs in other functions taking environment objects, allowing
2493 calls with an explicit NIL environment argument to be compiled
2495 * fixed bug 228: primary return values from
2496 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2497 COMPILE or FUNCTION.
2498 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2499 :TYPE will no longer signal errors on innocuous objects.
2500 * fixed bug 231b: SETQ is better at respecting type declarations in
2501 the lexical environment.
2502 * fixed a bug in DEFCLASS: classes named by symbols with no or
2503 unprintable packages can now be defined.
2504 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2505 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2506 * fixed compiler failure related to checking types of functions.
2507 (reported by Robert E. Brown)
2508 * the compiler is now much more consistent in its error-checking
2509 treatment of bounding index arguments to sequence functions: in
2510 (SAFETY 3) code, errors will be signalled in almost all cases if
2511 invalid sequence bounding indices are passed to functions defined
2512 by ANSI to operate on sequences.
2513 * fixed a bug in the build procedure: documentation of SBCL-specific
2514 packages is now preserved and available in the final Lisp image.
2515 * lifted FDEFINITION lookup out of loops in the implementation of
2516 many list operations. (thanks to Robert E. Brown)
2517 * fixed a bug in the reader: the #n# reader macro now works for
2518 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2519 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2520 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2521 E. Brown and Tony Martinez respectively)
2522 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2523 SLOT-BOUNDP: in particular, it is now able to optimize them much
2524 better, and is now not vulnerable to having packages renamed.
2525 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2526 always called when a slot is not present in an instance. (thanks
2528 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2529 longer signal an error on unreading a character following EOF on
2530 the previous constituent stream. (thanks to Tony Martinez)
2531 * fixed some bugs revealed by Paul Dietz' test suite:
2532 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2533 not just nonnegative fixnums;
2534 ** the logical bit-array operators such as BIT-AND now accept an
2535 explicit NIL for their "opt-arg" argument (to indicate a
2536 freshly-consed result bit-array);
2537 ** ELT now signals an error on an invalid sequence index in safe
2539 ** the type system is now cleverer about negations of numeric
2540 types, and consequently understands the BIGNUM and RATIO types
2542 ** the type system is now cleverer about the interaction between
2543 INTEGER and RATIO types: while bugs still remain, many more
2544 cases are accurately computed;
2545 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2546 if it is in the last clause;
2547 ** CONSTANTLY now correctly returns a side-effect-free function in
2549 ** DECLARE is no longer treated as a special-operator; in
2550 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2551 * incremented fasl file version number due to the change in the
2552 DEFSTRUCT-SLOT-DESCRIPTION structure.
2554 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2555 * minor incompatible change: code processed by the "interpreter" or
2556 EVAL now has a compilation optimization policy of (DEBUG 2)
2557 (changed from (DEBUG 1)) to improve debuggability of interactive
2558 development, and to allow the use of the debug RETURN command in
2560 * an experimental implementation of the RETURN command for the
2561 debugger has been included. (thanks to Frederik Kuivinen)
2562 * fixed bug 62: constraints were not propagated into a loop.
2563 * fixed bug in embedded calls of SORT (reported and investigated by
2565 * fixed some bugs revealed by Paul Dietz' test suite:
2566 ** printing and reading of arrays with some dimensions having
2567 length 0 (thanks to Gerd Moellmann);
2568 ** BOA constructor with &AUX argument without a default value does
2569 not cause a type error;
2570 ** CONSTANTP now returns true for all self-evaluating objects.
2572 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2573 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2574 accessors that are related by inheritance, as specified in the
2575 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2576 Valtteri Vuorikoski)
2577 * The compiler is now able to inline functions that were defined in
2578 a complex lexical environment (e.g. inside a MACROLET).
2579 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2581 * Support for the upcoming FreeBSD-5.0 release has been included.
2582 (thanks to Dag-Erling Smorgrav)
2583 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2584 effect when it is not in a toplevel context.
2585 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2586 stand a better chance of being correct. (thanks to Gerd
2588 * fixed bug in COERCE, which now signals an error on coercing a
2589 rational to a bounded real type which excludes the expected
2591 * The compiler is now able to derive types more accurately from the
2592 COERCE and COMPILE functions.
2593 * fixed bug 223: functional binding is considered to be constant
2594 only for symbols in the CL package.
2595 * fixed bug 231: SETQ did not check the type of a variable being set
2596 (reported by Robert E. Brown)
2597 * A new optimization for MAKE-INSTANCE has been included, fixing
2598 various bugs (including relating to :ALLOCATION :CLASS slots and
2599 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2601 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2602 ** As required by ANSI, LOOP now disallows anonymous collection
2603 clauses such as COLLECT I in conjunction with aggregate boolean
2604 clauses such as THEREIS (= I 1);
2605 ** LOOP now signals an error when any variable is reused in the
2606 same loop (including the potentially useful construct analogous
2607 to WITH A = 1 WITH A = (1+ A);
2608 ** IT is only a special loop symbol within the first clause of a
2609 conditional loop clause;
2610 ** LOOP with a typed iteration variable over a hashtable now
2611 signals a type error iff it should.
2612 * fixed some other bugs revealed by Paul Dietz' test suite:
2613 ** FILE-STREAM now names the class previously known as FD-STREAM;
2614 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2615 argument) no longer signals an error;
2616 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2617 of the slot symbol, rather than using the current package
2618 ((:CONC-NAME "") continues to intern the slot's name in the
2620 * incremented fasl file version number, because of the incompatible
2621 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2622 of the new implementation of DEFINE-COMPILER-MACRO.
2624 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2625 * Support for building SBCL for MIPS platforms running in
2626 little-endian mode has now been checked in, and basic
2627 functionality on said platforms verified.
2628 * minor incompatible change: PCL now records the pathname of a file
2629 in which methods and the like are defined, rather than its
2631 * minor incompatible change: TRUENAME now considers the truename of
2632 a file naming a directory to be the pathname with :DIRECTORY
2633 component indicating that directory.
2634 * minor incompatible change: a NAMED clause in the extended form of
2635 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2636 reason for the previous behaviour is unclear.
2637 * more systematization and improvement of CLOS and MOP conformance
2638 in PCL (thanks to Gerd Moellman and Pierre Mai):
2639 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2641 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2642 primary methods with no specializers;
2643 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2645 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2646 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2647 FORWARD-REFERENCED-CLASSes; error reporting on
2648 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2650 ** SXHASH on CLOS instances now uses a slot internal to the
2651 instance to return different numbers on distinct instances,
2652 while preserving the same return value through invocations of
2654 ** DEFMETHOD signals errors when methods with longer incongruent
2655 lambda lists are added to generic functions;
2656 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2657 CLASS, as specified in AMOP;
2658 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2659 on the order returned by the primary method for classes of
2660 class STANDARD-CLASS;
2661 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2662 * fixed some bugs shown by Paul Dietz' test suite:
2663 ** DOLIST puts its body in TAGBODY;
2664 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2666 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2667 value producing form;
2668 ** if more variables are given to PROGV than values, extra
2669 variables are bound and made to have no value;
2670 ** NSUBSTITUTE on list arguments gets the right answer with
2672 ** ELT signals an error of type TYPE-ERROR when the index argument
2673 is not a valid sequence index;
2674 ** LOOP signals (at macroexpansion time) an error of type
2675 PROGRAM-ERROR when duplicate variable names are found;
2676 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2677 ** FUNCALL of special-operators now cause an error of type
2679 ** PSETQ now works as required in the presence of side-effecting
2680 symbol-macro places;
2681 ** NCONC accepts any object as its last argument;
2682 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2684 ** loop-for-as-package does not require a package to be explicitely
2686 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2688 * fixed bug 166: compiler preserves "there is a way to go"
2689 invariant when deleting code.
2690 * fixed bug 172: macro lambda lists with required arguments after
2691 &REST arguments now cause an error to be signalled. (thanks to
2693 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2694 bug. (thanks to Matthew Danish)
2695 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2697 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2698 arguments are lexically rebound. (thanks to Gerd Moellmann and
2700 * fixed bug 194: error messages are now more informative when there
2701 is no primary method applicable in a call to a generic
2702 function. (thanks to Gerd Moellmann)
2703 * fixed bug in command line argument checking (thanks to Julian
2705 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2706 (thanks to Matthew Danish)
2707 * incremented fasl file version number, because of the
2708 SXHASH-related changes in the layout of CLOS data structures
2710 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2711 * minor incompatible change: The runtime (the Unix executable named
2712 "sbcl") is now much pickier about the .core files it will load.
2713 Essentially it now requires .core files to descend from the same
2714 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2715 as the runtime does. (The intent is to prevent the crashes which
2716 can occur, and which can even be reported as mysterious failures,
2717 when people patch the sources or change the build parameters
2718 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2719 sbcl and .core files.)
2720 * fixed bug: VALUES-LIST is no longer optimized away.
2721 * fixed bug 142: The FFI conversion of C string values to Lisp
2722 string values no longer conses excessively. (thanks to Nathan
2723 Froyd porting Raymond Toy's fix to CMU CL)
2724 * began to systematize and improve MOP conformance in PCL (thanks to
2725 Nathan Froyd, Gerd Moellman and Pierre Mai):
2726 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2728 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2729 ** FINALIZE-INHERITANCE is now called on class finalization;
2730 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2731 argument precedence order.
2732 * fixed bug 202: The compiler no longer fails on functions whose
2733 derived types contradict their declared type.
2734 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2735 so it can be non-toplevel.
2736 * The fasl file version number has changed (because of the new
2737 implementation of DEFMACRO).
2738 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2739 safe code, that any length requirement by their type-specifier
2740 argument is valid. The exceptions to this are described in bug
2742 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2743 functions now check the types of their inputs as required by ANSI.
2744 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2745 introduced symbol is DECLAREd to be SPECIAL.
2746 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2747 * fixed bug: :COUNT argument to sequence functions may be negative
2748 * fixed bug: body of DO-SYMBOLS may contain declarations
2749 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2750 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2751 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2752 symbol macro only once
2753 * fixed printing of call frame when argument list is unavailable
2754 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2755 * compiler no longer signals WARNING on unknown keyword
2758 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2759 * A beta-quality port to the mips architecture running Linux,
2760 based on the old CMUCL backend, has been made. It has been tested
2761 on a big-endian kernel, and works sufficiently well to be able to
2762 rebuild itself; it has not been tested in little-endian mode.
2763 * fixed an inconsistency between gencgc.c and purify.c which made
2764 dumping/loading .core files unreliable
2765 * fixed bug 120a: The compiler now deals correctly with IFs where
2766 the consequent is the same as the alternative, instead of
2767 misderiving the return type. (thanks to Alexey Dejneka)
2768 * fixed bug 113: Logical pathnames are now dumpable (the logical
2769 host is resolved at load-time, throwing an error if it is not
2771 * fixed bug 174: FORMAT's error message is slightly clearer when a
2772 non-printing character is used in a format directive.
2773 * fixed several bugs in compiler checking of type declarations, i.e.
2774 violations of the Python "declarations are assertions" principle
2775 (thanks to Alexey Dejneka)
2776 * fixed several bugs in PCL's error checking (thanks to Gerd
2778 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2780 * fixed bug in compilation of functions as first class values
2781 (thanks to Antonio Martinez-Shotton)
2782 * The compiler's handling TYPE-ERRORs which it can prove will
2783 inevitably happen at runtime has been cleaned up and corrected
2784 in several ways. (thanks to Alexey Dejneka)
2785 * improved argument type checking for various basic arithmetic
2786 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2787 lavished on them in the past that they can be compiled in many
2788 ways in different special cases
2789 * fixed bug 181: compiler checks validity of user supplied type
2791 * cleaned up code flushing in optimization: Function calls which
2792 should signal errors for safety purposes (e.g. which ANSI says
2793 should signal errors when their arguments are of incorrect type)
2794 are no longer optimized away.
2795 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2796 * incremented fasl file version number, because changes in the
2797 implementation of sequence functions like COERCE caused
2798 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2799 in old inline expansions) to become undefined. (Actually these
2800 changes were later undone, so we might very well be binary
2801 compatible with 0.7.7 after all, but leaving the version number
2802 incremented seemed like the simplest and most conservative
2805 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2806 * An alpha-quality port to the parisc architecture running Linux,
2807 based on the old CMUCL backend, has been made. This, even more so
2808 than the other backends, should be considered still a work in
2809 progress; known problems include that the Linux kernel in 64-bit
2810 mode does not propagate the correct sigcontext structure to
2811 userspace, and consequently SBCL on a parisc64 kernel will not
2813 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2814 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2815 suppressed.) Also now that the compiler is looking at declarations
2816 in the environment, it checks optimization declarations as well,
2817 and suppresses inlining when (> DEBUG SPEED).
2818 * More fixes have been made to treatment of floating point exception
2819 treatment and other Unix signals. In particular, floating point
2820 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2821 * The detection and handling of control stack exhaustion (infinite
2822 or very deeply nested recursion) has changed. Stack exhaustion
2823 detection is now done by write-protecting pages at the OS level
2824 and applies at all optimization settings; when found, a
2825 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2826 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2827 be caught using IGNORE-ERRORS.
2828 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2829 correctly when transforming recursive LABELS functions to LETs.
2830 (thanks to Alexey Dejneka)
2831 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2832 that are names of constants or global variables.
2833 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2834 alien routines with docstrings.
2835 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2836 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2838 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2839 (thanks to Eric Marsden)
2840 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2841 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2842 * Bug fix: The SPARC backend can now compile functions involving
2843 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2844 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2845 to a functions expecting a PATHNAME-DESIGNATOR.
2846 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2847 lambda lists. (thanks to Alexey Dejneka)
2848 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2849 functions so that EXPORTing the name of the class doesn't cause
2850 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2851 to Antonio Martinez for reporting this)
2852 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2853 constant-folded without failing an assertion. (thanks to Einar
2854 Floystad Dorum for reporting this)
2855 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2856 types are now dealt with more correctly by the compiler.
2857 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2858 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2859 behaviour analogous to RENAME-FILE. This puts its behaviour more
2860 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2862 * The fasl file version number has changed again. (because of the
2863 bug fix involving the names of PCL MAKE-INSTANCE functions)
2865 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2866 * bug fix: Floating point exceptions are treated much more
2867 consistently on the x86/Linux and PPC/Linux platforms.
2868 * Array initialization with :INITIAL-ELEMENT is now much faster for
2869 cases when the compiler cannot open code the array creation, but
2870 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2871 array accesses have also seen a speed increase.
2872 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2873 specified. (This was at the root of some bad interactions between
2874 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2876 * bug fix: Internal error arguments for undefined functions are now
2877 computed correctly on the PPC/Linux platform.
2878 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2879 Raymond Toy's patch for CMU CL)
2880 * Support for the Solaris 9 operating environment has been included
2881 (thanks to Daniel Merritt)
2882 * A very ugly but hopefully complete draft of the missing FFI chapter
2883 of the manual has been created by reformatting the corresponding
2884 CMU CL manual chapter into (currently very ugly and incoherent)
2885 DocBook and bringing it up to date for SBCL behavior. Thus, the
2886 manual is now essentially complete, at least by my extreme
2887 once-and-only-once standards, whereby it's acceptable to refer to
2888 the doc strings of SB-EXT functions as the primary documentation.
2889 * The fasl file version number has changed again, due to cleanup of
2890 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2891 is no longer a static symbol.)
2893 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2894 * SBCL now builds with OpenMCL (version 0.12) as the
2895 cross-compilation host; also, more progress has been made toward
2896 bootstrapping under CLISP.
2897 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2899 * bug 158 fixed: The compiler can now deal with integer loop
2900 increments different from 1; fixing this turned out also to fix
2902 * bug 169 fixed: no more bogus warnings about using lexical bindings
2903 despite the presence of perfectly good SPECIAL declarations (thanks
2904 to David Lichteblau)
2905 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2906 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2907 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2909 * bug 180 fixed: Method combination specifications no longer ignore
2910 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2911 * bug fix: Structure type predicate functions now check their argument
2912 count as they should.
2913 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2914 correctly. (thanks to Pierre Mai)
2915 * minor incompatible change: The --noprogrammer option is deprecated
2916 in favor of the new --disable-debugger option, which is very similar.
2917 (The major difference is that it takes effect at a slightly different
2918 time at startup, causing handling of errors in --sysinit and
2919 --userinit files will be affected differently.) The
2920 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2921 been added to allow this functionality to be controlled from ordinary
2922 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2923 maintainers, who might want to run non-interactive scripts to
2924 build SBCL cores which will later be used interactively.)
2925 * minor incompatible change: The LOAD function no longer, when given
2926 a wild pathname to load, loads all files matching that pathname.
2927 Instead, an error of type FILE-ERROR is signalled.
2929 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2930 * bug 147 fixed: The compiler preserves its block link/count
2931 invariants more correctly now so that it doesn't crash. (thanks
2933 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2935 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2936 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2937 no-value-for-this-character as NIL (as specified by ANSI).
2938 * HOST-NAMESTRING on physical pathnames now returns a string that is
2939 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2940 (thanks to Christophe Rhodes)
2941 * The Alpha port handles icache flushing more correctly. (thanks to
2943 * More progress has been made toward bootstrapping under CLISP. (thanks
2944 to Christophe Rhodes)
2945 * The fasl file format has changed again, because dynamic loading
2946 on OpenBSD (which has non-ELF object files) motivated some cleanups
2947 in the way that foreign symbols are transformed and passed around.
2948 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2949 is no longer treated as whitespace by the reader, but instead as
2950 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2951 |A
\7fB|, instead of A as it used to.
2953 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2954 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2955 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2956 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2957 this as of 0.7.1.45, but was left out of the previous news section
2958 (thanks to Dan Barlow)
2959 * SBCL now runs on the Solaris operating system on SPARC architectures
2960 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2961 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2962 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2963 * SPARC backend cleanups, allowing builds of cores optimized for V8
2964 and V9 SPARCS, and also emission of code targeted to a particular
2965 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2967 * SBCL is closer to bootstrapping under CLISP, thanks to various
2968 fixes by Christophe Rhodes.
2969 * The fasl file format has changed again, to allow the compiler's
2970 INFO database to support symbol macros.
2971 * The user manual (in doc/) is formatted into HTML more nicely.
2972 (thanks to coreythomas)
2973 * The system is smarter about SUBTYPEP relationships, especially
2974 those involving NOT types (including types such as ATOM which are
2975 represented internally using NOT types). Thus SUBTYPEP is less
2976 likely to return (VALUES NIL NIL) in general, and in particular
2977 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2979 * The fasl file format has changed again, because the internal
2980 representation of types now includes a new slot to support the new
2981 SUBTYPEP-of-NOT-types logic.
2982 * (not a change in the main branch of SBCL, but a related prototype
2983 which can hopefully be merged into the main branch of SBCL in the
2984 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2985 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2986 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2987 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2988 dispatch macro character combinations. (thanks to Alexey Dejneka)
2989 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2990 that superficially look like logical namestrings correctly.
2991 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2993 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2994 default supplied-p) for &optional and &key arguments. (thanks to
2997 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2998 * incompatible change: The compiler is now less aggressive about
2999 tail call optimization, doing it only when (> SPACE DEBUG) or
3000 (> SPEED DEBUG). (This is an incompatible change because there are
3001 programs which relied on the old CMU-CL-style behavior to optimize
3002 away their unbounded recursion which will now die of stack overflow.)
3003 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3004 for non-GENCGC systems has been increased to 20M (since that
3005 seems much closer to the likely performance optimum for modern
3006 systems than the old 4M value was)
3007 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3008 and *DEBUG-PRINT-LEVEL*
3009 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3010 of CMU CL's support for SPARC, and various endianness and other
3011 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3012 * new syntactic sugar for the Unix command line: --load foo.bar is now
3013 an alternate notation for --eval '(load "foo.bar")'.
3015 ** The system now detects stack overflow and handles it gracefully,
3016 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3017 optimization settings. (This is a good thing in general, and
3018 its introduction in this version should be particularly timely
3019 for anyone whose code fails because of suppression of tail
3021 ** The system now hunts for the C variable "environ" in a more
3022 devious way, to avoid segfaults when the C library version differs
3023 between compile time and run time. (thanks to Christophe Rhodes)
3024 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3025 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3026 ** The compiler no longer issues bogus style warnings for undefined
3027 classes in the same source file as the DEFCLASSes which defined
3028 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3030 ** fixes in CONDITION class precedence list for undefined function
3031 errors (thanks to Alexei Dejneka)
3032 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3033 correctly. (thanks to Dan Barlow)
3034 ** portability fixes aiming at bootstrapping under CLISP (thanks
3035 to Dave McDonald and Christophe Rhodes)
3036 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3037 ** fixes in type translation and and type inference (thanks to
3039 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3040 ** various fixes in the new ports (thanks to Dan Barlow)
3041 * several changes related to debugging:
3042 ** suppression of tail recursion, as noted above
3043 ** stack overflow detection, as noted above
3044 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3045 is now the default. (For some time encapsulation has been more
3046 reliable than the breakpoint-based :ENCAPSULATE NIL
3047 implementation, at least on X86 systems; and I just noticed that
3048 encapsulation also seems closer to the spirit of the ANSI
3051 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3053 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3054 up properly again. (There was a packaging bug in 0.7.0 which
3055 left their definitions in SB-SYS::LOAD-FOREIGN and
3056 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3057 vital for most things which interface to C-level interfaces,
3058 like extensions working with sockets or databases or
3059 Perl-compatible regexes or whatever, and the need to fix
3060 this bug is the main reason that 0.7.1 was released so
3062 ** DEFGENERIC is now choosier about the methods it redefines, so that
3063 reLOADing a previously-LOADed file containing DEFGENERICs does
3064 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3065 cycle now works as it should. (thanks to Alexey Dejneka)
3066 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3067 fixed by Christophe Rhodes. (He actually submitted this patch
3068 months ago, and I delayed until after 0.7.0.)
3069 ** Bug 111 (internal compiler confusion about runtime checks on
3070 FUNCTION types) was fixed by Alexey Dejneka.
3071 * Some internal cleanups (getting rid of variables which aren't
3072 needed now that the byte interpreter is gone) caused the fasl
3073 file format number to change again.
3075 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3076 * major incompatible change: The default fasl file extension, i.e. the
3077 default extension for files produced by COMPILE-FILE, has changed
3078 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3080 ** There are many changes in the implementation of the compiler.
3081 SBCL is now essentially a compiler-only implementation of ANSI
3082 Common Lisp. EVAL still "interprets" a few special cases, but
3083 almost all the interesting cases are handled by creating
3084 a LAMBDA expression, calling COMPILE on it, then calling
3085 FUNCALL on the result.
3086 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3087 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3088 Since the code is newer, there might still be some new bugs
3089 (though not as many as before Martin Atzmueller's fixes:-). But
3090 the new code is substantially simpler and clearer, and hopefully
3091 any remaining bugs will be simpler, less fundamental, and more
3092 fixable then the bugs in the old code.
3093 ** The revised compiler is still a little unsteady on its feet.
3095 *** The debugging information it produces (particularly the names
3096 of FUNCTION objects) is sometimes much less useful than what
3097 the old compiler produced.
3098 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3099 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3100 within a MACROLET) has been temporarily weakened.
3101 ** There are new compiler optimizations for various functions:
3102 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3103 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3104 *** the math functions TRUNCATE, FLOOR, and CEILING
3105 *** the function-of-all-trades COERCE
3106 Mostly these should be transparent, but there's one
3107 potentially-annoying problem (bug 117): when the compiler
3108 inline-expands a function and does type analysis on the result,
3109 it can create control paths which have type mismatches, and
3110 when it can't prove that those control paths aren't taken,
3111 it will issue WARNINGs about the type mismatches. This is
3112 a particular problem in practice for the new sequence functions.
3113 It's not clear how this should be fixed, and for now, a
3114 workaround is given in the entry for 117 in the BUGS file.
3115 ** (Because of the interaction between the two previous items --
3116 occasional inlining problems and new inline expansions -- some
3117 of the new sequence function optimizations won't really kick in
3118 completely until debugging information, and then inlining, are
3119 straightened out in some future version.)
3120 * minor incompatible changes:
3121 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3122 DIRECTORY no longer implicitly promotes NIL slots of its
3123 pathname argument to :WILD. In particular, when you ask for the
3124 contents of a directory (which you used to be able to do without
3125 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3126 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3127 ** changes in behavior that ANSI explicitly defines to be
3128 implementation dependent:
3129 *** The new compiler-only implementation still conforms with ANSI,
3130 but acts a little different than before. Besides the obvious
3131 changes in performance tradeoffs (that the cost per form passed
3132 to EVAL has gone up, and the cost per form executed by EVAL
3133 has gone down), the behavior of the system changes a little
3134 because there are no longer any interpreted function objects.
3135 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3136 e.g. doing COMPILE on the output of interactive DEFUN is
3138 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3140 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3141 has changed from (:CL) to NIL.
3142 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3143 e.g. #\Newline and #\Tab, don't have names specified in the
3144 ANSI Common Lisp standard, is now based on their ASCII symbolic
3145 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3146 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3147 are no longer used for output.
3148 ** changes in internal implementation constants:
3149 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3150 4 million. (If your application spends a lot of time GCing and
3151 you have a lot of RAM, you might want to experiment with
3152 increasing it even more.)
3153 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3154 However, almost all old code should still continue to work without
3155 immediate update, as SB-C-CALL is now a (deprecated) nickname
3157 ** Old operator names in the style DEF-FOO are now deprecated in
3158 favor of new corresponding names DEFINE-FOO, for consistency with
3159 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3160 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3161 internal symbols, but a few supported extensions like
3162 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3163 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3164 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3165 etc. as you get deeper into recursive calls to the debugger
3166 command loop, instead of the old "5]", "5]]", "5]]]"
3167 sequence. (I was motivated to do this when squabbles between
3168 ILISP and SBCL left me very deeply nested in the debugger. In the
3169 short term, this change will probably provoke more ILISP/SBCL
3170 squabbles, but hopefully it will be an improvement in the long run.)
3171 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3172 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3173 function, since declaiming return types involving aliens is
3174 (1) annoyingly messy to do by hand and (2) vital to efficient
3175 compilation of code which calls such functions.
3176 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3177 longer reexported by the SB-EXT package. They're solely useful
3178 for alien code, so it seems more logical that you should get
3179 them from the SB-ALIEN package, not in SB-EXT.
3180 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3181 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3182 features. Instead, the code that they used to control is always
3183 built into the system.
3184 * many other bug fixes
3185 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3186 advantage of the new EVAL-WHEN stuff and to clean them up in
3187 general, and they are now more ANSI-compliant in a number of
3188 ways. Martin Atzmueller is responsible for a lot of this.
3189 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3191 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3192 *** correct ERROR type for various file operations
3193 *** some fixes for Lisp streams
3194 *** DEFMETHOD syntax checking
3195 *** changing old weird representation of debug information as
3196 strings (which, among their other deficiencies, don't transform
3197 correctly when you rename packages, and don't change their
3198 print representation when you change things like *PACKAGE*
3199 and *PRINT-LENGTH*) to symbols and lists of symbols
3200 He also made several improvements and fixed several bugs in DESCRIBE.
3201 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3203 *** misbehavior of WRITE-STRING/WRITE-LINE
3204 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3205 and several other LOOP problems as well
3206 *** DIRECTORY when similar filenames are present
3207 *** DEFGENERIC with :METHOD options
3208 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3209 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3210 *** argument ordering in FIND with :TEST option
3211 *** mishandled package designator argument in APROPOS-LIST
3212 *** various problems in the backquote readmacro
3213 *** a bug in APROPOS
3214 *** probably some others that I'm not describing very well here,
3215 since the CVS log documents them by reference to sbcl-devel
3216 messages, and the SourceForge archives aren't working well.:-(
3217 ** Dan Barlow improved the Alpha port (and is making progress on the
3218 PPC port, for those of you who think different).
3219 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3220 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3221 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3222 continues to work on the SPARC port (for those of you in a position
3223 to look down upon our little PC-compatible boxes from a great height).
3224 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3225 and uses that copy, rather than the current dynamic value, when
3226 it's trying to decide whether to truncate output. Thus e.g.
3227 (let ((*print-lines* 50))
3228 (pprint-logical-block (stream nil)
3230 (let ((*print-lines* 8))
3231 (print (aref possiblybigthings i) stream)))))
3232 should now truncate the logical block only at 50 lines, instead of
3233 often truncating it at 8 lines, as it did before.
3234 * The doc/cmucl/ directory, containing old CMU CL documentation
3235 from the time of the fork, is no longer part of the base system.
3236 SourceForge has shut down its anonymous FTP service, and with it
3237 my original plan for distributing the old CMU CL documentation
3238 there. For now, if you need these files you can download an old
3239 SBCL source release and extract them from it.
3240 * The fasl file version number changed again, for dozens of reasons,
3241 some of which are apparent above.
3243 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3244 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3245 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3246 patch, so that DISASSEMBLE output is much nicer.
3247 * The code in the SB-PROFILE package now seems reasonably stable.
3248 I still haven't decided what the final interface should look like
3249 (I'd like PROFILE to interact cleanly with TRACE, since both
3250 facilities use function encapsulation) but if you have a need
3251 for profiling now, you can probably use it successfully with
3252 the current CMU-CL-style interface.
3253 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3254 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3255 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3256 link now returns the dangling link itself, and for similar
3257 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3258 link itself. (In these cases the old code signalled an error and
3259 looped endlessly, respectively.) Thus, DIRECTORY now works even
3260 in the presence of dangling and cyclic symbolic links.
3261 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3262 is now a supported extension again, since the consensus on
3263 sbcl-devel was that it can be useful for ordinary development
3264 work, not just for debugging SBCL itself.
3265 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3266 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3267 that currently-defined functions might be redefined later with
3268 different return types.
3269 * Hash tables can be printed readably, as inspired by CMU CL code
3270 of Eric Marsden and SBCL code of Martin Atzmueller.
3271 * better error handling in CLOS method combination, thanks to
3272 Martin Atzmueller porting Pierre Mai's CMU CL patches
3273 * more overflow fixes for >16Mbyte I/O buffers
3274 * A bug in READ has been fixed, so that now a single Ctrl-D
3275 character suffices to cause end-of-file on character streams.
3276 In particular, now you only need one Ctrl-D at the command
3277 line (not two) to exit SBCL.
3278 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3280 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3281 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3282 does the right thing.
3283 * The implementation of some type tests, especially for CONDITION
3284 types, is now tidier and maybe faster, due to CMU CL code
3285 originally by Douglas Crosher, ported by Martin Atzmueller.
3286 * Some math functions have been fixed, and there are new
3287 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3288 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3289 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3290 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3291 likely to be built into sbcl-0.7.0.)
3292 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3293 CL, ported to SBCL by Martin Atzmueller)
3294 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3295 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3296 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3297 * There's a new slam.sh hack to shorten the edit/compile/debug
3298 cycle for low-level changes to SBCL itself, and a new
3299 :SB-AFTER-XC-CORE target feature to control the generation of
3300 the after-xc.core file needed by slam.sh.
3301 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3302 COMPILE-FILE is no longer supported, so that now every function
3303 gets an entry point, so that block compilation looks a little
3304 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3305 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3306 returns the number of bytes consed since the system started,
3307 rather than the number consed since the first time the function
3308 was called. (The new definition parallels ANSI functions like
3309 CL:GET-INTERNAL-RUN-TIME.)
3310 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3311 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3312 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3313 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3314 * Fasl file version numbers are now independent of the target CPU,
3315 since historically most system changes which required version
3316 number changes have affected all CPUs equally. Similarly,
3317 the byte fasl file version is now equal to the ordinary
3320 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3321 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3322 is no longer recognized. I apologize for this, because it was
3323 listed in SB-EXT as a supported extension, but I found that
3324 its existing behavior was poorly specified, as well as incorrectly
3325 specified, and it looked like too much of a mess to straighten it
3326 out. I have enough on my hands trying to get ANSI stuff to work..
3327 * many patches ported from CMU CL by Martin Atzmueller, with
3328 half a dozen bug fixes in pretty-printing and the debugger, and
3329 half a dozen others elsewhere
3330 * fixed bug 13: Floating point infinities are now supported again.
3331 They might still be a little bit flaky, but thanks to bug reports
3332 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3333 as flaky as they were.
3334 * The --noprogrammer command line option is now supported. (Its
3335 behavior is slightly different in detail from what the old man
3336 page claimed it would do, but it's still appropriate under the
3337 same circumstances that the man page talks about.)
3338 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3339 are now supported, and enabled by default. Thus, the compiler can
3340 handle many floating point and complex operations much less
3341 inefficiently. (Thus e.g. you can implement a complex FFT
3343 * The compiler now detects type mismatches between DECLAIM FTYPE
3344 and DEFUN better, and implements CHECK-TYPE more correctly, and
3345 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3346 thanks to patches from Martin Atzmueller.
3347 * various fixes to make the cross-compiler more portable to
3348 ANSI-conforming-but-different cross-compilation hosts (notably
3349 Lispworks for Windows, following bug reports from Arthur Lemmens)
3350 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3351 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3352 fixed thanks to Pierre Mai's CMU CL patches.
3353 * fixes to keep the system from overflowing internal counters when
3354 it tries to use i/o buffers larger than 16M bytes
3355 * fixed bug 45a: Various internal functions required to support
3356 complex special functions have been merged from CMU CL sources.
3357 (When I was first setting up SBCL, I misunderstood a compile-time
3358 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3359 * improved support for type intersection and union, fixing bug 12
3360 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3361 more obscure bugs as well
3362 * some steps toward byte-compiling non-performance-critical
3363 parts of the system, courtesy of patches from Martin Atzmueller
3364 * Christophe Rhodes has made some debian packages of sbcl at
3365 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3366 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3367 stable, but are nonetheless usable. When he's ready, I'd be happy
3368 to add them to the SourceForge "File Releases" section. (And if
3369 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3370 * new fasl file format version number (because of changes in
3371 internal representation of (OR ..) types to accommodate the new
3372 support for (AND ..) types, among other things)
3374 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3375 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3377 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3379 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3380 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3381 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3382 * DESCRIBE now works on structure objects again.
3383 * Most function call argument type mismatches are now handled as
3384 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3385 know whether the function will be redefined before the call is
3386 executed. (The compiler could flag local calls with full WARNINGs,
3387 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3388 it doesn't keep track of enough information to know whether calls
3389 are local in this sense.)
3390 * Compiler output is now more verbose, with messages truncated
3391 later than before. (There should be some supported way for users
3392 to override the default verbosity, but I haven't decided how to
3393 provide it yet, so this behavior is still controlled by the internal
3394 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3395 src/compiler/ir1util.lisp.)
3396 * Fasl file format version numbers have increased again, because
3397 support for the Gray streams extension changes the layout of the
3398 system's STREAM objects.
3399 * The Gray subclassable streams extension now works, thanks to a
3400 patch from Martin Atzmueller.
3401 * The full LOAD-FOREIGN extension (not just the primitive
3402 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3403 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3404 but like most other programs, it defaults to copying the Unix
3405 environment from the original process instead of starting the
3406 new process in an empty environment.
3407 * Extensions which manipulate the Unix environment now support
3408 an :ENVIRONMENT keyword option which doesn't smash case or
3409 do other bad things. The CMU-CL-style :ENV option is retained
3410 for porting convenience.
3411 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3412 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3414 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3416 * A patch from Martin Atzmueller seems to have solved the SIGINT
3417 problem, and as far as we know, signal-handling now works cleanly.
3418 (If you find any new bugs, please report them!)
3419 * The system no longer defaults Lisp source file names to types
3420 ".l", ".cl", or ".lsp", but only to ".lisp".
3421 * The compiler no longer uses special default file extensions for
3422 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3423 seems to expect a single default extension for all compiled code,
3424 and there's no compelling reason to try to stretch the standard
3425 to allow two different extensions.) Instead, byte-compiled files
3426 default to the same extension as native-compiled files.
3427 * Fasl file format version numbers have increased again, because
3428 a rearrangement of internal implementation packages made some
3429 dumped symbols in old fasl files unreadable in new cores.
3430 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3431 many fewer weird special cases.
3432 * Bug #17 (differing COMPILE-FILE behavior between logical and
3433 physical pathnames) has been fixed, and some related misbehavior too,
3434 thanks to a patch from Martin Atzmueller.
3435 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3436 by Tim Moore, ported to SBCL by Martin Atzmueller.
3437 * Martin Atzmueller fixed several filesystem-related problems,
3438 including bug #36, in part by porting CMU CL patches, which were
3439 written in part by Paul Werkowski.
3440 * More compiler warnings in src/runtime/ are gone, thanks to
3441 more patches from Martin Atzmueller.
3442 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3445 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3447 * DESCRIBE now works on CONDITION objects.
3448 * The debugger now handles errors which arise when trying to print
3449 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3451 * The build system now uses an additional file, customize-target-features.lisp,
3452 to allow local modifications to the target *FEATURES* list. (The point of
3453 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3454 debugging features enabled, without having to worry about propagating it
3455 into everyone's system when I do a "cvs update".) When no
3456 customize-target-features.lisp file exists, the target *FEATURES* list
3457 should be constructed the same way as before.
3458 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3459 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3460 warn about behavior which is undefined under the ANSI spec. Note: This
3461 is specified by ANSI, but it's not very popular with programmers.
3462 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3463 macro in the SBCL sources for an example of a workaround which you
3464 might use to make portable ANSI-standard code which does what you want.
3465 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3466 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3467 when it's used as a non-toplevel form. (This is required in order
3468 to implement the DEFCONSTANT-EQX macro.)
3469 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3470 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3471 funny things, due to bugs in the implementation of EVAL-WHEN
3472 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3473 non-toplevel DEFCONSTANTs are uncommon.)
3474 * The core file version number and fasl file version number have been
3475 incremented, because the old noncompliant DEFCONSTANT behavior involved
3476 calling functions which no longer exist, and because I also took the
3477 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3478 * fixed bug 1 (error handling before read-eval-print loop starts), and
3479 redid debugger restarts and related debugger commands somewhat while
3481 ** The QUIT debugger command is gone, since it did something
3482 rather different than the SB-EXT:QUIT command, and since it never
3483 worked properly outside the main toplevel read/eval/print loop.
3484 Invoking the new TOPLEVEL restart provides the same functionality.
3485 ** The GO debugger command is also gone, since you can just invoke
3486 the CONTINUE restart directly instead.
3487 ** The TOP debugger command is also gone, since it's redundant with the
3488 FRAME 0 command, and since it interfered with abbreviations for the
3490 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3492 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3493 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3494 some porting work by Martin Atzmueller)
3495 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3496 with a boatload of patches which he ported to SBCL. Now that those
3498 ** The system tries to make sure that its low-priority messages
3499 are prefixed by semicolons, to help people who like to use
3500 syntax highlighting in their ILISP buffer. (This patch
3501 was originally due to Raymond Toy.)
3502 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3503 patches originally written by Raymond Toy.
3504 ** The compiler understands coercion between single-value and
3505 multiple-VALUES type expressions better, getting rid of some very
3506 weird behavior, thanks to patches originally by Robert MacLachlan
3507 and Douglas Crosher.
3508 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3509 lambda lists, thanks to a patch originally by Pierre Mai.
3510 ** The system no longer bogusly warns about "abbreviated type
3512 ** The compiler gets less confused by inlining and RETURN-FROM,
3513 thanks to some patches originally by Tim Moore.
3514 ** The system no longer hangs when dumping circular lists to fasl
3515 files, thanks to a patch originally from Douglas Crosher.
3516 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3517 undefined function error.
3518 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3519 in structure slot initforms) for now, documented workaround instead:-|
3520 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3521 * fixed bug 5: added stubs for various Gray stream functions called
3522 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3523 installed, at least appropriate type errors are generated
3524 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3525 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3526 consistently in DEFMETHOD forms.
3527 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3528 it doesn't seem to affect SBCL after all
3529 * The C runtime system now builds with better optimization and many
3530 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3532 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3534 * The system is now under CVS at SourceForge (instead of the
3535 CVS repository on my home machine).
3536 * The new signal handling code has been tweaked to treat register
3537 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3538 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3539 downstream, e.g. in debug-int.lisp, has implicit dependencies
3540 on the unsignedness of integer representation of machine words,
3541 and that caused the system to bomb out with infinite regress
3542 when trying to recover from type errors involving signed values,
3543 e.g. (BUTLAST '(1 2 3) -1).)
3544 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3545 (This was one of the bugs Peter Van Eynde reported back in July.)
3546 * The system now uses code inspired by Colin Walters' O(N)
3547 implementation of MAP (from the cmucl-imp@cons.org mailing
3548 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3549 inline the MAP operation, and there is more than one
3550 sequence argument to the MAP call (so that it can't just
3551 do ETYPECASE once and for all based on the type of the
3552 single sequence argument). (The old non-inline implementation
3553 of the general M-argument sequence-of-length-N case required
3554 O(M*N*N) time when any of the sequence arguments were LISTs.)
3555 * The QUIT :UNIX-CODE keyword argument has been renamed to
3556 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3557 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3559 * Raymond Wiker's patches to port dynamic loading from Linux to
3560 FreeBSD have been added.
3561 * The BUGS file is now more nearly up to date, thanks in large part
3562 to Martin Atzmueller's review of it.
3563 * The debugger now flushes standard output streams before it begins
3564 its output ("debugger invoked" and so forth).
3565 * The core version number and fasl file version number have both
3566 been incremented, because of incompatible changes in the layout
3568 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3569 used to not be called for a saved Lisp image.)
3570 * Martin Atzmueller's version of a patch to fix a compiler crash,
3571 as posted on sbcl-devel 13 September 2000, has been installed.
3572 * Instead of installing Martin Atzmueller's patch for the
3573 compiler transform for SUBSEQ, I deleted the compiler transform,
3574 and transforms for some similar consing operations.
3575 * A bug in signal handling which kept TRACE from working on OpenBSD
3577 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3578 away by constant folding
3579 * The system now defines its address space constants in one place
3580 (in the Lisp sources), and propagates them automatically elsewhere
3581 (through GENESIS and the sbcl.h file). Therefore, patching the
3582 address map is less unnecessarily tedious and error-prone. The
3583 Lisp names of address space constants have also been systematized.
3584 * CVS tags like dollar-Header-dollar have been removed from
3585 the sources, because they have never saved me trouble and
3586 they've been source of trouble working with patches and other
3587 diff-related operations.
3588 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3589 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3591 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3593 * The system has been ported to OpenBSD.
3594 * The system now compiles with a simple "sh make.sh" on the systems
3595 that it's supported on. I.e., now you no longer need to tweak
3596 text in the target-features.lisp-expr and symlinks in src/runtime/
3597 by hand, the make.sh takes care of it for you.
3598 * The system is no longer so grossly inefficient when compiling code
3599 involving vectors implemented as general (not simple) vectors (VECTOR T),
3600 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3601 which dares to use the various sequence functions on non-simple
3602 vectors, takes less of a performance hit.
3603 * There is now a primitive type predicate VECTOR-T-P
3604 to test for the (VECTOR T) type, so that e.g.
3605 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3606 can now be compiled with some semblance of efficiency. (The old code
3607 turned the type declaration into a full call to %TYPEP at runtime!)
3608 * AREF on (VECTOR T) is still not fast, since it's still compiled
3609 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3610 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3611 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3612 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3613 without making full calls to SUBTYPEP at runtime.
3614 (Some analogous efficiency issues for non-simple vectors specialized to
3615 element types other than T, or for non-simple multidimensional arrays,
3616 have not been addressed. They could almost certainly be handled the
3617 same way if anyone is motivated to do so.)
3618 * The changes in array handling break binary compatibility, so
3619 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3620 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3621 instead of (VALUES T T).
3622 * By following the instructions that Dan Barlow posted to sbcl-devel
3623 on 2 July 2000, I was able to enable primitive dynamic object
3624 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3625 functionality is not implemented (since it calls ld to resolve
3626 library references automatically, requiring RUN-PROGRAM for its
3627 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3628 try to resolve library references) is now supported.
3629 * The system now flushes the standard output streams when it terminates,
3630 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3631 them at several other probably-convenient times, e.g. in each pass of
3632 the toplevel read-eval-print loop, and after evaluating a form given
3633 as an "--eval" command-line option. (These changes were motivated by a
3634 discussion of stream flushing issues on cmucl-imp in August 2000.)
3635 * The source transform for TYPEP of array types no longer assumes
3636 that an array whose element type is a not-yet-defined type
3637 is implemented as an array of T, but instead punts, so that the
3638 type will be interpreted at runtime.
3639 * There is now some support for cross-compiling in make.sh: each of
3640 the phases of make.sh has its own script. (This should be transparent
3641 to people doing ordinary, non-cross-compile builds.)
3642 * Since my laptop doesn't have hundreds of megabytes of memory like
3643 my desktop machine, I became more motivated to do some items on
3644 my to-do list in order to reduce the size of the system a little:
3645 ** Arrange for various needed-only-at-cold-init things to be
3646 uninterned after cold init. To support this, those things have
3647 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3648 symbols with such names are now uninterned after cold init).
3649 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3650 fasl files for cold load.
3651 ** Remove the old compiler structure pooling code (which used to
3652 be conditional on the target feature :SB-ALLOC) completely.
3653 ** Redo the representation of some data in cold init to be more compact.
3654 (I also looked into supporting byte compiled code at bootstrap time,
3655 which would probably reduce the size of the system a lot, but that
3656 looked too complicated, so I punted for now.)
3657 * The maximum signal nesting depth in the src/runtime/ support code has
3658 been reduced from 4096 to 256. (I don't know any reason for the very
3659 large old value. If the new smaller value turns out to break something,
3660 I'll probably just bump it back up.)
3661 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3663 * Many, many bugs reported by Peter Van Eynde have been added to
3664 the BUGS list; some have even been fixed.
3665 * While enabling dynamic object file loading, I tried to make the
3666 code easier to understand, renaming various functions and variables
3667 with less ambiguous names, and changing some function calling
3668 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3669 * While trying to figure out how to do the OpenBSD port, I tried to
3670 clean up some of the code in src/runtime/. In particular, I dropped
3671 support for non-POSIX signal handling, added various comments,
3672 tweaked the code to reduce the number of compilation warnings, and
3673 renamed some files to increase consistency.
3674 * To support the new automatic configuration functionality in make.sh,
3675 the source file target-features.lisp-expr has been replaced with the
3676 source file base-target-features.lisp-expr and the machine-generated
3677 file local-target-features.lisp-expr.
3678 * fixed a stupid quoting error in make.sh so that using CMU CL
3679 "lisp -batch" as cross-compilation host works again
3681 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3683 * DESCRIBE no longer tries to call itself recursively to describe
3684 bound/fbound values, so that it no longer fails on symbols which are
3685 bound to themselves (like keywords, T, and NIL).
3686 * DESCRIBE now works on generic functions.
3687 * The printer now prints less-screwed-up representations of closures
3688 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3689 * A private symbol is used instead of the :EMPTY keyword previously
3690 used to mark empty slots in hash tables. Thus
3691 (DEFVAR *HT* (MAKE-HASH-TABLE))
3692 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3693 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3694 now does what ANSI says that it should. (You can still get
3695 similar noncompliant behavior if bang on the hash table
3696 implementation with all the symbols you get back from
3697 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3698 This breaks binary compatibility, since tests for equality to
3699 :EMPTY are wired into things like the macroexpansion of
3700 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3702 * There's now a minimal placeholder implementation for CL:STEP,
3703 as required by ANSI.
3704 * An obscure bug in the interaction of the normal compiler, the byte
3705 compiler, inlining, and structure predicates has been patched
3706 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3707 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3709 * Missing ordinary arguments in a macro call are now detected even
3710 when the macro lambda list contains &KEY or &REST.
3711 * The debugger no longer complains about encountering the top of the
3712 stack when you type "FRAME 0" to explicitly instruct it to go to
3713 the top of the stack. And it now prints the frame you request even
3714 if it's the current frame (instead of saying "You are here.").
3715 * As specified by ANSI, the system now always prints keywords
3716 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3718 * The default initial SIZE of HASH-TABLEs is now smaller.
3719 * Type information from CLOS class dispatch is now propagated
3720 into DEFMETHOD bodies, so that e.g.
3721 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3723 is now basically equivalent to
3724 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3725 (DECLARE (TYPE SINGLE-FLOAT X))
3727 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3728 operation, without having to do run-time type dispatch.
3729 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3730 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3731 * Since I don't seem to be making big file reorganizations very often
3732 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3733 is overflowing my ability to conveniently back them up), I've finally
3734 checked the system into CVS. (The CVS repository is on my home system,
3735 not at SourceForge -- putting it on SourceForge might come later.)
3736 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3737 high-level GC-NOTIFY-FOO functions send their output. (There's
3738 still very little control of where low-level verbose GC functions
3739 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3740 less than it used to -- the GC-NOTIFY-FOO functions are now under
3741 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3742 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3743 in only one place in the source code, and propagates it automatically
3744 everywhere that it's needed. Thus e.g. when I bump the version from
3745 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3746 * The C source files now include boilerplate legalese and documentation
3747 at the head of each file (just as the Lisp source files already did).
3748 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3749 to his page will be replaced with a link to his new CLiki service.
3751 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3753 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3754 * The build process now looks for GNU make under the default name "gmake",
3755 instead of "make" as it used to. If GNU make is not available as "gmake"
3756 on your system, you can change this default behavior by setting the
3757 GNUMAKE environment variable.
3758 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3759 can build without error under CMU CL.
3761 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3763 * There is now a partial SBCL user manual (with some new text and some
3764 text cribbed from the CMU CL manual).
3765 * The beginnings of a profiler have been added (starting with the
3766 CMU CL profiler and simplifying and cleaning up). Eventually the
3767 main interface should be through the TRACE macro, but for now,
3768 it's still accessed through vaguely CMU-CL-style functions and macros
3769 exported from the package SB-PROFILE.
3770 * Some problems left over from porting CMU CL to the new
3771 cross-compilation bootstrap process have been cleaned up:
3772 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3773 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3774 part when getting rid of the compiler *BACKEND* stuff.)
3775 ** The value of *NULL-TYPE* was screwed up, because it was
3776 being initialized before the type system knew the final
3777 definition of the 'NULL type. This screwed up several key
3778 optimizations in the compiler, causing inefficiency in all sorts
3779 of places. (I found it because I wanted to understand why
3780 GET-INTERNAL-RUN-TIME was consing.)
3781 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3782 PROCLAIM FTYPE information. Unfortunately this broke binary
3783 compatibility again, since now the forms output by DEFGENERIC
3784 to refer to functions which didn't exist in 0.6.3.
3785 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3786 can use the new (as of 0.6.3) transform for SOME into MAP into
3788 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3789 values to (INTEGER 0 1000000), so that the time code will no longer
3790 occasionally get blown up by Linux returning 1000000 microseconds
3791 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3792 its output conform to the ANSI spec. (Alas, this makes its output
3793 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3794 * A full call to MAP NIL with a single sequence argument no longer conses.
3795 * fixes to problems pointed out by Martin Atzmueller:
3796 * The manual page no longer talks about multiprocessing as though
3797 it were currently supported.
3798 * The ILISP support patches have been removed from the distribution,
3799 because as of version 5.10.1, ILISP now supports SBCL without us
3800 having to maintain patches.
3801 * added a modified version of Raymond Toy's recent CMU CL patch for
3802 EQUALP comparison of HASH-TABLE
3804 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3806 * The system still can't cross-compile itself with
3807 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3808 entails), but at least it can compile more of itself that way
3809 than it used to be able to, and various buglets which were uncovered
3810 by trying to cross-compile itself that way have now been fixed.
3811 * This release breaks binary compatibility again. This time
3812 at least I've incremented the FASL file format version to 2, so that the
3813 problem can be detected reliably instead of just causing weird errors.
3814 * various new style warnings:
3815 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3816 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3817 ** using the lexical binding of a variable named in the *FOO* style
3818 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3819 as specified by ANSI.
3820 * *RANDOM-STATE* is no longer automatically initialized from
3821 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3822 default behavior of the system is to repeat its behavior every time
3823 it's run. If you'd like to change this behavior, you can always
3824 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3825 old convention there was no comparably easy way to get the system to
3826 repeat its behavior every time it was run.
3827 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3828 FTYPE declarations has been removed, in favor of their ANSI
3829 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3830 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3831 the types of their sequence arguments can be determined at compile time.
3832 This is done through a new open code expansion for MAP which eliminates
3833 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3834 argument types can be determined at compile time.
3835 * The optimizer now transforms COERCE into an identity operation when it
3836 can prove that the coerced object is already of the correct type. (This
3837 can be a win for machine generated code, including the output of other
3838 optimization transforms, such as the MAP transform above.)
3839 * Credit information has been moved from source file headers into CREDITS.
3840 * Source file headers have been made more standard.
3841 * The CASE macro now compiles without complaining even when it has
3844 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3846 * (Note that the way that the PCL macroexpansions were rewritten
3847 to accommodate the change in DEFGENERIC below breaks binary
3848 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3849 not run under sbcl-0.6.2. Once we get out of alpha releases,
3850 i.e. hit release 1.0.0, we'll probably try to maintain binary
3851 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3852 and sbcl-1.4.4. Until then, however, it might be fairly common
3853 for maintenance releases to break binary compatibility.)
3854 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3856 * The reporting of unhandled signals has been changed to print some
3857 explanatory text as well as the report form. (Previously only
3858 the report form was printed.)
3859 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3860 it defines, so that the compiler no longer issues undefined function
3861 warnings for compiled-but-not-yet-loaded generic functions.
3862 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3863 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3864 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3865 section of the ANSI spec are supported.
3866 * The "" nickname for the "KEYWORD" package has been removed.
3867 The reader still handles symbol tokens which begin with a package marker
3868 as keywords, but it doesn't expose its mechanism for doing so in the
3869 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3870 * The system now issues STYLE-WARNINGs for contradictory TYPE
3871 proclamations. (Warnings for contradictory FTYPE proclamations would
3872 be nice too, but those can't be done usefully unless the type system
3873 is made smarter about FUNCTION types.)
3874 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3875 systematized, so that "*target-*.lisp is supposed to exist only on the
3876 target and imply that there's a related file which exists on the
3877 host, and *host-*.lisp is supposed to exist only on the host and imply
3878 that there's a related file which exists on the target. This involves a
3879 lot of renaming. Hopefully the acute confusion caused by the renaming
3880 will be justified by the reduction in chronic confusion..
3881 ** runtime-type.lisp -> early-target-type.lisp
3882 ** target-type.lisp -> late-target-type.lisp
3883 ** early-host-format.lisp -> early-format.lisp
3884 ** late-host-format.lisp -> late-format.lisp
3885 ** host-error.lisp -> misc-error.lisp
3886 ** early-error.lisp -> early-target-error.lisp
3887 ** late-error.lisp -> late-target-error.lisp
3888 ** host-defboot.lisp -> early-defboot.lisp
3889 ** code/misc.lisp -> code/target-misc.lisp
3890 ** code/host-misc.lisp -> code/misc.lisp
3891 ** code/numbers.lisp -> code/target-numbers.lisp
3892 ** code/early-numbers.lisp -> numbers.lisp
3893 ** early-host-type.lisp -> early-type.lisp
3894 ** late-host-type.lisp -> late-type.lisp
3895 ** host-typep.lisp -> typep.lisp
3896 ** load.lisp -> target-load.lisp
3897 ** host-load.lisp -> load.lisp
3898 ** host-disassem.lisp -> disassem.lisp
3899 ** host-insts.lisp -> insts.lisp
3900 ** byte-comp.lisp -> target-byte-comp.lisp
3901 ** host-byte-comp.lisp -> byte-comp.lisp
3902 ** host-signal.lisp -> signal.lisp
3903 ** host-defstruct.lisp -> defstruct.lisp
3904 ** late-target-type.lisp -> deftypes-for-target.lisp
3905 Furthermore, several other previously target-only files foo.lisp (e.g.
3906 hash-table.lisp and random.lisp) have been split into a target-and-host
3907 foo.lisp file and a target-only target-foo.lisp file, with their key type
3908 definitions in the target-and-host part, so that the cross-compiler will
3909 know more about target types.
3910 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3911 gone away. In their place are various *BACKEND-FOO* variables
3912 corresponding to the slots of the old structure.
3913 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3914 into the target SBCL has been fixed.
3915 * The chill.lisp system for loading cold code into a running SBCL
3917 * Support for the CMU CL "scavenger hook" extension has been removed.
3918 (It was undocumented and unused in the CMU CL sources that SBCL was
3919 derived from, and stale in sbcl-0.6.1.)
3920 * Various errors in the cross-compiler type system were detected
3921 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3922 (enabling various consistency checks). Many of them were fixed,
3923 but some hard problems remain, so the compiler is back to
3924 running without *TYPE-SYSTEM-INITIALIZED* for now.
3925 * As part of the cross-compiler type system cleanup, I implemented
3926 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3927 * I have started adding UNCROSS calls throughout the type system
3928 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3929 on cross-compiler input files will be able to go away, and various
3931 * CONSTANTP now returns true for quoted forms (as explicitly required
3934 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3936 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3937 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3938 (MAKE-LIST -1). In the longer term, it ought to become true
3939 that declarations are assertions even at SAFETY 1. For now, it's not
3940 quite true even at SAFETY 3, but it's at least more nearly true..
3941 (Note that this change seems to increases the size of the system by
3942 O(5%) and to decrease the speed of the compiler by 20% or more.)
3943 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3944 for the problem of printing dozens of lines of distracting information
3945 about low-level system machinery as part of the top stack frame
3946 on entry to the debugger when an undefined function was called.
3947 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3948 is not reset to COMMON-LISP-USER.
3949 * Compilation of stuff related to dyncount.lisp has been made conditional
3950 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3951 smaller. The various dyncount-related symbols have been moved into
3952 a new "SB-DYNCOUNT" package.
3953 * tty-inspect.lisp has been renamed to inspect.lisp.
3954 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3955 feature has gone away. (When we eventually port to other flavors of
3956 libc and/or Unix, we'll try to make the differences between flavors
3957 invisible at the user level.)
3958 * Various other *FEATURES* tags, and/or their associated conditionals,
3959 have been removed if obsolescent, or given better documentation, or
3960 sometimes given more-mnemonic names.
3962 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3964 * tidied up "make.sh" script
3965 * tidied up system directory structure
3966 * better "clean.sh" behavior
3967 * added doc/FOR-CMUCL-DEVELOPERS
3968 * many many small tweaks to output format, e.g. removing possibly-confusing
3969 trailing #\. character in DESCRIBE-INSTANCE
3970 * (EQUALP #\A 'A) no longer signals an error.
3971 * new hashing code, including EQUALP hashing
3972 * tidied up Lisp initialization and toplevel
3973 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3974 * command line argument processing
3975 * added POSIX-GETENV function to deal with Unix-ish environment variables
3976 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3977 terminating SBCL on EOF
3978 * non-verbose GC by default
3979 * There is no more "sbcl" shell script; the sbcl file is now the C
3980 runtime executable (just like CMU CL).
3981 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3983 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3984 symbol and package information as Lisp native symbol and package objects
3985 instead of strings naming symbols and strings naming packages. This way,
3986 whenever packages are renamed (as in warm init), debug information is
3987 transformed along with everything else.
3988 * tweaked the optimization policy declarations which control the building
3989 of SBCL itself. Now, among other things, the system no longer saves
3990 source location debugging information. (This helps two problems at once
3991 by reducing SBCL size and by keeping SBCL from trying to look for its
3992 sources -- which may not exist -- when reporting errors.)
3993 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3994 debugging and testing purposes
3995 * cleaned up printing, making the printer call PRINT-OBJECT for
3996 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3997 methods, giving nearly-ANSI behavior
3998 * converted almost all special variables to use *FOO* naming convention
3999 * deleted PARSE-TIME functionality, since it can be done portably
4000 * moved some files out of cold init into warm init
4001 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4003 * regularized formatting of source files
4004 * added an install.sh script
4005 * fixed ridiculous memory usage of cross-compiler by making
4006 compiler/alloc.lisp not try to do pooling unless it can hook
4007 itself into the GC of the cross-compilation host. Now the system
4008 builds nicely on my old laptop.
4009 * added :SB-ALLOC in target-features.lisp-expr
4010 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4011 was not implemented)
4012 * re-did condition handling and note reporting in the compiler. Notes
4013 are no longer handled by signalling conditions. Style warnings
4014 and warnings are handled more correctly and reported in such a way
4015 that it's easy to find one or the other in your output (so that you
4016 can e.g. figure out which of many problems caused COMPILE-FILE to
4018 * changed the severity of several compiler warnings from full WARNING
4019 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4020 compiler note reporting so that it doesn't use the condition system
4021 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4022 COMPILE-FILE command)
4023 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4024 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4025 no longer supported, since their implementation was deeply intertwingled
4026 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4027 * removed bogus "support" for compiler macros named (SETF FOO), and
4028 removed the compiler macro for SETF INFO (but only after making a fool
4029 of myself on the cmucl-imp mailing list by posting a bogus patch for
4030 DEFINE-COMPILER-MACRO..)
4031 * Compiled files containing forms which have side effects on the Lisp
4032 reader (such as DEFPACKAGE forms) are now handled more correctly.
4033 (Compiler queuing of top level lambdas has been suppressed by setting
4034 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4035 * deleted various currently-unused source files, e.g. gengc.lisp. They
4036 may be added back at some point e.g. when porting to other architectures,
4037 but until they are it's distracting to distribute them and to try to
4039 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4040 there were so many of them they're just distractions, and UNCROSS is
4041 known to be able to handle the current sources
4042 * moved PROFILE functionality into TRACE, so that it will be clear
4043 how the wrapping and unwrapping of functions when you profile them
4044 interacts with the wrapping and unwrapping of functions when you
4045 trace them. (Actually, the functionality isn't there yet, but at least
4046 the interface specification is there. Hopefully, the functionality will
4047 arrive with some future maintenance release.)
4048 * removed host-oops.lisp
4049 * changed signature of QUIT function to allow UNIX-CODE argument
4050 * fixed READ-SEQUENCE bug
4051 * tweaked verbose GC output so that it looks more like the progress
4052 output that ANSI specifies for functions like LOAD
4053 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4054 * added <http://sbcl.sourceforge.com> to the banner information printed by