1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
3 * documentation: unwinding from asyncronous events has been
5 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
6 in multithreaded application code.
7 * bug fix: GC deadlocks from asynchronous interrupts has been fixed
8 by disabling interrupts for the duration of any
9 SB-SYS:WITHOUT-GCING section.
11 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
12 * new platform: experimental support for x86-64/darwin (MacOS).
13 * incompatible change: the thread-safe (on most platforms) getaddrinfo
14 and getnameinfo sockets functions are used instead of gethostbyaddr
15 and gethostbyname, on platforms where the newer functions are available.
16 As a result, the ALIASES field of HOST-ENT will always be NIL on these
18 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
19 don't print the startup banner, but behave as if --noinform was passed
20 as a command line argument. (thanks to Kevin Reid)
21 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
22 * optimization: code using alien values with undeclared types is much faster.
23 * optimization: the compiler is now able to open code SEARCH in more cases.
24 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
25 * bug fix: using standardized COMMON-LISP special variables as loop
26 variables no longer signals bogus package lock violations. (reported
28 * bug fix: declaring local loop variables to be of a range-limited type
29 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
30 (reported by Andras Simon)
31 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
32 bugs remain on x86-64.)
33 * bug fix: the #= and ## reader macros now interact reasonably with
34 funcallable instances.
35 * bug fix: type-checks for function arguments were compiled using the
36 compiler policy settings of the wrong lexical environment (in 1.0.2
38 * bug fix: SHADOW accepts characters as string designators, as required
39 by the spec (thanks to Eric Marsden)
40 * bug fix: fixed GC safety issues when foreign functions are called with
41 non-base strings as arguments
42 * bug fix: more consistent error output for fatal-compiler-warnings, like
44 * bug fix: functions with proper names were showing up as NIL in
45 backtraces (reported by Edi Weitz, regression in 1.0.2)
47 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
48 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
49 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
50 produce more efficient code on the x86 if the compiler can determine
51 that (+ INDEX <constant>) does not require a bounds check and FOO
52 has an element type at least 8 bits wide.
53 * bug fix: references to undefined variables are handled the same way
54 in toplevel forms as in the normal compiler
55 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
57 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
58 (thanks to Jon Buller)
59 * improvement: faster compilation times for complex functions
60 * improvement: added readlink support to SB-POSIX (thanks to Richard
63 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
64 * improvement: experimental support for mach exception handling on
65 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
67 * improvement: support for GBK external format.
68 (thanks to Chun Tian (binghe))
69 * improvement: the debugger now displays variables that have been closed
70 over, in code compiled with (DEBUG 2) or higher
71 * improvement: support for executable cores on NetBSD (thanks to
73 * new feature: added a RESTART-FRAME debugger command
74 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
75 be used to provide an implementation for FILE-POSITION on Gray streams
76 (thanks to Eric Marsden)
77 * new feature: add syslog, openlog and closelog support to SB-POSIX
78 (thanks to Richard Kreuter)
79 * optimization: the function call overhead in code compiled with
80 a high DEBUG optimization setting is significantly
81 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
82 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
83 (thanks to Tony Martinez)
84 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
85 evaluated once (reported by Kevin Reid)
86 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
87 works again (thanks to Kilian Sprotte)
88 * bug fix: an error is signaled for tagbodies with duplicate tags
89 (thanks to Stephen Wilson)
90 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
91 * bug fix: Win32 port can now handle foreign code unwinding Lisp
92 stack frames from alien callbacks.
93 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
94 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
95 65k pages (thanks to David Woodhouse)
96 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
98 changes in sbcl-1.0.1 relative to sbcl-1.0:
99 * new platform: FreeBSD/x86-64, including support for threading.
100 * new feature: the compiler stores cross-referencing information
101 abount function calls (who-calls), macroexpansion (who-macroexpands)
102 and special variables (who-binds, who-sets, who-references) for code
103 compiled with (< SPACE 3). This information is available through the
104 sb-introspect contrib.
105 * new feature: users may subclass SEQUENCE, and have instances of
106 these classes interoperate with standard Common Lisp functions if
107 a number of methods are defined. (This feature is experimental
108 and the interface subject to change based on feedback from SBCL
109 users and the general community)
110 * improvement: sb-sprof traces call stacks to an arbitrary depth on
111 x86/x86-64, rather than the previous fixed depth of 8
112 * improvement: another pthread back-end of mutex "pthread-futex".
113 * bug fix: non-ascii command-line arguments are processed correctly
114 (thanks to Yaroslav Kavenchuk)
115 * bug fix: non-required arguments were not passed correctly when a method
116 defined using DEFMETHOD was called from a mop-generated method using
117 CALL-NEXT-METHOD (reported by Pascal Costanza)
118 * bug fix: recursion is now permitted in accessors through
119 SLOT-UNBOUND. (reported by Pascal Costanza)
120 * bug fix: an error was signaled at startup if the HOME environment
121 variable was defined, but had an empty value (reported by Peter Van Eynde)
122 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
123 signal a PROGRAM-ERROR, not a BUG.
124 * bug fix: SB-POSIX:READDIR works when built with large file support.
125 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
127 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
128 for better automatic indentation support. (thanks to Matt Pillsbury)
129 * optimization: loading generic functions no longer takes O(n^2) time,
130 proportional to the amount of methods in the generic function
131 (reported by Todd Sabin and Jeremy Brown)
132 * optimization: the FIND and POSITION family of sequence functions
133 are significantly faster on arrays whose element types have been
135 * improvements to the Windows port:
136 ** Intermittent heap corruption problems have been fixed. (thanks
137 to Alastair Bridgewater)
138 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
140 ** Lisp is able to unwind foreign exception frames from alien
141 callbacks. (thanks to Alastair Bridgewater)
143 changes in sbcl-1.0 relative to sbcl-0.9.18:
144 * improvement: experimental support for threading on FreeBSD/x86.
145 (thanks to NIIMI Satoshi)
146 * improvement: runtime option --dynamic-space-size can be used
147 to set the size of the dynamic space reserved on startup.
148 * improvement: floating point modes in effect are now saved in
149 core, and restored on startup.
150 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
151 startup, not time since first call to GET-INTERNAL-REAL-TIME.
152 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
153 threads are not running after *SAVE-HOOKS* have run.
154 * improvement: writes to CLOS instance slots are type-checked in code
155 compiled with (SAFETY 3)
156 * improvement: floating-point exception handling on FreeBSD (thanks to
158 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
159 (thanks to Zach Beane)
160 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
162 * improvement: added support for the Shift-JIS external format.
163 (contributed by NIIMI Satoshi)
164 * improvement: callbacks are supported on Linux/PPC. (thanks to
166 * bug fix: compiler bug triggered by a (non-standard) VALUES
167 declaration in a LET* was fixed. (reported by Kaersten Poeck)
168 * bug fix: file compiler no longer confuses validated and already
169 dumped structurres. (reported by Kaersten Poeck)
170 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
171 fill-pointer signals a type-error as required. (thanks to
173 * bug fix: disassemly of funcallable instances works.
174 * bug fix: single stepping on PPC.
175 * bug fix: fix thread-safety problems in the type system (generally
176 manifesting as nonsensical errors like "STRING is a bad type specifier
177 for sequences" or "The value 1 is not of type FIXNUM")
178 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
179 (reported by Josip Gracin)
180 * bug fix: an error is signaled for attempts to displace arrays with
181 incompatible element types (thanks to Mario Mommer)
182 * bug fix: more correct handling of wide characters in debug info
183 (bug reported by Attila Lendvai and fixed by Juho Snellman)
184 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
185 and don't cause extra consing
186 * optimization: MAP and MAP-INTO are significantly faster on vectors
187 whose elements types have been declared.
188 * Improvements to SB-SPROF:
189 ** Support for allocation profiling
190 ** Reduced profiling overhead, especially for long profiling runs
191 * Improvements to the Windows port:
192 ** floating point exceptions are now reported correctly.
193 ** stack exhaustion detection works partially.
194 ** more accurate GET-INTERNAL-REAL-TIME.
195 ** better RUN-PROGRAM behaviour in case of errors in the child
197 ** PROBE-FILE now simplifies pathnames correctly.
198 ** DIRECTORY now works correctly with :WILD-INFERIORS.
199 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
200 1st 1970 (although time timezone data will be unreliable).
201 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
203 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
204 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
205 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
207 * bug fix: two potential GC deadlocks affecting threaded builds.
208 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
209 returns T (reported by Anton Kazennikov)
210 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
211 with non-variable places
212 * bug fix: remove a race condition in the setting of
213 funcallable-instance functions, this should make threaded CLOS
214 code more stable against memory faults.
215 * bug fix: corruption of specials when unbinding is interrupted by an
216 asynchronous unwind (reported by Hannu Koivisto)
217 * improvement: the debugger will now also display local variables that
218 are only used once, for code compiled with a DEBUG optimization quality
221 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
222 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
223 * incompatible change: External-format support for FFI calls. The
224 SB-ALIEN:C-STRING no longer implies an ASCII
225 external-format. Instead the string is subject to external-format
226 conversion. Additionally return values of type SB-ALIEN:C-STRING
227 are by default of type (SIMPLE-ARRAY CHARACTER), not
228 SIMPLE-BASE-STRING. For an alien type that behaves like the old
229 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
230 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
231 * incompatible change: SB-EXT package no longer contains the
232 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
233 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
235 * incompatible change: the single-stepper is no longer available
236 on Alpha, Mips and Sparc platforms.
237 * minor incompatible change: the direct superclasses of
238 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
239 not (STANDARD-OBJECT FUNCTION). This makes the
240 class-precedence-lists of GENERIC-FUNCTION and
241 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
243 * minor incompatible change: the default stream external format on
244 non-unicode SBCL is no longer detected from the locale, but is always
246 * new feature: Add a version of evaluator that uses an interpreter instead
247 of the compiler. EVAL still uses the compiler by default, to switch it
248 to use the interpreter, set the value of the variable
249 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
250 * minor incompatible change: the single-stepper REPL has been merged
251 with the normal debugger (see the "Stepping" heading of the debugger help
252 for more details). The debugger command STEP will no longer switch
253 to the single-stepper REPL.
254 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
255 for a type now works.
256 * bug fix: loading of default sysinit file works. (thanks to Leonid
258 * bug fix: better detection of circularities in the file-compiler.
259 (reported by Marco Monteiro)
260 * bug fix: the CL pathname functions now work with files that have
261 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
262 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
263 systems with Unix98 pty semantics.
264 * bug fix: ASDF-INSTALL will now work with bsd tar.
265 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
267 * bug fix: timers expiring in dead threads no longer cause a
268 type-error (reported by Paul "Nonny Mouse").
269 * bug fix: thanks to more lightweight single-stepper instrumentation,
270 code compiled with (DEBUG 3) will compile and execute significantly faster,
271 and will have more accurate type-inferencing than before
272 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
273 whose bindings are modified
274 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
275 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
276 ** replace dummy implementations of CL:MACHINE-INSTANCE and
277 CL:SOFTWARE-VERSION with proper version
279 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
280 * feature: implemented the READER-METHOD-CLASS and
281 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
282 as specified by AMOP.
283 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
285 * optimization: faster LOGCOUNT implementation on x86 and x86-64
286 (thanks to Lutz Euler)
287 * optimization: hashing of general arrays and vectors has been
288 improved. (reported by Any Fingerhut)
289 * enhancement: SB-INTROSPECT is now able to find definitions of
290 profiled functions. (thanks to Troels Henriksen)
291 * enhancement: compiler-macro expansion applies now to FUNCALL forms
292 as well, allowing compiler-macros for SETF-functions to expand.
293 * enhancement: step-instrumentation no longer wraps known
294 single-value functions with multiple-value context, allowing
295 better type inference.
296 * fixed bug #337: use of MAKE-METHOD in method combination now works
297 even in the presence of user-defined method classes. (reported by
298 Bruno Haible and Pascal Costanza)
299 * fixed bug #339(c): if there are applicable methods not part of any
300 long-form method-combination group, call INVALID-METHOD-ERROR.
301 (reported by Bruno Haible)
302 * fixed bug #361: the :FUNCTION initarg in the protocol for
303 initialization of methods can now be used to override
304 internally-produced optimized functions. (reported by Bruno
306 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
307 system-provided lambda expression no longer cause warnings about
308 unbound #:|pv-table| symbols.
309 * bug fix: improved the handling of type declarations and the
310 detection of violations for keyword arguments with non-constant
312 * bug fix: potentially erronous calls to PATHNAME and
313 MERGE-PATHNAMES were being flushed in some cases.
314 (reported by Richard Kreuter)
315 * bug fix: compiled calls to TYPEP were mishandling obsolete
316 instances. (reported by James Bielman and Attila Lendvai)
317 * bug fix: format strings with ~> without matching ~< no longer
318 trigger an AVER, but signal an understandable error instead.
319 (reported by Antonio Martinez)
320 * bug fix: specifying an output-file without a pathname-type for
321 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
322 of input-file instead of "fasl". (reported by Robert Dodier)
323 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
324 (reported by James Y Knight).
325 * bug fix: compiler-macros-function did not consider the environment
326 argument for shadowing by local functions.
327 * bug fix: compiler-macros expansion was inhibited by local INLINE
329 * bug fix: inline expansions of known functions were subject to
330 step-instrumentation in high debug policies, leading to problems
332 * bug fix: compiler failed to differentiate between different CONS
334 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
335 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
336 * bug fix: #368: incorrect use of expressed vs. upgraded array
338 * bug fix: #306a: more precise unions of array types.
339 * thread-safety improvements:
340 ** CONDITION-WAIT could return early on Linux, if the thread was
341 interrupted and subsequently continued with SIGCONT.
342 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
344 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
345 * added support for the ucs-2 external format. (contributed by Ivan
347 * minor incompatible change: pretty printing of objects of type
348 (cons symbol) is, in the default pprint-dispatch-table, now
349 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
351 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
352 called later in a class's lifetime, possibly as late as when the
353 first instance of the class is created. Previously,
354 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
355 class became finalizeable.
356 * fixed bug: FILE-POSITION sometimes returned inconsistent results
357 for multibyte external-format streams. (thanks to "vbzoli")
358 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
359 with :ALLOCATION :CLASS inherited from superclasses of the
361 * fixed bug: anonymous classes can now be created using the :NAME
362 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
363 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
364 * fixed bug: core-files saved with :EXECUTABLE T can again be
365 executed when SBCL_HOME isn't set. (reported by James Knight)
366 * fixed bug: toplevel LOCALLY forms with declarations could
367 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
368 * fixed bug: printing from several different threads using different
369 values of *print-case* could cause invalid output, due to
370 some internal special variables of the printer not being bound
371 thread-locally (reported by Max Mikhanosha)
372 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
373 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
374 (reported by Pascal Costanza)
375 * fixed bug: FINALIZE-INHERITANCE is called from
376 REINITIALIZE-INSTANCE on classes when the class has previously
377 been finalized, as required by AMOP.
378 * minor code generation optimizations:
379 ** better register allocation in CLOS dispatching functions
380 ** overflow detection when coercing signed bytes to fixnums on x86-64
381 ** is now implemented with one IMUL instruction instead of three shifts
382 ** more efficient bit-vector access on x86 and x86-64
383 ** more efficient access to raw structure slots on x86 and x86-64
384 * fixed some bugs revealed by Paul Dietz' test suite:
385 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
388 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
389 * feature: thread support on Solaris/x86, and experimental thread support
391 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
393 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
394 default method for SLOT-UNBOUND.
395 * minor incompatible change: prevent the user from specializing the
396 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
397 somewhat counter to the intent of the protocol, I (CSR) think, and
398 additionally it just doesn't work in SBCL as currently
399 implemented, thanks to optimizations (that are always valid for
400 the other three SLOT-VALUEish functions, but not for the setter).
401 * minor incompatible change: the :SB-LDB feature is now enabled by
402 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
403 the low-level debugger.
404 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
405 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
406 down to an SB-SYS:FD-STREAM, allowing direct communication in
407 more cases. Temporary files and pipes are still used as a fallback
409 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
410 on the non-reentrant C functions localtime(3) and gmtime(3).
411 * bug fix: global optimization policy was not visible in LOCALLY and
413 * bug fix: class objects can be used as specializers in methods.
414 (reported by Pascal Costanza)
415 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
416 with a type error. (reported by blitz_ on #lisp)
417 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
418 failed with a type error. (reported by Pascal Bourguignon)
419 * bug fix: merging pathnames against defaults with :DIRECTORY
420 starting with '(:RELATIVE :BACK) should preserve the :BACK.
421 (reported by James Y Knight)
422 * bug fix: saving large (>2GB) cores on x86-64 now works
423 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
424 constant index and a (simple-array (signed-byte 32)) array
425 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
426 error (patch by Robert J. Macomber)
427 * feature: TIME now displays the amount of run-time spent in GC
428 * bug fix: The debugger now does a better job of respecting
429 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
430 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
431 workaround for bug 403.)
432 * bug fix: floating point exception handling now works on FreeBSD
433 (thanks to NIIMI Satoshi)
434 * fixed some bugs revealed by Paul Dietz' test suite:
435 ** MISC.641: LET-conversion were not supposed to work in late
437 * improvements to the Win32/x86 port:
438 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
439 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
441 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
442 * new feature: source path information is generated for macro-expansion
443 errors for use in IDE's like Slime (thanks to Helmut Eller)
444 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
445 * bug fix: compiler does not loop forever on an invalid type in
447 * improvement: compilation of most CLOS applications is significantly
449 * optimization: added a limited bytecode compiler for simple toplevel
450 forms, speeding up compilation and FASL loading
451 * bug fix: the statistical profiler now properly distinguishes anonymous
454 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
455 * minor incompatible change: in sbcl-0.9.11 (but not earlier
456 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
457 into subclasses of the finalized class. Now user calls to
458 FINALIZE-INHERITANCE finalize just the one class, and calls by the
459 system return before any subclasses are finalized.
460 * minor incompatible change: The reader no longer ignores errors
461 regarding non-existent packages in #+ and #- feature tests.
462 * new feature: command line options --no-sysinit, --no-userinit to
463 inhibit loading the corresponding init files
464 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
465 for sending data through UDP sockets (thanks to François-René Rideau)
466 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
467 error is signalled instead (thanks to François-René Rideau)
468 * minor incompatible change: Error signalling behaviour of lexical
469 operations violating package locks has changed slightly. Refer to
470 documentation on package locks for details.
471 * enhancement: EVAL can process IF-expressions without resorting to the
473 * bug fix: Some operations involving SIN, COS, or TAN weren't being
474 constant-folded properly due to missing out-of-line functions.
475 (reported by Mika Pihlajamäki)
476 * bug fix: LISTEN sometimes returned T even in cases where no data was
477 immediately available from the stream
478 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
479 were reversed. (reported by Levente Mészáros)
480 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
481 when given a vector displaced to an adjustable array. (reported
483 * bug fix: derivation of float boundaries from numbers outside the
484 appropriate float range (reported by John Wiseman)
485 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
487 * bug fix: precomputing cache entries for generic functions with
488 some subclasses of specializers as yet invalid does not attempt to
489 fill a cache line with a negative offset. (reported by Levente
491 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
492 allow condition class objects as arguments to DOCUMENTATION and
493 (SETF DOCUMENTATION); only find and set documentation for
494 structure names for the STRUCTURE doc-type. (suggested by Gary
496 * improvements to the Win32/x86 port:
497 ** (user-homedir-pathname) and default initialization file
498 locations now know about the user's "Documents and Settings"
499 directory (thanks to Yaroslav Kavenchuk)
500 ** run-program is implemented (thanks to Mike Thomas)
501 ** sockets support (thanks to Timothy Ritchey)
502 ** better backtrace support (thanks to Alastair Bridgewater)
503 ** sb-grovel supported
504 ** asdf-install and sb-posix work somewhat
505 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
506 * improvements to the Solaris/x86 port:
507 ** works on Solaris 11/Solaris Express
508 ** floating-point exception handling support
509 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
510 * fixed some bugs revealed by Paul Dietz' test suite:
511 ** REMOVE-METHOD returns its generic function argument even when
512 the method is not one of the generic functions' methods.
513 ** objects declared as MEMBER types can be admissible arguments to
515 ** printing characters should simply be printed by the FORMAT ~:C
517 ** compiler failure when compiling functions with hairy constant
518 defaults for optional parameters.
519 ** compiler produces wrong code when MAYBE-INLINE-expanding a
520 function, which is already optimized.
522 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
523 * new platform: experimental support for SBCL x86/Darwin, including
524 MacOS X 10.4.5 on Intel.
525 * new feature: Unicode character names are now known to the system
526 (through CHAR-NAME and NAME-CHAR).
527 * new feature: the filesystem location of SBCL's core file is
528 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
529 * minor incompatible change: the contrib modules SB-POSIX and
530 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
531 this change is to make it easier to distribute
532 location-independent binaries.
533 * bug fix: as implied by AMOP, standardized classes no longer have
534 slots named by external symbols of public packages. (reported by
536 * optimization: calling functions via constant symbols -- as in
537 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
538 via the function object as in (FUNCALL #'FOO).
539 * optimization: on x86 and x86-64, the calling convention for the
540 case of dealing with an unknown number of values has been altered
541 to be friendlier to the prediction heuristics implemented,
542 particularly with reference to CALL and RET pairing. (thanks to
543 Alastair Bridgewater)
544 * optimization: on x86, the code for access to thread-local symbol
545 values and binding/unbinding of thread-local symbols is smaller.
546 (thanks to Alastair Bridgewater)
547 * enhancement: CONSTANTP is now able to determine constantness of
548 more complex forms, including calls to constant-foldable standardized
549 functions and some special forms beyond QUOTE.
550 * fixed bug: occasional GC crashes on Solaris/x86
551 * optimization: x86-64 supports stack allocation of results of simple
552 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
553 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
554 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
555 (thanks to James Knight)
556 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
557 by the standard (thanks to Stephen Compall)
559 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
560 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
561 be used for bundling the runtime and the core file into one
562 executable binary. This feature is not currently supported on all SBCL
563 platforms. (thanks to James Bielman and NIIMI Satoshi)
564 * new feature: a generational or ephemeral garbage collector is now
565 the default on the PowerPC platform (both Linux and Darwin). The
566 old Cheney (stop and copy) collector is a build-time option.
567 (thanks to Cyrus Harmon, after Raymond Toy)
568 * minor incompatible change: the method by which SBCL finds its
569 contributed modules has changed; it no longer relies on symbolic
570 links from an $SBCL_HOME/systems directory, but searches directly
571 in the subdirectories of $SBCL_HOME.
572 * enhancement: the dynamic heap size on the Linux/PPC platform is
573 markedly larger, even using the older Cheney garbage collector.
574 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
575 certain complicated string types. (reported by Gary King)
576 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
577 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
578 * fixed bug: DOCUMENTATION on structures no longer signals an
579 error if no structure type of the right name exists. (reported by
581 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
582 applicable non-standard primary or :AROUND method on
583 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
584 no longer cause unbound variable errors. (reported by Kevin Reid)
585 * fixed bug: metacircle resolution in cases where methods have slots
586 added before the slots from STANDARD-METHOD. (reported by Jean
588 * fixed bug: the Power PC assembly code for calling into lisp
589 sometimes computed its return address wrongly (depending on the
590 15th bit of the address, fixed at link time). This used to
591 manifest itself in a segmentation violation while building PCL.
592 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
593 many others over the years)
594 * fixed bug: ROOM no longer reports silly numbers for stack usage if
595 the stack pointer (treated as a signed integer) is negative.
596 (thanks to Peter van Eynde)
598 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
599 * new platform: experimental support for the Windows operating
600 system has been added. (thanks to Alastair Bridgewater)
601 * fixed several bugs in and robustified the PPC FFI (including
602 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
603 * bug fix: allow non-simple string symbol names (reported by Paul
605 * bug fix: interrupt handling on NetBSD (thanks to Richard M
607 * bug fix: saving a core corrupted callbacks on x86/x86-64
608 * bug fix: closed a loophole in metacircularity detection and
609 grounding in the PCL implementation of CLOS.
610 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
611 (reported by tomppa on #lisp)
612 * bug fix: clear the direction flag on Lisp -> C transitions, as
613 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
614 (reported by Andrej Grozin and Hendrik Maryns)
615 * optimization: major improvements to GC efficiency on GENCGC platforms
616 * optimization: faster implementation of EQUAL
617 * optimization: emit more efficient opcodes for some common
618 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
619 * optimization: several other minor code-generation improvements on x86-64
620 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
622 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
623 * minor incompatible change: (SETF CLASS-NAME) and (SETF
624 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
625 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
626 by ANSI as a generic function, and for consistency (SETF
627 GENERIC-FUNCTION-NAME) is treated likewise.
628 * fixed bug #233.b: make constraint propagation notice when a variable
629 value is changed after it is referenced but before it is used
630 * fixed bug #296: no more arbitrary behaviour from filenames with
631 odd characters as --load arguments. (reported by Adam Warner)
632 * fixed bug #390: :CHARACTER-SET pathname components now work as
633 expected. (reported by Tim Daly Jr)
634 * fixed bug #391: complicated :TYPE intersections in slot
635 definitions no longer cause an error in PCL internals.
636 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
637 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
638 returning the number of octets which would be written to the
639 file-stream. (thanks to Robert J. Macomber)
640 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
641 arguments to the internal stream functions. (thanks to David
643 * bug fix: stack exhaustion is now handled on the x86/SunOS
644 platform. (thanks to Daisuke Homma)
645 * bug fix: more accurate ROOM results on GENCGC platforms
646 * optimization: improved type inference for arithmetic-for
647 index variables in LOOP
648 * optimization: faster floating-point SQRT on x86-64
649 * fixed some bugs revealed by Paul Dietz' test suite:
650 ** DOCUMENTATION returns NIL instead of "" for method combinations
651 that don't have a docstring
653 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
654 * minor incompatible change: (SETF CLASS-NAME) and (SETF
655 GENERIC-FUNCTION-NAME) are no longer generic functions, and
656 therefore are not customizeable by user code (as seems to be at
657 least permitted and maybe required by AMOP). As a consolation,
658 however, the SBCL implementation of these functions now calls
659 REINITIALIZE-INSTANCE as specified by AMOP.
660 * bug fix: slot-definition documentation is propagated as per ANSI
661 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
662 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
663 argument, as do eslotd initialization methods. (from Pascal
664 Costanza's "Closer" project)
665 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
666 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
668 * bug fix: it is now possible to have more than one subclass of
669 STANDARD-GENERIC-FUNCTION without causing stack overflow.
670 (reported by Bruno Haible, Pascal Costanza and others)
671 * bug fix: the dependent update protocol now works for generic
672 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
674 * bug fix: condition-class instances corresponding to
675 DEFINE-CONDITION forms are now created eagerly. (reported by
676 Kalle Olavi Niemitalo on comp.lang.lisp)
677 * bug fix: floating point printing is more accurate in some
678 circumstances. (thanks to Simon Alexander)
679 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
680 merged with *DEFAULT-PATHNAME-DEFAULTS*.
681 * bug fix: callbacks on OS X now preserve stack-alignment.
682 (thanks to Cyrus Harmon)
683 * enhancement: the x86-64 disassembler is much better at
684 disassembling SSE instructions. (thanks to Lutz Euler)
685 * enhancement: CHECK-TYPE on a variable now allows type-inference
686 (thanks to Vincent Arkesteijn)
687 * optimization: improved performance of EUC-JP external format.
688 (thanks to NIIMI Satoshi)
689 * optimization: performance improvements to IO on file streams of
690 :ELEMENT-TYPE CHARACTER
691 * optimization: much faster memory allocation on x86-64
692 * optimization: faster garbage collections (latency and throughput)
694 * optimization: faster bignum subtraction and division on x86 and x86-64
695 (thanks to Lutz Euler)
696 * optimization: more accurate type-inference for loops that use a
697 floating point index variable or a negative step.
699 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
700 * bug fix: add a workaround to SBCL looping infinitely at startup on
701 Linux kernels with apparently buggy implementations of personality().
702 (thanks to Svein Ove Aas)
703 * bug fix: Unicode symbols are correctly printed in LDB backtraces
704 (thanks to David Lichteblau)
705 * bug fix: local bindings shadow global symbol macros for
706 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
707 * optimization: non-open coded uses of numeric comparison operators
708 (e.g. >) no longer cons when called with more than one parameter
709 on platforms supporting dynamic-extent allocation.
710 * enhancement: saving cores with foreign code loaded is now supported
711 on MIPS/Linux in addition to the previously supported platforms.
712 * bug fix: FIND-RESTART now tests for activity, not applicability when given
713 a restart object as identifier. (reported by Helmut Eller for CMUCL)
714 * bug fix: division by zero in sb-sprof when no samples were collected
715 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
716 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
717 (thanks to Alexander Kjeldaas)
718 * new feature: ignoring the return values of destructive functions
719 like NREVERSE signals a compile-time style-warning.
720 (thanks to Kevin Reid)
721 * new feature: the alignment of alien structure fields can be
722 explicitly specified. (thanks to Cyrus Harmon)
723 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
724 and dump core on SIGQUIT
726 ** incompatible change: threads do not inherit values of specials
727 from their parents (see manual)
728 ** bug fix: threads stacks belonging to dead threads are freed by the
729 next exiting thread, no need to gc to collect thread stacks anymore
730 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
731 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
732 ** bug fix: dynamic variable and thread start related gc lossage
733 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
735 * fixed some bugs revealed by Paul Dietz' test suite:
736 ** SUBTYPEP is slightly more accurate on heinously complicated
737 CONS types where some of the members have uncertain (in the
738 NIL, NIL sense) type relationships to each other.
740 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
741 no longer purified unless :PURIFY T is explicitly specified.
742 ** Non-purified cores are significantly smaller than before
744 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
745 * new feature: timers based on Zach Beane's excellent timer package
746 * added support for the following external formats: koi8-u,
747 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
748 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
749 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
750 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
751 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
752 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
753 * incompatible change: a threaded SBCL will no longer revert to
754 non-threaded mode on non-NPTL systems, but refuse to start entirely.
755 * bug fix: interrupts are disabled until startup is complete; no
756 more sigsegvs when receiving a signal to soon
757 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
759 * bug fix: add a workaround for the memory randomization features in
760 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
761 workaround will only be in effect on systems with the proc filesystem
763 * bug fix: printing objects of type HASH-TABLE signals a
764 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
766 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
767 have been running for over 50 days. (reported by Gilbert Baumann)
768 * bug fix: the logic for getting names of functions gets less
769 confused when confronded with alternate-metaclass
770 funcallable-instances. (reported by Cyrus Harmon)
771 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
772 consistent, even on internal alternate-metaclass objects.
773 * bug fix: flush closure information collected by physical
774 environment analysis prepass before the main pass. (bug reported
776 * bug fix: compiler pack phase does not modify a hashtable, which is
777 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
778 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
779 may be expanded. (reported by Alexander <malishev> on c.l.l)
780 * bug fix: no more style-warnings when compiling code using LAST
781 under high-SPEED low-SPACE settings. (reported by David Wallin
783 * fixed some incompatibilities between SBCL's MOP and the MOP
785 ** the METAOBJECT class is now implemented;
786 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
787 STANDARD-OBJECT, as required;
788 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
789 now compatible (as required by VALIDATE-SUPERCLASS); there
790 remains a consistency requirement over the presence or absence
791 of the FUNCTION class in the superclass list of finalized
792 classes; see the manual for more details;
793 ** the :AROUND method for COMPUTE-SLOTS on
794 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
795 requested slot ordering.
797 ** bug fix: parent thread now can be gc'ed even with a live
799 ** bug fix: binding a special with PROGV to no value is not
800 the same as that symbol not having been bound (thanks to
802 ** bug fix: binding specials is thread safe (thanks to
804 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
805 and SB-SPROF do not die with 'no handler for signal XX in
806 interrupt_handle_now(..)' anymore
807 ** bug fix: WITH-TIMEOUT works with multiple running threads
808 * fixed some bugs revealed by Paul Dietz' test suite:
809 ** ENSURE-GENERIC-FUNCTION should take a method class object for
810 the :method-class keyword argument.
812 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
813 * new port: the Solaris operating system on x86 processors is now
814 mostly supported, though some rough edges in the environment
815 remain. (thanks to Daisuke Homma)
816 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
817 space for its heap. (thanks to Thiemo Seufer)
818 * optimization: SBCL on MIPS platforms now supports dynamic-extent
819 closures. (thanks to Thiemo Seufer)
820 * minor incompatible change: eof selects abort in the debugger.
821 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
822 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
824 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
825 deprecated, and will go away in a future revision of SBCL.
826 * minor incompatible change: GC-ON and GC-OFF are no longer
827 implemented with a counter, it does not matter how many times gc
828 is switched on or off
829 * bug fix: discriminating functions for generic function classes
830 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
831 make invalid assumptions about method precedence order. (reported
833 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
834 platforms. (thanks to Thiemo Seufer)
835 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
836 * bug fix: correct pathname printing with printer escaping is on.
837 (thanks to Kevin Reid)
838 * bug fix: complex VOP definitions in "user-space" no longer trigger
839 package locks. (reported by Zach Beane)
840 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
841 for funcallable-instance functions, and is no different from
843 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
844 characters in the current readtable. (reported by Nicholas Neuss)
845 * bug fix: SUBTYPEP on various CONS types returns more a more
846 accurate acknowledgment of its certainty.
847 * optimizations: REMOVE-DUPLICATES now runs in linear time on
848 lists in some cases. This partially fixes bug 384.
849 * flush all standard streams before prompting in the REPL and the
851 * bug fix: signal handling and triggering gc do not conflict
852 directly anymore, in particular a high frequency sb-sprof does
853 not prevent gc from running
854 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
855 approximation for timezone and DST information between the
856 universal time epoch and the smallest negative 32-bit time_t.
857 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
858 year 1899 so long as the encoded time is non-negative.
859 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
860 the correct answer for non-constant rotate amounts when performing
861 an inline 32-bit rotation.
863 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
864 there is only one thread in the session
865 ** bug fix: memory leak for streams created in one thread and
866 written to in another
867 ** bug fix: lockup when compiled with gcc4
868 ** bug fix: race that allows the gc to be triggered when gc is
870 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
871 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
873 ** bug fix: finished threads are gc'ed properly
874 * fixed some bugs revealed by Paul Dietz' test suite:
875 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
876 has arguments with a different set of applicable methods from
877 the orignal arguments.
878 ** The type error thrown by MAP now has a correct expected-type
880 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
881 name a compiled function.
882 ** Three MISC tests where a large bignum was improperly coerced to
883 a float that couldn't represent that bignum during type
884 derivation were fixed.
885 ** SUBTYPEP can now handle types involving the negation of a
886 list-form FUNCTION type.
887 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
888 as (COMPLEX (AND RATIO (NOT FIXNUM))).
889 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
891 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
892 * New feature: Experimental support for bivalent streams: streams
893 opened with :element-type :default now allow character and binary
894 (unsigned-byte 8) I/O
895 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
896 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
898 * bug fix: no more highly sporadic "couldn't check whether ~S is
899 readable" when reading a stream and an interrupt hits in the middle
900 of a select system call
901 * compiler better recognizes complex arrays (reported by Eduardo
903 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
904 for variables on SBCL built with linkage-tables. (reported by Luis
906 * various error reporting improvements.
907 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
908 (Thanks to Thiemo Seufer)
909 * bug fix: nested reader invokations work correctly
910 * bug fix: it is possible to have simultaneous references to foreign
911 code and foreign data with the same name.
913 ** added x86-64 support
914 ** incompatible change: the threading api now works with thread
915 objects instead of thread ids
916 ** bug fix: threads are protected from signals and interruption when
917 starting up or going down
918 ** bug fix: a race where an exiting thread could lose its stack to gc
919 ** bug fix: don't halt on infinite error in threads if possible
920 ** fixed numerous gc deadlocks introduced in the pthread merge
921 ** bug fix: fixed thread safety issues in read and print
922 ** bug fix: debugger doesn't hang on session lock if interrupted at
923 an inappropriate moment
924 ** bug fix: run-program is now thread safe(r)
925 ** bug fix: inner with-recursive-lock no longer releases the mutex
926 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
927 * fixed some bugs revealed by Paul Dietz' test suite:
928 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
929 EXPECTED-TYPE slots filled.
930 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
932 ** MAKE-CONDITION accepts classes as type-designators.
933 ** COMPILE may never return NIL.
934 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
935 range before calling Unix time functions
937 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
938 * numerous signal handling fixes to increase stability
939 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
940 * minor incompatible change: we now correctly canonize default
941 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
942 as per the MOP, rather than the historical (INITARG INITFUNCTION
943 INITFORM). (reported by Bruno Haible)
944 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
945 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
946 for more information.
947 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
948 pathname is a directory pathname.
949 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
950 * dynamic space size on PPC has been increased to 768Mb. (thanks to
952 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
953 :METACLASS argument in addition to a class name. (reported by
954 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
955 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
956 (reported by Bruno Haible; patch from Gerd Moellmann)
957 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
959 * bug fix: more cleanups to the floating point exception handling on
960 x86-64 (thanks to James Knight)
961 * bug fix: the compiler does not try to constant fold calls of
962 COERCE and BIT-* functions when they return freshly constructed
963 objects. (reported by Nikodemus Siivola and Paul Dietz)
964 * optimization: improved the MIPS versions of generic (in the
965 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
966 * optimization: direct conversion of (unsigned-byte 32) to floats on
967 the PowerPC platform.
968 * optimization: structure instances with raw slots now use less
969 memory, and probably show better memory locality. (thanks to
971 * optimization: DYNAMIC-EXTENT declarations for lists and closures
972 are treated as requests for stack allocation on the x86-64,
973 Alpha-32, PPC and SPARC platforms.
974 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
975 the wrong file descriptor; implementation of SOCKET-OPEN-P.
976 (thanks to Tony Martinez)
978 ** gcing a dead thread can no longer lead to lockups
979 ** threads block signals until they are set up properly
980 ** errno is no longer shared by threads
981 ** interrupt-thread restores the eflags register on x86
982 ** fixed some lockups due to gc/thread interaction
983 * fixed some bugs revealed by Paul Dietz' test suite:
984 ** invalid dotted lists no longer raise a read error when
986 ** signal an error if a symbol that names a declaration is used
987 as the name of a type, or vice versa
988 ** allow using the (declare (typespec var*)) abbreviation for
989 (declare (type typespec var*)) with all type specifiers
990 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
991 ** always return NIL from PPRINT-POP when OBJECT is NIL
992 ** don't signal errors when pretty-printing malformed LABELS,
993 FLET or MACROLET forms
994 ** declarations in a DOLIST body are also in scope for the
996 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
998 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1001 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1002 * fixed cross-compiler leakages that prevented building a 32-bit
1003 target with a 64-bit host compiler.
1004 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1005 opened with :IF-EXISTS :OVERWRITE.
1006 * fixed bug 281: error for an invalid qualifier in a short-form method
1007 combination method is not signalled until the faulty method is called.
1008 * bug fix: iteration variable type inferrer failed to deal with open
1009 intervals. (reported by Alan Shields)
1010 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1011 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1012 or more runtime options were provided to the sbcl binary.
1013 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1014 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1015 * optimization: there's now a fast-path for fixnum arguments in the
1016 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1017 * optimization: the code generated on x86-64 is more compact thanks
1018 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1019 * bug fix: floating-point exception handling now partly works on
1020 x86-64. (Thanks to James Knight)
1021 * improvement to the MIPS backend from Thiemo Seufer: C-style
1022 64-bit long long arguments and return values to alien functions
1024 * fixed some bugs revealed by Paul Dietz' test suite:
1025 ** the type-error signalled from WARN has a filled-in DATUM slot.
1026 ** the type-error required when a stream is not associated with
1027 a file has the stream as its datum.
1028 ** type-errors on single-floats on x86-64 no longer have
1029 :INVALID-OBJECT as the datum
1030 ** the type-errors signalled for invalid function names now have
1031 a correct expected type
1032 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1033 for structures defined with a :TYPE. Documentation strings for
1034 typed structures are no longer immediately discarded
1035 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1036 broadcast streams and synonym streams. FILE-LENGTH now also works
1037 on broadcast streams.
1039 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1040 * incompatible change: the --noprogrammer option, deprecated since
1041 version 0.7.5, has been removed. Please use the equivalent
1042 --disable-debugger option instead.
1043 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1045 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1046 inoperational for a while now) has been completely removed.
1047 * null lexical environments are now printed as #<NULL-LEXENV>,
1048 significantly reducing the amount of clutter in typical backtraces.
1049 * documentation on weak pointers, finalization, and after GC hooks
1050 has been added to the manual.
1051 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1052 as other specialized array types, is much faster. SUBSEQ and
1053 COPY-SEQ on such arrays have also been sped up.
1054 * optimization: EQL is now more efficient when at least other argument
1055 is known to be of type (OR FIXNUM (NOT NUMBER)).
1056 * fixed bug: the runtime is now less vulnerable to changes in the
1057 size of the SBCL object on OS X, and virtual memory is reserved for
1058 all spaces that need to be at a fixed address.
1059 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1060 * fixed bug: finalizers and after GC hooks that cause consing are now
1062 * fixed bug: compiler error messages and summaries are now printed to
1063 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1064 * fixed inference of the upper bound of an iteration variable.
1065 (reported by Rajat Datta).
1066 * fixed bug 211e: calling local functions with duplicated constant
1067 keyword argument no longer causes a bogus style warning about an
1069 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1070 ftype declaration to be disregarded. (reported by Dave Roberts)
1071 * fixed bug 373: caused by erronous compilation of references to alien
1072 variables in the runtime on ppc/darwin.
1073 * fixed bug 376: CONJUGATE type deriver.
1074 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1075 (reported by Baughn on #lisp)
1076 * a cross-compiler bug on non-x86 platforms has been identified and
1077 fixed. (thanks to Bruno Haible)
1078 * improvements to the MIPS runtime code for increased stability.
1079 (thanks to Thiemo Seufer)
1080 * increased the maximimum compact environment size to allow
1081 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1082 * partial workaround for bug 135: don't name the function we're
1083 calling for hairy cases of EVAL, so as not to accumulate one environment
1084 entry per call to EVAL. (reported by Kevin Reid)
1085 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1086 * optimization: type testing for non-vector arrays should be faster.
1087 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1089 * bug fix: setting 31st element of a bit vector to zero did not work
1091 * bug fix: redefining a class definition which failed due to a
1092 previous accessor / function clash now works (but see BUGS entry
1093 #380 for more problems in this area). (thanks to Zach Beane)
1094 * the long form of DEFINE-METHOD-COMBINATION disables method group
1095 checking when given a single method group with pattern *. (thanks
1097 * on x86 compiler supports stack allocation of results of simple
1098 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1099 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1100 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1101 * fixed some bugs related to Unicode integration:
1102 ** the restarts for recovering from input and output encoding
1103 errors only appear when there is in fact such an error to
1105 * fixed some bugs revealed by Paul Dietz' test suite:
1106 ** MISC.549 and similar: late transformation of unsafe type
1107 assertions into derived types caused unexpected code
1109 ** SCALE-FLOAT type deriver is less wrong.
1110 ** type derivers for EXP, LOG and similar functions compute result
1111 types for complex arguments better.
1112 ** (MISC.563) CONJUGATE type deriver works for very restricted
1114 ** out-of-line type testers for character strings are available.
1115 ** EQUAL compiler transform understands specialness of objects
1117 ** accessing double-floats stored on the stack now works on x86-64.
1118 ** debugger internals could sometimes create invalid lispobjs,
1119 resulting in GC crashes.
1120 ** MISC.548: type check weakening can convert required type into
1122 ** initialization forms for bindings are not in scope of free special
1125 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1126 * incompatible change: thread support for non-NPTL systems has
1127 been removed - locking is buggy and unreliable. A threaded
1128 SBCL build will now warn at startup and refuse to create
1129 new threads, unless futex support is detected in the kernel
1130 * incompatible change: the top level REPL now has only an ABORT
1131 restart associated with it, not TOPLEVEL and ABORT as it used to.
1132 TOP and TOPLEVEL are now available as debugger commands for
1133 returning to the top level.
1134 * incompatible change: forms evaluated in the REPL now use the
1135 global optimization policy.
1136 * incompatible change: user- and system-initialization files are
1137 no longer processed with LOAD, but by READ and EVAL; hence the
1138 global optimization policy, startup package, readtable, etc,
1140 * The .fasl file format number has been incremented because of
1141 various incompatible changes.
1142 * internal entry point details and argument counts no longer appear
1143 in backtraces unless explicitly requested by setting
1144 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1145 * built-in and standard functions no longer have names like "top
1146 level local call to FOO".
1147 * fixed bug 32: functions defined in non-null lexical environments
1148 now have more legible printed representation
1149 * fixed bug 33: functions defined in non-null lexical environemnts
1150 are now more amenable to inspection by INSPECT.
1151 * workaround for bug 354: XEPs no longer appear in backtraces unless
1152 explicitly requested.
1153 * fixed bug: receiving the signal which results from attempting to
1154 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1155 SIGBUS on other *BSDs) on architectures where the C stack is also
1156 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1157 to debugger confusion or wild instability. (reported by Cheuksan
1159 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1160 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1161 (reported by Lutz Euler)
1162 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1163 compiled code. (reported by Paul Dietz)
1164 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1165 specialized parameter to a method to something that is not TYPEP
1166 the specializer is now possible.
1167 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1168 face of package deletion.
1169 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1170 pathname. (reported by David Tolpin on c.l.l)
1171 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1172 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1173 STRING-STREAM and FILE-STREAM, respectively.
1174 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1175 than previously; in addition, its test suite is now run on build.
1176 (thanks to Luke Gorrie)
1177 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1178 * optimization: added a immediate representation for single-floats
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1182 ** attempting to create a package with a colliding nickname causes
1183 correctable errors to be signalled.
1184 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1185 ** several x86-64 backend bugs related to sign-extension of immediate
1188 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1189 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1190 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1192 * fixed loading of multiply forward-referenced layouts.
1193 (thanks to Cheuksan Wang)
1194 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1195 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1196 (see COMPILE-FILE documentation for details of :PRINT option)
1197 * fixed bugs 19 and 317: fixed-format floating point printing is
1198 more accurate. This also fixes a bug reported by Adam Warner
1199 related to the ~@F format directive.
1200 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1202 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1203 dispatch table if the from-char is a dispatch macro character.
1204 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1205 by the wordsize no longer ignore the last word. (reported by Lutz
1207 * fixed bug in type checking of dynamic-extent variables. (reported
1209 * optimization: sequence traversal functions use their freedom to
1210 coerce function designators to functions.
1211 * optimization: code with many calls to CLOS methods specialized on
1212 CLOS classes has had redundant type checks removed.
1213 * optimization: added declarations to speed up operations that access
1214 the internal character database (for example STRING-UPCASE)
1215 * optimization: comparison operations between floats and sufficiently small
1216 fixnums no longer create extra rationals
1217 * fixed some bugs related to Unicode integration:
1218 ** portions of multibyte characters at the end of buffers for
1219 character-based file input are correctly transferred to the
1220 start of the buffer at the next read.
1221 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1222 passing it through to OPEN.
1223 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1224 argument given to internal calls to OPEN.
1225 ** the built-in comment readers (introduced by character sequences
1226 ";" and "#|") are more forgiving to encoding errors; they will
1227 STYLE-WARN and then attempt to resync the stream at a character
1228 boundary. (thanks to Teemu Kalvas)
1229 * fixed some bugs in the x86-64 port:
1230 ** Negative short int return values from c-calls are sign-extended
1232 ** The stack is aligned to 16-bytes for c-calls, as required by
1233 the ABI. (thanks to Cheuksan Wang)
1234 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1235 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1236 for enums. (thanks to Vincent Arkesteijn)
1237 ** Multiple small optimizations and bugfixes for floating point
1239 * fixed some bugs revealed by Paul Dietz' test suite:
1240 ** Space, Tab, Linefeed, Return and Page have the invalid
1241 secondary constituent character trait.
1242 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1244 ** WITH-INPUT-FROM-STRING should only update the index place on
1246 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1247 works more reliably.
1248 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1249 with one-dimensional arrays with fill pointers.
1250 ** TYPE-OF failed on a complex with an integer realpart and a
1252 ** compiler failure during type inference for the code of form
1253 (IF (EQL X (THE ...)) ...) (MISC.535).
1255 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1256 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1257 platform. The port supports 61-bit fixnums, large memory spaces
1258 and reloading shared object files.
1259 * enhancement: saving cores with foreign code loaded is now
1260 supported on ppc/Darwin in addition to the previously supported
1262 * enhancement: the statistical profiler now walks deeper into the
1263 call stack for more meaningful call-graphs and accrued time
1264 reports (x86/x86-64 only). It also now reports time spent in
1266 * enhancement: it is now possible to trace most individual methods
1267 of a generic function in addition to tracing the generic function
1269 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1271 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1272 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1273 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1274 produces an error. (thanks to Vincent Arkesteijn)
1275 * bug fix: NAMESTRING on pathnames with :WILD components in their
1276 directories works correctly. (thanks to Artem V. Andreev)
1277 * fixed bug 125: compiler preserves identity of closures. (reported
1279 * bug fixed: functions with &REST arguments sometimes failed with
1280 "Undefined function" when compiled with (DEBUG 3). (reported by
1282 * bug fix: overflow during compiling of setting element of a bit
1283 vector with constant index and value. (reported by Timmy Douglas)
1284 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1285 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1286 * on x86 compiler partially supports stack allocation of dynamic-extent
1288 * GO and RETURN-FROM do not check the extent of their exit points
1289 when compiled with SAFETY 0.
1290 * fixed some bugs related to Unicode integration:
1291 ** encoding and decoding errors are now much more robustly
1292 handled; it should now be possible to recover even from invalid
1293 input or output to the terminal. (thanks to Teemu Kalvas)
1294 ** provided a first cut at implementing STRING-TO-OCTETS and
1295 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1296 ** altered the SB-MD5 contributed module slightly, changing the
1297 interface just enough for it to be supportable for builds where
1298 lisp characters are not eight bits.
1299 * fixed some bugs revealed by Paul Dietz' test suite:
1300 ** the FORMATTER-generated functions for ~V[ conditionals require
1301 the correct number of arguments.
1302 ** READ-FROM-STRING returns the mandated second value when applied
1303 to displaced strings.
1304 ** the #\Rubout and #\Backspace characters are treated as invalid
1305 constituent characters by the tokenizer.
1307 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1308 * new feature: reloading changed shared object files with
1309 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1310 * new feature: references to foreign variables and functions
1311 can now be compiled and loaded before the corresponding shared
1312 object file is loaded, as long as the foreign definitions are
1313 available at runtime.
1314 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1315 supported. (thanks to Dan Debertin)
1316 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1317 just keywords. (thanks to Vincent Arkesteijn)
1318 * fixed bug #331: structure-class instances corresponding to
1319 DEFSTRUCT forms are now created eagerly.
1320 * fixed bug #345: backtraces from calls to undefined functions work
1321 on x86 as well. Related bug #61 is now also partially fixed on x86
1322 (backtraces from throws to unknown catch tags.)
1323 * bug fix: lambda-list parsing is now stricter vrt. order and number
1324 of lambda-list keywords.
1325 * bug fix: as specified by AMOP, an error is signalled if a
1326 class-option appears multiple times in a DEFCLASS form. (reported
1328 * bug fix: RANDOM can be compiled when the compiler derives the type
1329 of its numeric argument as a disjoint set of small integers.
1330 (reported by Paul Dietz)
1331 * bug fix: starting a core saved with shared objects loaded when
1332 those objects are not available no longer causes threaded SBCL to
1333 hang. (reported by Sean Ross)
1334 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1336 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1337 argument, which must always be NIL. (reported by Kalle Niemitalo)
1338 * bug fix: printing 1.0d+23 no longer results in an error.
1339 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1340 * bug fix: structure slot setters preserve evaluation order. (thanks
1342 * fixed some bugs related to Unicode integration:
1343 ** RUN-PROGRAM can allow its child to take input from a Lisp
1344 stream. (reported by Stefan Scholl)
1345 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1346 locales. (reported by Ken Causey)
1347 * fixed some bugs revealed by Paul Dietz' test suite:
1348 ** INCF, DECF and REMF evaluate their place form as specified in
1350 ** FORMATTER expands ~{ ~} iteration directives with V or #
1351 parameters correctly.
1352 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1353 consequent uses no arguments correctly.
1354 ** the system has a partial understanding of the (COMPLEX RATIO)
1356 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1359 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1360 * new feature: a build-time option (controlled by the :SB-UNICODE
1361 keyword feature, enabled by default) for building the system with
1362 support for the entire 21-bit character space defined by the
1364 * new feature: the system now has rudimentary external-format
1365 support; the primary user-visible change at this time is that
1366 characters with the high bit set (such as lower-case-e-acute) will
1367 print correctly to a terminal in a UTF-8 environment.
1368 * minor incompatible change: BASE-CHAR no longer names a class;
1369 however, CHARACTER continues to do so, as required by ANSI.
1370 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1371 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1372 has been moved to the SB-EXT package (temporarily re-exported from
1374 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1375 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1377 * fixed bug #351: better error-handlind and reporting for malformed
1379 * fixed bug #350: bignum-printing is now more memory-efficient,
1380 allowing printing of very large bignums, eg. (expt 2 10000000).
1381 (reported by Bruno Haible)
1382 * fixed bug #302: better primitive-type selection for intersection
1384 * fixed bug #308: non-graphic characters now all have names, as
1385 required. (reported by Bruno Haible)
1386 * bug fix: redefining a class with different superclasses now correctly
1387 removes it from the direct-subclasses of its previous superclasses.
1388 (reported by David Morse)
1389 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1390 new value now works. (reported by Bruno Haible)
1391 * bug fix: correct canonicalization of multiple non-standard slot
1392 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1393 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1394 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1395 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1396 now exists, an signals an error.
1397 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1398 during macroexpansion for non-iteration variables as well. (reported
1399 by Bruno Haible for CMUCL)
1400 * bug fix: Cyclic structures and unprintable objects in compiler
1401 messages no longer cause errors. (reported by Bruno Haible)
1402 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1403 and READ-FROM-STRING all now return a primary value of NIL if
1404 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1405 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1406 (reported by Bruno Haible for CMUCL)
1407 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1408 do not designate a condition. (reported by Bruno Haible for
1410 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1411 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1412 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1413 permitting resolution in favour of any of the conflicting symbols.
1414 (reported by Bruno Haible for CMUCL)
1415 * bug fix: EQUAL compiler optimizations is less aggressive on
1416 strings which can potentially compare true despite having distinct
1417 specialized array element types.
1418 * bug fix: unit enumerations can be defined without dividing by
1419 zero. (thanks to Vincent Arkesteijn)
1420 * FORMAT compile-time argument count checking has been enhanced.
1421 (report from Bruno Haible for CMUCL)
1422 * a partial workaround for the bug 262: the compiler does not try to
1423 inline-expand a local function doing RETURN-FROM from a deleted
1424 BLOCK. (thanks to Peter Denno for the bug report and to David
1425 Wragg for the simple test case)
1426 * fixed some bugs revealed by Paul Dietz' test suite:
1427 ** RENAME-PACKAGE allows all package designators as new package
1429 ** constraint propagation and lambda variable substitution are
1430 more cautious in dealing with partially deleted code.
1431 ** compiler sometimes lost reoptimization passes.
1432 ** CERROR, given a condition as condition designator, passes its
1433 remaining arguments to the continue format control without
1435 ** Case-altering FORMAT directives work correctly on non-ASCII
1437 ** The REAL type specifier handles bounds outside the floating
1438 point ranges without signalling FLOATING-POINT-OVERFLOW.
1439 ** Functions with IR1-transformations can create intercomponent
1440 references to global functions.
1441 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1443 ** FORMAT ~R treats a nil value for its first parameter correctly.
1445 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1446 * enhancement: saving cores with foreign code loaded is now
1447 supported on x86/NetBSD and sparc/Linux in addition to the previously
1448 supported platforms.
1449 * bug fix: on some platforms repeated installations caused multiple
1450 copies of HTML documentation to be installed -- should not happen
1451 any more. (reported by Stefan Scholl)
1452 * bug fix: parsing self-recursive alien record types multiple times
1453 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1454 original patch by Helmut Eller for CMUCL)
1455 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1456 (thanks to Richard Kreuter)
1457 * bug fix: defining classes whose accessors are methods on existing
1458 generic functions in other (locked) packages no longer signals
1459 bogus package lock violations. (reported by François-René Rideau)
1460 * bug fix: special variables as DEFMETHOD parameters no longer have
1461 associated bogus type declarations. (reported by David Wragg and
1463 * bug fix: read-write consistency on streams of element-type
1464 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1465 * bug fix: redefiniton of the only method of a generic function with
1466 no DEFGENERIC no longer emits a full WARNING. In addition,
1467 redefinition of generic functions with no DEFGENERIC to an
1468 incompatible lambda list now signals an error. (thanks to Zach
1470 * bug fix: DEFGENERIC now works even when there's a function of the
1471 same name in an enclosing lexical environment. (thanks to Zach
1473 * fixed compiler failure, caused by instrumenting code during
1474 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1475 * optimization: added loop analysis and improved register allocation
1476 * fixed some bugs revealed by Paul Dietz' test suite:
1477 ** POSITION on displaced vectors with non-zero displacement
1478 returns the right answer.
1479 ** (SIMPLE-STRING) is a valid type specifier for sequence
1481 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1483 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1484 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1486 ** The justification version of the FORMAT ~< directive treats
1487 non-zero minpad parameter correctly.
1489 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1490 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1491 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1492 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1493 the supported interface.
1494 * new feature: Single-stepping of code compiled with DEBUG 2 or
1495 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1496 * new feature: saving cores with foreign code loaded is now
1497 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1498 Timothy Moore's work for CMUCL)
1499 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1500 parameters to * instead of NIL if no initform is supplied.
1501 (reported by Johan Bockgaard)
1502 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1503 correctly when FUNCALL appears as the car of the form. Note:
1504 despite this FUNCALL forms are not currently subject to
1505 compiler-macro expansion. (port of Raymond Toy's fix for the
1506 same from CMUCL, reported by Johan Bockgaard)
1507 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1508 (thanks for Teemu Kalvas)
1509 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1510 case instead of terminating the iteration (thanks for Julian
1511 Squires, Sean Champ and Raymond Toy)
1512 * bug fix: incorrect expansion of defgeneric that caused a style
1513 warning. (thanks for Zach Beane)
1514 * bug fix: slot accessor effective method computation works properly
1515 for classes with multiple non-standard applicable methods on
1516 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1517 * on x86 compiler supports stack allocation of results of LIST and
1518 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1519 CMUCL implementation by Gerd Moellmann)
1520 * fixed some bugs revealed by Paul Dietz' test suite:
1521 ** FORMAT strings with both the ~<~:;~> form of the justification
1522 directive and pretty-printing directives cause an error.
1524 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1525 * incompatible change: the internal functions
1526 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1527 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1528 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1529 instead of the old functions.
1530 * new feature: on platforms where "dladdr" is available foreign
1531 function names now appear in backtraces. (based on Helmut Eller's
1533 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1534 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1536 * documentation: ASDF manual is now distributed with SBCL.
1537 * bug fix: SBCL can now load its contributed modules with REQUIRE
1538 even if the system-provided entries have been removed from
1539 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1541 * bug fix: backtraces involving undefined functions or assembly
1542 routines are more informative. (thanks to Brian Downing)
1543 * bug fix: mutually referent alien structures now work correctly.
1544 (reported by Rick Taube)
1545 * bug fix: structures defined by WITH-ALIEN can be referred to
1546 within other definitions in the same WITH-ALIEN.
1547 * bug fix: division operators (MOD, TRUNCATE and the like) with
1548 constant zero divisors and integer dividends no longer generate
1550 * bug fix: provide default methods for INPUT-STREAM-P and
1551 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1552 * bug fix: improve the detection and resolution of MOP metacycles
1553 (where effective method computation depends on the generic function
1554 having its effective method computed). (reported by Bruno Haible)
1555 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1556 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1557 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1558 * bug fix: pathnames with relative directory components can be
1559 represented relative to default pathnames.
1560 * optimization: in taking the GCD of bignums, reduce the two bignums
1561 to approximately the same size (using Euclid's algorithm) before
1562 applying the more sophisticated binary GCD. (thanks to Juho
1564 * optimization: COUNT on bitvectors now operates word-at-a-time.
1565 * optimization: ASH with a positive, but not necessarily constant,
1566 (leftwards) shift, when in a modular context, is compiled to a
1568 * fixed some bugs revealed by Paul Dietz' test suite:
1569 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1570 if the corresponding argument is NIL.
1571 ** FORMAT directives accepting a minpad parameter treat negative
1573 ** literal commas as character parameters to FORMAT directives are
1575 ** literal spaces directly after ~<Newline> directives within a
1576 format-logical-block (~:< ~@:>) do not induce :FILL-style
1577 conditional newlines.
1578 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1579 as specified: it no longer includes conditional newlines).
1580 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1582 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1583 :PER-LINE-PREFIX argument does not evaluate to a string.
1585 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1586 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1587 the manual for details; for now, package locks can be disabled by
1588 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1589 if no major problems are found then it is likely that they will be
1590 compiled in unconditionally.
1591 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1592 unsupported operators on all platforms. To load a shared library
1593 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1594 object file, link it into a shared library outside of SBCL and
1595 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1597 * bug workaround: ROOM T and the GC were not working together
1598 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1599 reporting facility aren't preserved. That reporting has been
1600 disabled, so now until and unless someone figures out how to make it
1601 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1602 * minor incompatible change: as threatened around sbcl-0.8.0, the
1603 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1604 The symbols are also exported from SB-PCL for backwards
1605 compatibility, but more so than before SB-PCL should be treated as
1606 an implementation-internal package.
1607 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1609 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1610 It is known as of this release not to work on the Alpha, however.
1611 * fixed bug #167: errors signalled due to illegal syntax in method
1612 bodies are now more legible.
1613 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1614 designators and can hence be used with TYPEP.
1615 * fixed bug #333: CHECK-TYPE now ensures that the type error
1616 signalled, if any, has the right object to be accessed by
1617 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1618 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1619 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1621 * fixed bug #334: programmatic addition of slots using specialized
1622 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1623 and :ALLOCATION :CLASS slots.
1624 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1625 float exponents. (rereported by Peter Seibel)
1626 * fixed a bug: #\Space (and other whitespace characters) are no
1627 longer considered to be macro characters in standard syntax by
1628 GET-MACRO-CHARACTER.
1629 * fixed bug: initialization of condition class metaobjects no longer
1630 causes an instance of the condition to be created. (reported by
1632 * fixed bug: it is now possible to have slots such that
1633 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1634 description is neither :INSTANCE nor :CLASS.
1635 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1636 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1637 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1638 non-standard SLOT-DEFINITION-ALLOCATION values.
1639 * fixed bugs: various race conditions handling exiting threads.
1640 CL-PPCRE's thread torture test now passes.
1641 * fixed arguably-a-bug: GC time proportional to number of threads in
1642 system even when most of them are idle
1643 * optimization: improved performance of BIT and SBIT on bit-vectors.
1644 * .fasl file incompatibility: The fasl file version number has
1645 been incremented because of changes associated with package locks.
1647 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1648 * minor incompatible change: the system no longer provides
1649 optimization hints (conditions which are TYPEP
1650 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1651 mode; these hints are still emitted when higher SPEED optimization
1653 * new contrib module: a sampling profiler (profiling by statistical
1654 sampling, rather than by instrumenting functions) is available as
1655 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1656 * the behaviour of the standard function ED is now customizeable by
1657 third parties through a hook variable: see ED's documentation
1658 string for information on the protocol.
1659 * the compiler no longer emits efficiency notes for (FUNCALL X)
1660 when the type of X is uncertain under default optimization
1662 * fixed bug 276: mutating a binding of a specialized parameter to a
1663 method to something that is not TYPEP the specializer is now
1665 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1666 exhibit the required behaviour on the x86 platform. (thanks to
1667 Peter van Eynde, Eric Marsden and Bruno Haible)
1668 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1669 even for difficult arguments. (reported by Peter Graves)
1670 * fixed bug 141a: the backquote printer now descends quoted
1672 * fixed another bug in backquote printing: no more destructive
1673 modification of the form's list structure. (reported by Brian
1675 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1676 move between its address being taken and the call to
1677 interrupt_thread, fixing a crashing race condition.
1678 * the SB-POSIX contrib implementation has been adjusted so that it
1679 no longer exhibits ridiculously poor performance when constructing
1680 instances corresponding to C structs.
1682 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1683 * minor incompatible change: the sb-grovel contrib now treats C
1684 structures as alien (in the sense of SB-ALIEN) objects rather than
1685 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1686 has implications for memory management of client code
1687 (sb-grovel-returned objects must now be manually managed) and for
1688 type safety (alien objects now have full types).
1689 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1690 used to control emission of compiler diagnostics, rather than the
1691 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1692 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1693 quality should be considered deprecated.
1694 * install.sh now installs the user manual as well
1695 * (not quite a new documentable feature, but worth considering in
1696 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1697 beginnings of a semantically meaningful condition hierarchy is
1698 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1700 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1701 designator as the defaults argument.
1702 * fixed bug: Displaced arrays whose displaced-to array has become
1703 too small now cause ARRAY-DIMENSION to signal an error, providing
1704 for safer bounds-checking. (reported by Bruno Haible)
1705 * fixed bug: DEFCLASS slot definitions with identical :READER and
1706 :WRITER names now signal a reasonable error. (reported by Thomas
1708 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1710 * fixed bug: Invalid slot specification errors now print correctly.
1711 (thanks to Zach Beane)
1712 * fixed bug 320: Shared to local slot value transfers in class
1713 redefinitions now happen corrently from superclasses as
1714 well. (reported by Bruno Haible)
1715 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1717 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1718 improper lists correctly. (reported by Bruno Haible)
1719 * fixed bug 313: source-transform for <fun-name> was erroneously
1720 applied to a call of a value of a variable with name <fun-name>.
1721 (reported by Antonio Menezes Leitao)
1722 * fixed bug 307: The obsolete instance protocol ensures that
1723 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1724 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1725 remove dead unknown-values globs from the middle of the stack.
1726 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1727 easier to resume long computations after using *BREAK-ON-SIGNALS*
1728 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1729 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1730 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1731 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1732 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1733 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1735 * fixed a (fairly theoretical) bug in string printing: if
1736 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1737 does not have array-element-type equal to the most general string
1739 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1740 function name. (thanks to Nikodemus Siivola)
1741 * fixed bug: compile-time format string checker failed on
1742 non-closed ~{. (reported by Thomas F Burdick)
1743 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1744 don't warn on legal loop code involving more than one aggregate
1745 boolean. (thanks to Nikodemus Siivola)
1746 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1747 in keyword position to MAKE-PACKAGE was not being reported as
1749 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1750 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1752 * fixed bugs 280 and 312: the checking for multiple definitions in a
1753 file is less likely to become confused by uses of inline
1755 * fixed bug: the #S reader macro performs the keyword coercion
1756 specified for slot names. (reported by Kalle Niemitalo)
1757 * fixed bug: lambda lists may contain symbols whose names start with
1758 & but are not lambda-list-keywords; their occurrence triggers a
1760 * fixed bug 321: define-method-combination argument lambda lists do
1761 not cause invalid code to be generated when &optional or &aux
1762 variables have default values. (reported by Bruno Haible)
1763 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1764 have CLOS classes; this appears also to have fixed reported
1765 instability in stack exhaustion detection.
1766 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1767 MUFFLE-WARNING when no associated restart is present is now
1769 * optimization: rearranged the expansion of various defining macros
1770 so that each expands into only one top-level form in a
1771 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1773 * optimization: used a previously little-used slot in symbols to
1774 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1776 * fixed some bugs revealed by Paul Dietz' test suite:
1777 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1779 ** Fixed an optimization bug related to inheritance of initforms
1780 from local to shared slots.
1781 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1782 ** CAST optimizer forgot to flush argument derived type.
1783 ** print/read consistency on floats is now orders of magnitude
1784 more likely. (thanks also to Bruno Haible for a similar report
1786 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1787 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1788 for objects of type REAL. Make it so.
1789 ** SXHASH is specified (infelicitously) to respect similarity,
1790 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1791 it so. (thanks to Markus Ziegler)
1792 ** on the Alpha, the compiler succeeds in compiling functions
1793 returning a known number of arguments greater than 63.
1794 ** fixed handling of invalid NIL arguments in keyword position
1796 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1798 ** provided a readably-printable representation for RANDOM-STATE
1800 ** ensured that pathnames, where they have a namestring, always
1801 print using #P"..." syntax.
1803 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1804 * Support for the forthcoming 2.0 version of the NetBSD kernel
1805 running on x86 hardware has been added. (thanks to Perry
1806 E. Metzger most immediately, and others for their past work)
1807 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1808 current release version; SBCL's OpenBSD support had been broken
1809 since about the time of OpenBSD's switch to ELF binary format.)
1810 * [placeholder for DX summary]
1811 ** user code with &REST lists declared dynamic-extent, under high
1812 speed or space and low safety and debug optimization policy.
1813 * The manual has been converted to Texinfo format and the debugger
1814 chapter from the cmucl manual has been added.
1815 * A facility has been added to extract documentation strings from
1816 sbcl and store them as Texinfo-formatted snippets for inclusion in
1817 the manual (via Texinfo's @include directive)
1818 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1819 the test case to Dave Roberts)
1820 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1821 pointers. (reported by Sean Ross)
1822 * bug fix: PROFILE output is printed nicely even for large numerical
1823 values. (thanks to Zach Beane)
1824 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1825 greater than 32 handle EOF correctly.
1826 * bug fix: on X86 an immediate argument of the IMUL instruction is
1827 correctly printed in disassembly. (thanks to Lutz Euler)
1828 * bug fix: class slots in redefined classes preserve their old
1829 values. (thanks to Bruno Haible and Nikodemus Siivola)
1830 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1831 succeeds. (reported by Marco Baringer)
1832 * fixed some bugs revealed by Paul Dietz' test suite:
1833 ** READ-SEQUENCE now works on ECHO-STREAMs.
1834 ** RATIONALIZE works more according to its specification. (thanks
1837 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1838 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1839 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1840 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1841 should matter to you only if you rebind the printer control
1842 variables and then find you want different bindings in the
1843 debugger than in the ordinary execution of your program.)
1844 * The runtime build system has been tweaked to support building
1845 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1846 assembler and linker. (thanks to Nikodemus Siivola)
1847 * Unbound, undefined, undeclared variables now trigger full
1848 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1849 more often programmer error than deliberate exploitation of undefined
1851 * optimization: the hash algorithm for strings has changed to one
1852 that is less vulnerable to spurious collisions. (thanks to Juho
1854 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1855 needless bounds checking. (thanks to Juho Snellman)
1856 * optimization: implemented multiplication as a modular
1857 (UNSIGNED-BYTE 32) operation on the PPC backend.
1858 * fixed some bugs revealed by Paul Dietz' test suite:
1859 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1861 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1862 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1863 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1865 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1867 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1869 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1870 their output stream on EOF from read.
1871 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1872 have been read to end-of-file.
1873 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1875 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1876 description of determination of which consecutive characters
1878 ** Printing the "Space" character with escaping on now yields "#\\ ",
1879 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1880 ** Reading floating-point numbers with *READ-BASE* set to a number
1881 less than 10 works correctly.
1882 ** Reading floating-point numbers with *READ-BASE* set to a number
1883 more than 10 works correctly.
1884 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1885 the readtable currently in effect.
1887 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1888 * minor incompatible change: parsing of namestrings on a physical
1889 (Unix) host has changed; numbers after the final #\. in a
1890 namestring are no longer interpreted as a version field. This is
1891 intented to be largely invisible to the user, except that the
1892 meaning of the namestring "*.*.*" has changed: it now refers to a
1893 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1894 should usually be replaced by
1895 (make-pathname :name :wild :type :wild :version :wild)
1896 with the added benefit that this is more likely to be portable.
1897 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1898 option to OPEN now signals an error if the file being opened
1899 exists; this may have an impact on existing code.
1900 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1901 signals received, on the PowerPC platforms (both Linux and
1902 Darwin). (thanks to Pierre Mai for pointing out the location of
1904 * several fixes on OS X: The system now builds and runs cleanly on
1905 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1906 (thanks to Brian Mastenbrook)
1907 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1908 second-resolution: integer multiples of 1/3600 between -24 and 24.
1909 (thanks to Vincent Arkesteijn)
1910 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1911 type when called with 1 argument; PEEK-CHAR checked type of
1912 PEEK-TYPE only after having read first character from a
1913 stream. (reported by Peter Graves)
1914 * bug fix: the garbage collector now has much better locality
1915 behaviour, and in particular no longer treats all memory as being
1916 exhausted when many small objects point to each other in a deeply
1918 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1919 recognized as being TYPEP their class.
1920 * bug fix: the PUSHNEW documentation string has been corrected.
1921 (thanks to Vincent Arkesteijn)
1922 * bug fix: defaulting of the value for the last of an atypically
1923 large number of multiple values being bound was not being
1924 performed correctly on the Alpha or PPC platforms
1925 * optimization: implemented multiplication as a modular
1926 (UNSIGNED-BYTE 32) operation on the x86 backend.
1927 * optimization: SEARCH on simple-base-strings can now be open-coded.
1928 (see also contrib/compiler-extras.lisp for inspiration for
1929 teaching the compiler about the Boyer-Moore algorithm).
1930 * value, returned by MAX (and MIN) called with several EQUALP, but
1931 not EQL, arguments now does not depend on compiler settings.
1932 * fixed some bugs revealed by Paul Dietz' test suite:
1933 ** in stack analysis liveness information is propagated from
1934 non-local entry points.
1935 ** pathwise CAST removing failed when the CAST node did not start
1937 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1938 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1940 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1941 host is already defined.
1942 ** RENAME-FILE works on streams instead of signalling an internal
1944 ** PEEK-CHAR uses the current readtable when determining whether
1945 or not a character is whitespace.
1946 ** MERGE-PATHNAMES handles the case when the pathname does not
1947 specify a name while the default-pathname specifies a version.
1948 ** Pathnames now stand a better chance of respecting print/read
1950 ** Attempting to use standardized file system operators with a
1951 pathname with invalid :DIRECTORY components signals a
1953 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1955 ** DIRECTORY on logical pathnames is more correct.
1956 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1957 signal a TYPE-ERROR if their argument is not a stream.
1958 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1959 designator argument does not designate a stream.
1960 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1961 examining the synonym.
1962 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1964 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1965 element-type larger than ([UN]SIGNED-BYTE 32).
1967 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1968 * When built with the :SB-FUTEX feature, threaded builds now take
1969 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1970 for faster/more reliable mutex and condition variable support.
1971 * Incompatible change (but one you probably shouldn't have been using
1972 anyway): the interface and code for arbitrating between multiple
1973 threads in the same user session has been redesigned.
1974 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1975 error when called without an explicit environment argument.
1976 (thanks to Nikodemus Siivola)
1977 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1978 confused about their position. (thanks to Adam Warner and Gerd
1980 * bug fix: Pretty printing backquoted forms with unquotations in the
1981 argument list position of various code constructs such as LAMBDA
1982 now works correctly. (reported by Paul Dietz)
1983 * bug fix: Pretty printing unquotations no longer loses all
1984 stream position information.
1985 * optimization: performance of string output streams is now less
1986 poor for multiple small sequence writes.
1987 * optimization: performance of CSUBTYPEP in the presence of complex
1988 expressions involving CONS and NOT many times has been improved.
1989 (reported by Paul Dietz)
1990 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1992 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1993 streams for the REPL, for the debugger, and so on. (thanks to
1995 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1996 * fixed some bugs revealed by Paul Dietz' test suite:
1997 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1998 is ignored if it is too small, rather than propagating through
1999 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2000 ** extremely complex negations of CONS types were not being
2001 sufficiently canonicalized, leading to inconsistencies in
2003 ** VALUES tranformer lost derived type.
2005 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2006 * fixed a bootstrapping bug: the build process no longer assumes
2007 that the various BOOLE-related constants have the same value in
2008 host and target lisps. (noted by Paul Dietz' test suite on an
2009 SBCL binary built from CLISP)
2010 * The system can now be dynamically linked on the MIPS platform,
2011 which enables dynamic loading of foreign code from Lisp. (thanks
2012 to Ralf Baechle for discussions on the MIPS ABI)
2013 * The system now records debugging information for its own source
2014 files in a filesystem-position-independent manner, relative to
2015 the "SYS" logical host.
2016 * fixed a compiler bug: MV-LET convertion did not check references
2017 to the "max args" entry point. (reported by Brian Downing)
2018 * tweaked disassembly notes to be less confident about proclaiming
2019 some instruction as an LRA. (thanks to Brian Downing)
2020 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2021 now each have their own history, command character, and other
2022 characteristics. (thanks to David Lichteblau)
2023 * fixed some bugs revealed by Paul Dietz' test suite:
2024 ** compiler failure in compiling LOGAND expressions including a
2026 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2027 shift greater than 32.
2028 ** FLUSH-DEST did not mark blocks for type check regeneration.
2029 ** HANDLER-CASE failed to accept declarations in handler clauses
2030 in some circumstances.
2032 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2033 * New code in contrib/sb-introspect (still probably not entirely
2034 stable yet) provides some support for smart Lisp development
2035 environments like SLIME.
2036 * The conditions signalled for errors occurring when loading .fasl
2037 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2038 in a way which should help ASDF recover gracefully.
2039 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2040 stuff has been cleaned up. If you code contrib/ stuff, this might
2041 affect you, and you can look at contrib/README, contrib/STANDARDS,
2042 and/or the 0.8.4.27 diff to check.
2043 * In full calls the compiler now does not generate checks for declared
2044 argument types for all arguments.
2045 * various threading fixes
2046 ** and some experimental patches which didn't make it into
2047 the main tree for this release, but which are shipped in
2048 contrib/experimental-thread.patch as a possible fix for some
2049 failures (deadlock, spinning...) in GC-intensive multithreaded
2051 * fixed PPC build problem (source code incompatibility of different
2052 library versions): added offsetof() hackery which attempts to divine
2053 where glibc maintainers put uc_mcontext today
2054 * fixed bug 282: compiler does not trust type assertions while passing
2055 arguments to a full call.
2056 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2057 &OPTIONAL VALUES type parameter.
2058 * fix bug 214: algorithm for noting rejected templates is now more
2059 similar to that of template seletion. (also reported by rydis on
2061 * fixed bug 141b: printing backquoted information readably and prettily
2062 inserts a space where necessary.
2063 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2064 (CHARACTER 10) are now reported as errors, rather than propagated
2065 as unknown types. (reported by piso on #lisp)
2066 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2067 with values NIL and :ERROR. (thanks to Milan Zamazal)
2068 * fixed bug 191c: CLOS now does proper keyword argument checking as
2069 described in CLHS 7.6.5 and 7.6.5.1.
2070 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2071 longer raise an error; further, using a list as a for-as-arithmetic
2072 counter now raises a meaningful error.
2073 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2074 now understood by sequence creation functions such as MAKE-SEQUENCE
2076 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2077 a STRING-INPUT-STREAM.
2078 * compiler enhancement: SIGNUM is now better able to derive the type
2080 * type declarations inside WITH-SLOTS are checked. (reported by
2082 * fixed some bugs revealed by Paul Dietz' test suite:
2083 ** incorrect optimization of TRUNCATE for a positive first
2084 argument and negative second.
2085 ** compiler failure in let-convertion during flushing dead code.
2086 ** compiler failure while deriving type of TRUNCATE on an
2087 interval, containing 0.
2088 ** ASH of a negative bignum by a negative bignum count now returns
2090 ** intersection of CONS types now canonicalizes properly, fixing
2091 inconsistencies in SUBTYPEP.
2093 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2094 * incompatible change: The --disable-debugger command line
2095 option now clobbers the debugger at a more fundamental
2096 level, by redefining #'INVOKE-DEBUGGER instead of by
2097 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2098 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2099 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2100 enter the debugger REPL and then suffer recursive errors
2101 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2102 while under the new system, BREAK in --disable-debugger mode
2103 terminates the system just as an unhandled error would.
2104 * fixed compiler performance when processing loops with a step >1;
2105 * bug fix: DOCUMENTATION now retrieves generic function
2106 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2107 support has been systematized, and now supports the methods
2108 specified by ANSI, along with a default method and a method for
2109 slot documentation. (reported by Nathan Froyd)
2110 * bug fix: effective methods associated with a generic function are
2111 no longer cached over a change of that generic function's method
2112 combination. (reported by Andreas Fuchs)
2113 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2114 names a non-existent directory. (thanks to Andreas Fuchs)
2115 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2116 the PPC platform, lead to stack corruption; this has been fixed.
2117 (reported by Rainer Joswig)
2118 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2119 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2120 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2121 arithmetic in general, is now much more reliable on the PPC
2123 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2124 platform now returns the right answer.
2125 * optimization: restored some effective method precomputation in
2126 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2127 precomputation is now tunable.
2128 * optimization: compiler-internal data structure use has been
2129 reviewed, and changes have been made that should improve the
2130 performance of the compiler by about 20%.
2131 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2132 simple-base-strings and simple-bit-vectors is improved.
2133 * optimization: the optimization of 32-bit logical and arithmetic
2134 functions introduced in version 0.8.3 on the x86 has been
2135 implemented on the mips, ppc and sparc platforms; an
2136 implementation of the same facility, but for 64-bit arithmetic,
2137 has been added for the alpha.
2138 * microoptimization: the compiler is better able to make use of the
2139 x86 LEA instruction for multiplication by constants.
2140 * bug fix: in some situations compiler did not report usage of
2141 generic arithmetic in (SPEED 3) policy.
2142 * bug 145b fix: compiler used wrong type specifier while converting
2143 MEMBER-types to numeric.
2144 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2146 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2147 index argument now works properly on the Alpha platform.
2148 * bug fix: floating point exception treatment on the Alpha platform
2150 * bug fix: FILE-POSITION works much better on string input and
2151 output streams. (thanks to Nikodemus Siivola)
2152 * bug fix: many threading/garbage collection symptoms sorted.
2153 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2154 might be pseudo-atomic.
2155 * internal change: Stopping for GC is now done with signals not
2156 ptrace. GC is now done in whichever thread wanted it, instead of
2158 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2160 * bug fix: result form in DO is not contained in the implicit
2162 * incompatible change: ICR structure is changed; the value part of
2163 CONTINUATION is now called LVAR; corresponding functions are
2164 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2165 * added type deriver for ISQRT (thanks to Robert E. Brown).
2166 * bug fix: better support for loading from the command line when an
2167 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2169 * fixed some bugs revealed by Paul Dietz' test suite:
2170 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2171 ** ROUND and FROUND now give the right answer when given very
2172 small float arguments.
2173 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2175 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2176 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2177 ** LCM should return a non-negative integer.
2178 ** PARSE-INTEGER returned the index of a terminator instead of the
2179 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2180 ** PARSE-INTEGER returned an incorrect index being applied to a
2182 ** LCM with two arguments of 0 returns 0 rather than signalling
2184 ** unsigned addition of a 32-bit constant with the high bit set no
2185 longer causes an internal compiler error on the x86.
2186 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2187 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2188 with negative last argument.
2189 ** byte specifiers with zero size and position no longer cause
2190 an error during type derivation.
2191 ** bignum multiplication on the Alpha platform now returns the
2193 * porting: The system now builds on SuSE AMD64, although it still
2194 generates a 32-bit binary.
2195 * .fasl file incompatibility: The fasl file version number has
2196 been incremented (because of the changes to internal compiler
2197 data structures referred to above).
2199 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2200 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2201 more accurately, on the Darwin kernel running on PowerPC hardware.
2202 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2203 * Compiler code deletion notes now signal a condition of type
2204 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2205 an associated MUFFLE-WARNING restart.
2206 * The compiler now performs limited argument count validation of
2207 constant format strings in FORMAT, and where appropriate in ERROR,
2208 CERROR and WARN. (thanks to Gerd Moellmann)
2209 * New ASDF-INSTALL contrib can be used for automatic download and
2210 installation of third-party Lisp code from CCLAN or other sites
2212 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2213 which forces another thread to execute a function supplied by the
2215 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2216 now accept and act upon their :ELEMENT-TYPE keyword argument.
2217 (reported by Martin Atzmueller, Edi Weitz)
2218 * bug fix: FILE-POSITION now accepts position designators up to
2219 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2220 is the greater. (thanks to Patrik Nordebo)
2221 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2222 Kalle Olavi Niemitalo)
2223 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2224 printable. (reported by Eric Marsden)
2225 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2226 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2227 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2228 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2229 (thanks to Stig Sandoe)
2230 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2231 UNREAD-CHAR now correctly includes the unread character in the
2232 target sequence. (thanks to Gerd Moellmann)
2233 * bug fix in threaded builds: the system can now be suspended and
2234 resumed by shell job control with minimal disruption.
2235 * bug fixes in times and timezones >2038AD
2236 * better handling of "where is GNU make?" problem in build scripts
2237 (thanks to Nikodemus Siivola)
2238 * new optimization: inside a named function any reference to a
2239 function with the same name is considered to be a self-reference;
2240 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2241 optimization quality.
2242 * new optimization on x86: logical functions and + now have
2243 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2244 used when the result is truncated to 32 bits.
2245 * VALUES declaration is partially enabled.
2246 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2247 * fixed some bugs revealed by Paul Dietz' test suite:
2248 ** The system now obeys the constraint imposed by
2249 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2250 types form a lattice under type intersection.
2251 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2252 ** ASSOC now ignores NIL elements in an alist.
2253 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2254 and (1+ MOST-POSITIVE-FIXNUM) answers.
2255 ** The addition of a method with invalid qualifiers to a generic
2256 function does not cause an error to be signalled immediately;
2257 a warning is signalled, and the error is generated only on
2258 calling the generic function.
2259 * changed .fasl file version number, in part to add type codes for
2260 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2261 obscure ANSI requirements
2263 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2264 * fixed bug 148: failure to inline-expand a local function left
2265 garbage, confusing the compiler.
2266 * fixed bugs 3cd: structure slot readers perform type check if the
2267 slot can have an invalid value (i.e. it is either not initialized
2268 or can be written with a less specific slot writer).
2269 * bug fix: the compiler now traps array references to elements off
2270 the end of an array; previously, the bounds checking in some
2271 circumstances could go off-by-one.
2272 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2274 * type declarations for array element types now obey the description
2275 on the CLHS page "Declaration TYPE", as per discussions on
2276 sbcl-help around 2003-05-08. This means that a declaration
2277 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2278 declaration, all references to BAR will be asserted or assumed
2279 (with THE, so dependent on compiler policy) to involve objects of
2280 type FOO. Note that no such declaration is implied in
2281 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2282 * declared types of functions from the "Conditions"
2283 chapter. (reported by Paul Dietz)
2284 * bug fix: CERROR accepts a function as its first argument.
2285 * bug fix: NTH an NTHCDR accept a bignum as index
2286 arguments. (reported by Adam Warner)
2287 * optimization: character compare routines now optimize comparing
2288 against a constant character. (reported by Gilbert Baumann)
2289 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2290 argument now works properly on the MIPS platform.
2291 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2292 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2293 declared to have a functional type, does not cause an error
2295 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2296 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2297 * fixed bug: if last continuation of a deleted block has a
2298 destination, this destination should be deleted too. (reported by
2300 * fixed a bug in the bootstrap process: the host compiler's values
2301 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2302 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2303 test case from Patrik Nordebo)
2304 * improved the ability of the disassembler on the PPC platform to
2305 provide helpful disassembly notes.
2306 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2307 the class in more cases than previously.
2308 * bug fix: FILE-POSITION now understands :START and :END for
2309 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2310 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2311 8) data. (thanks to David Lichteblau)
2312 * bug fix: it is possible to add a method to a generic function
2313 without lambda list.
2314 * bug fix: reader failed to signal END-OF-FILE inside an
2315 object representation. (reported by Nikodemus Siivola)
2316 * fixed some bugs revealed by Paul Dietz' test suite:
2317 ** LAST and [N]BUTLAST should accept a bignum.
2318 ** condition slot accessors are methods.
2319 ** (VECTOR NIL) is a subtype of STRING.
2321 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2322 * minor incompatible change: some nonsensical specialized lambda
2323 lists (used in DEFMETHOD) which were previously ignored now signal
2325 * minor incompatible change: the system is now aware of the types of
2326 variables in the COMMON-LISP package, and will signal errors for
2327 most violations of these type constraints (where previously they
2328 were silently accepted).
2329 * minor incompatible change: COMPILE-FILE now uses the freedom
2330 afforded (ANSI 3.2.2.3) to use derived function types for
2331 functions defined in the same file. This also permits the system
2332 to warn on static type mismatches and function
2333 redefinition. (Currently it does not work with high DEBUG level.)
2334 * minor incompatible change: VALUES declaration is disabled.
2335 * When issuing notes, the compiler now signals a condition of type
2336 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2337 restart for use in user handlers. It is expected that the
2338 COMPILER-NOTE condition will eventually become a condition
2339 supertype to a hierarchy of note types, which will then be
2340 handleable in a similar fashion. However, at the moment, no such
2341 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2342 but it's an implementation detail, not a classification for the
2344 * Changes in type checking closed the following bugs:
2345 ** type checking of unused values (192b, 194d, 203);
2346 ** template selection based on unsafe type assertions (192c, 236);
2347 ** type checking in branches (194bc).
2348 * A short form of VALUES type specifier has ANSI meaning (it has
2349 increased the number of situations when SBCL cannot perform type
2351 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2352 keywords or constants is permissible.
2353 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2354 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2355 classes. (thanks to Antonio Martinez)
2356 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2357 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2358 argument to a pretty-print stream. Instead, it leaves any such
2359 operations to DESCRIBE-OBJECT methods.
2360 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2362 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2363 lambda lists. (thanks to David Lichteblau)
2364 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2365 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2366 (thanks to Gerd Moellmann)
2367 * fixed bug 246: increased compilation speed of long
2368 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2370 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2372 * DEFCONSTANT now throws a condition of type
2373 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2374 constant to a non-EQL value; CONTINUE and ABORT restarts
2375 respectively change and preserve the value.
2376 * fixed bug 63: The code walker, part of the implementation of CLOS,
2377 is now better at handling symbol macros.
2378 * bug fix: There is no longer an internal implementation type named
2379 CL:LENGTH. (reported by Raymond Toy)
2380 * bug fix: In macro-like defining macros/special operators the
2381 implicit block does not enclose the lambda list.
2382 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2383 enclosed in parentheses) are not suitable as type specifiers, and
2384 their use properly signals an error now.
2385 * bug fix: An argument count mismatch for a type specifier in code
2386 being compiled no longer causes an unhandled error at compile
2387 time, but signals a compile-time warning.
2388 * fixed simple vector readable printing
2389 * bug fix: DESCRIBE takes more care over whether the class
2390 precedence list slot of a class is bound before accessing it.
2391 (reported by Markus Krummenacker)
2392 * bug fix: FORMATTER can successfully compile pretty-printer format
2393 strings which use variants of the ~* directive inside.
2394 * bug fix: SEARCH now applies its TEST predicate to the elements of
2395 the arguments in the correct order. (thanks to Wolfhard Buss)
2396 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2397 in type inference. (thanks to Robert E. Brown)
2398 * bug fix: Reading in symbols with an explicit package name of ""
2399 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2400 * fixed some bugs revealed by Paul Dietz' test suite:
2401 ** NIL is now allowed as a structure slot name.
2402 ** Arbitrary numbers, not just REALs, are allowed in certain
2403 circumstances in LOOP for-as-arithmetic clauses.
2404 ** Multiple class redefinitions before slot access no longer
2405 causes a type error.
2406 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2407 association between the name and a class.
2408 ** Generic functions with non-standard method-combination and over
2409 five methods all of which return constants no longer return NIL
2410 after the first few invocations. (thanks to Gerd Moellmann)
2411 ** CALL-NEXT-METHOD with no arguments now passes the original
2412 values of the arguments, even in the presence of assignment.
2413 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2415 ** DISASSEMBLE works with closures and funcallable instances.
2416 ** ADD-METHOD now returns the generic function, not the new method.
2417 ** FIND-METHOD signals an error if the lengths of the specializers
2418 is incompatible with the generic function, even if the ERRORP
2420 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2421 which its argument is a member.
2422 ** DEFCLASS only redefines the class named by its class-name
2423 argument if that name is the proper name of the class;
2424 otherwise, it creates a new class.
2425 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2426 of the UNBOUND-SLOT condition to the name of the slot.
2427 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2428 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2429 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2430 SLOT-MAKUNBOUND in the specified fashion.
2432 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2433 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2434 cross-compilation host. As a consequence, we can now bootstrap our
2435 way up to SBCL starting with a bare gcc toolchain and human-readable
2436 source code (first the source to CLISP, then the source to SBCL).
2437 * A contributed module containing a partial implementation of the
2438 simple-streams interface has been included. (thanks to Rudi
2440 * A contributed module implementing the RFC1321 Message Digest
2441 Algorithm, known as MD5, has been included.
2442 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2443 no longer has any effect, as the code controlled by this feature
2444 has been deleted. (As far as we know, no-one has ever built using
2445 this feature, and its semantics were confused in any case).
2446 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2447 work on conditions (as required by the ANSI specification),
2448 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2449 expected behaviour on conditions. Users should note, however,
2450 that such behaviour is not required by the ANSI specification,
2451 and so use of this behaviour may render their code unportable.
2452 * fixed some bugs revealed by Paul Dietz' test suite:
2453 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2455 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2456 ** SXHASH on condition objects no longer returns NIL.
2457 ** :ALLOCATION :CLASS slots are better treated; their values are
2458 updated on class redefinition, and initforms inherited from
2459 superclasses are applied.
2460 ** REMOVE-METHOD returns its generic function argument even when
2461 no method was removed.
2462 ** SHARED-INITIALIZE now initializes the values of the requested
2463 slots, including those with :ALLOCATION :CLASS.
2464 ** ALLOCATE-INSTANCE now works on structure classes defined via
2465 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2467 ** SLOT-EXISTS-P now works on conditions, as well as structures
2469 ** MAKE-LOAD-FORM now has the required methods on
2470 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2471 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2472 keyword, and now implements the SLOT-NAMES argument.
2473 ** methods with &OPTIONAL arguments no longer allow too many
2474 arguments to be passed in the call without error.
2475 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2476 option is consistent with the required arguments of the generic
2477 function lambda list.
2478 * bug fix: REQUIRE accepts a string designator. (Thanks to
2480 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2481 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2482 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2483 * bug fix: The compiler now checks for duplicated variables in macro
2485 * bug fix: SETQ on globals returns the correct value.
2486 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2487 not a COMPILER-ERROR (followed by some other strange error on
2488 choosing the CONTINUE restart).
2489 * bug fix: make.sh and friends are now more consistent in the way that
2490 they look for GNU "make".
2492 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2493 * experimental native threads support (on x86 Linux >=2.4 only).
2494 This is not compiled in by default: you need to add :SB-THREAD to
2495 the target features. See the "Beyond ANSI" chapter of the manual
2497 * fix for longstanding nonANSIism: The old distinction between
2498 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2499 The return value from CL:FIND-CLASS is now a CLOS class, and
2500 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2501 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2502 * An interface to the MetaObject Protocol, as described in Kiczales,
2503 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2504 MIT Press, 1991, is available from the SB-MOP package.
2505 * incompatible change: the SB-PCL package should now be considered
2506 a private implementation detail, and no longer a semi-private MOP
2508 * minor incompatible change: due to rearrangement for threads, the
2509 control stack and binding stack are now allocated at arbitrary
2510 addresses instead of being hardcoded per-port. Users affected by
2511 this probably have to be doing advanced things with shared
2512 libraries, and will know who they are.
2513 * minor incompatible change: Previously, all --eval forms used were
2514 processed with READ before any of them were processed with EVAL.
2515 Now each --eval form is processed with both READ and EVAL before
2516 the next --eval form is processed. (Thus package operations like
2517 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2518 work as the user might reasonably expect.)
2519 * minor incompatible change: *STANDARD-INPUT* is now only an
2520 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2522 * minor incompatible change: Y-OR-N-P is now character-oriented,
2523 not line oriented. Also, YES-OR-NO-P now works without errors.
2524 (thanks to Antonio Martinez)
2525 * sb-aclrepl module improvements: an integrated inspector, added
2526 repl features, and a bug fix to :trace command.
2527 * Known functions, which cannot be open coded by the backend, are
2528 considered to be able to check types of their arguments. (fixing
2529 a bug report by Nathan J. Froyd)
2530 * fixed a bug in computing method discriminating functions: It is
2531 now possible to define methods specialized on classes which have
2532 forward-referenced superclasses. (thanks to Gerd Moellmann)
2533 * fixed evaluation order in optional entries (reported by Gilbert
2535 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2536 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2537 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2538 required-by-AMOP NAME argument, as well as CLASS and
2539 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2540 * fixed bug 20: DEFMETHOD can define methods using names that are
2541 not the proper names of classes to designate class specializers.
2542 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2543 Unix file descriptors, instead of blowing up. (thanks to Antonio
2545 * Garbage collection refactoring: user-visible change is that a
2546 call to the GC function during WITHOUT-GCING will not do garbage
2547 collection until the end of the WITHOUT-GCING. If you were doing
2548 this you were probably losing anyway.
2549 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2550 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2551 TYPEP the latter but not the former.
2552 * The compiler issues a full WARNING for calls to undefined functions
2553 with names from the CL package.
2554 * MAP-INTO for a vector destination is open coded. (reported by
2555 Brian Downing on c.l.l)
2556 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2557 documentation string.
2558 * fixed some bugs revealed by Paul Dietz' test suite:
2559 ** COPY-ALIST now signals an error if its argument is a dotted
2561 ** Condition slots are now accessed more correctly in the presence
2562 of multiple initargs for a given slot.
2563 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2564 exclude restarts of the same name associated with a different
2566 ** DEFCLASS of forward-referenced classes with another
2567 forward-referenced class in the superclasses list no longer
2569 ** Condition slots are now initialized once each, not multiple
2570 times. (thanks to Gerd Moellmann)
2571 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2573 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2574 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2575 arguments contain duplicated elements.
2576 ** RESTART-CASE understands local macros.
2577 ** RESTART-CASE associates exactly its own restarts with a condition.
2578 ** ENDP in safe mode checks its argument to be of type LIST.
2579 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2580 in question is unbound.
2581 ** Optimized MAKE-INSTANCE functions no longer cause internal
2582 assertion failures in the presence of duplicate initargs.
2583 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2584 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2585 propagated even in the ctor optimized implementation of
2587 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2589 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2590 * a better implementation of SXHASH on (simple) bit vectors,
2591 measured both in execution speed and in distribution of results
2592 over the positive fixnums, has been installed. Likewise, a better
2593 implementation of EQUAL for simple bit vectors is now available.
2594 * fixed CEILING optimization for a divisor of form 2^k.
2595 * fixed bug 240 (emitting extra style warnings "using the lexical
2596 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2597 by Antonio Martinez)
2598 * fixed SXHASH, giving different results for NIL depending on type
2599 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2600 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2601 variable. (found by Rolf Wester)
2602 * SBCL does not ignore type declarations for special
2603 variables. (reported by rif on c.l.l 2003-03-05)
2604 * some bug fixes in contrib/sb-aclrepl/
2605 * fixed some bugs revealed by Paul Dietz' test suite:
2606 ** a bug in the CONS type specifier, whereby the CAR and CDR
2607 types got intertwined, has been fixed;
2608 ** the type system is now able to reason about the interaction
2609 between INTEGER and RATIO types more completely;
2610 ** APPEND, [N]REVERSE and NRECONC check that those their
2611 arguments, which must be proper lists, are really so;
2612 ** An array specialized to be unable to hold elements has been
2613 implemented, as required -- yes, really -- by ANSI;
2614 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2615 on malformed property lists;
2617 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2618 * incompatible packaging change: in line with Unix convention,
2619 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2620 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2621 * REQUIRE and PROVIDE are now optionally capable of doing something
2622 useful. See the documentation string for REQUIRE.
2623 * infrastructure for a managed SBCL contrib system: contributed
2624 modules in this release include:
2625 ** the ASDF system definition facility;
2626 ** an interface to the BSD Sockets API;
2627 ** an ACL-like convenience interface to the repl;
2628 (thanks to Kevin Rosenberg)
2629 ** an implementation of ROTATE-BYTE, with efficient implementation
2631 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2632 gives a closer approximation to the right answer than previously.
2633 (thanks to Raymond Toy)
2634 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2635 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2636 optional environment arguments, as required by ANSI.
2637 * fixed bugs in other functions taking environment objects, allowing
2638 calls with an explicit NIL environment argument to be compiled
2640 * fixed bug 228: primary return values from
2641 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2642 COMPILE or FUNCTION.
2643 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2644 :TYPE will no longer signal errors on innocuous objects.
2645 * fixed bug 231b: SETQ is better at respecting type declarations in
2646 the lexical environment.
2647 * fixed a bug in DEFCLASS: classes named by symbols with no or
2648 unprintable packages can now be defined.
2649 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2650 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2651 * fixed compiler failure related to checking types of functions.
2652 (reported by Robert E. Brown)
2653 * the compiler is now much more consistent in its error-checking
2654 treatment of bounding index arguments to sequence functions: in
2655 (SAFETY 3) code, errors will be signalled in almost all cases if
2656 invalid sequence bounding indices are passed to functions defined
2657 by ANSI to operate on sequences.
2658 * fixed a bug in the build procedure: documentation of SBCL-specific
2659 packages is now preserved and available in the final Lisp image.
2660 * lifted FDEFINITION lookup out of loops in the implementation of
2661 many list operations. (thanks to Robert E. Brown)
2662 * fixed a bug in the reader: the #n# reader macro now works for
2663 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2664 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2665 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2666 E. Brown and Tony Martinez respectively)
2667 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2668 SLOT-BOUNDP: in particular, it is now able to optimize them much
2669 better, and is now not vulnerable to having packages renamed.
2670 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2671 always called when a slot is not present in an instance. (thanks
2673 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2674 longer signal an error on unreading a character following EOF on
2675 the previous constituent stream. (thanks to Tony Martinez)
2676 * fixed some bugs revealed by Paul Dietz' test suite:
2677 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2678 not just nonnegative fixnums;
2679 ** the logical bit-array operators such as BIT-AND now accept an
2680 explicit NIL for their "opt-arg" argument (to indicate a
2681 freshly-consed result bit-array);
2682 ** ELT now signals an error on an invalid sequence index in safe
2684 ** the type system is now cleverer about negations of numeric
2685 types, and consequently understands the BIGNUM and RATIO types
2687 ** the type system is now cleverer about the interaction between
2688 INTEGER and RATIO types: while bugs still remain, many more
2689 cases are accurately computed;
2690 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2691 if it is in the last clause;
2692 ** CONSTANTLY now correctly returns a side-effect-free function in
2694 ** DECLARE is no longer treated as a special-operator; in
2695 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2696 * incremented fasl file version number due to the change in the
2697 DEFSTRUCT-SLOT-DESCRIPTION structure.
2699 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2700 * minor incompatible change: code processed by the "interpreter" or
2701 EVAL now has a compilation optimization policy of (DEBUG 2)
2702 (changed from (DEBUG 1)) to improve debuggability of interactive
2703 development, and to allow the use of the debug RETURN command in
2705 * an experimental implementation of the RETURN command for the
2706 debugger has been included. (thanks to Frederik Kuivinen)
2707 * fixed bug 62: constraints were not propagated into a loop.
2708 * fixed bug in embedded calls of SORT (reported and investigated by
2710 * fixed some bugs revealed by Paul Dietz' test suite:
2711 ** printing and reading of arrays with some dimensions having
2712 length 0 (thanks to Gerd Moellmann);
2713 ** BOA constructor with &AUX argument without a default value does
2714 not cause a type error;
2715 ** CONSTANTP now returns true for all self-evaluating objects.
2717 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2718 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2719 accessors that are related by inheritance, as specified in the
2720 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2721 Valtteri Vuorikoski)
2722 * The compiler is now able to inline functions that were defined in
2723 a complex lexical environment (e.g. inside a MACROLET).
2724 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2726 * Support for the upcoming FreeBSD-5.0 release has been included.
2727 (thanks to Dag-Erling Smorgrav)
2728 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2729 effect when it is not in a toplevel context.
2730 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2731 stand a better chance of being correct. (thanks to Gerd
2733 * fixed bug in COERCE, which now signals an error on coercing a
2734 rational to a bounded real type which excludes the expected
2736 * The compiler is now able to derive types more accurately from the
2737 COERCE and COMPILE functions.
2738 * fixed bug 223: functional binding is considered to be constant
2739 only for symbols in the CL package.
2740 * fixed bug 231: SETQ did not check the type of a variable being set
2741 (reported by Robert E. Brown)
2742 * A new optimization for MAKE-INSTANCE has been included, fixing
2743 various bugs (including relating to :ALLOCATION :CLASS slots and
2744 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2746 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2747 ** As required by ANSI, LOOP now disallows anonymous collection
2748 clauses such as COLLECT I in conjunction with aggregate boolean
2749 clauses such as THEREIS (= I 1);
2750 ** LOOP now signals an error when any variable is reused in the
2751 same loop (including the potentially useful construct analogous
2752 to WITH A = 1 WITH A = (1+ A);
2753 ** IT is only a special loop symbol within the first clause of a
2754 conditional loop clause;
2755 ** LOOP with a typed iteration variable over a hashtable now
2756 signals a type error iff it should.
2757 * fixed some other bugs revealed by Paul Dietz' test suite:
2758 ** FILE-STREAM now names the class previously known as FD-STREAM;
2759 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2760 argument) no longer signals an error;
2761 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2762 of the slot symbol, rather than using the current package
2763 ((:CONC-NAME "") continues to intern the slot's name in the
2765 * incremented fasl file version number, because of the incompatible
2766 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2767 of the new implementation of DEFINE-COMPILER-MACRO.
2769 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2770 * Support for building SBCL for MIPS platforms running in
2771 little-endian mode has now been checked in, and basic
2772 functionality on said platforms verified.
2773 * minor incompatible change: PCL now records the pathname of a file
2774 in which methods and the like are defined, rather than its
2776 * minor incompatible change: TRUENAME now considers the truename of
2777 a file naming a directory to be the pathname with :DIRECTORY
2778 component indicating that directory.
2779 * minor incompatible change: a NAMED clause in the extended form of
2780 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2781 reason for the previous behaviour is unclear.
2782 * more systematization and improvement of CLOS and MOP conformance
2783 in PCL (thanks to Gerd Moellman and Pierre Mai):
2784 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2786 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2787 primary methods with no specializers;
2788 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2790 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2791 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2792 FORWARD-REFERENCED-CLASSes; error reporting on
2793 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2795 ** SXHASH on CLOS instances now uses a slot internal to the
2796 instance to return different numbers on distinct instances,
2797 while preserving the same return value through invocations of
2799 ** DEFMETHOD signals errors when methods with longer incongruent
2800 lambda lists are added to generic functions;
2801 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2802 CLASS, as specified in AMOP;
2803 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2804 on the order returned by the primary method for classes of
2805 class STANDARD-CLASS;
2806 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2807 * fixed some bugs shown by Paul Dietz' test suite:
2808 ** DOLIST puts its body in TAGBODY;
2809 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2811 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2812 value producing form;
2813 ** if more variables are given to PROGV than values, extra
2814 variables are bound and made to have no value;
2815 ** NSUBSTITUTE on list arguments gets the right answer with
2817 ** ELT signals an error of type TYPE-ERROR when the index argument
2818 is not a valid sequence index;
2819 ** LOOP signals (at macroexpansion time) an error of type
2820 PROGRAM-ERROR when duplicate variable names are found;
2821 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2822 ** FUNCALL of special-operators now cause an error of type
2824 ** PSETQ now works as required in the presence of side-effecting
2825 symbol-macro places;
2826 ** NCONC accepts any object as its last argument;
2827 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2829 ** loop-for-as-package does not require a package to be explicitely
2831 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2833 * fixed bug 166: compiler preserves "there is a way to go"
2834 invariant when deleting code.
2835 * fixed bug 172: macro lambda lists with required arguments after
2836 &REST arguments now cause an error to be signalled. (thanks to
2838 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2839 bug. (thanks to Matthew Danish)
2840 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2842 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2843 arguments are lexically rebound. (thanks to Gerd Moellmann and
2845 * fixed bug 194: error messages are now more informative when there
2846 is no primary method applicable in a call to a generic
2847 function. (thanks to Gerd Moellmann)
2848 * fixed bug in command line argument checking (thanks to Julian
2850 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2851 (thanks to Matthew Danish)
2852 * incremented fasl file version number, because of the
2853 SXHASH-related changes in the layout of CLOS data structures
2855 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2856 * minor incompatible change: The runtime (the Unix executable named
2857 "sbcl") is now much pickier about the .core files it will load.
2858 Essentially it now requires .core files to descend from the same
2859 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2860 as the runtime does. (The intent is to prevent the crashes which
2861 can occur, and which can even be reported as mysterious failures,
2862 when people patch the sources or change the build parameters
2863 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2864 sbcl and .core files.)
2865 * fixed bug: VALUES-LIST is no longer optimized away.
2866 * fixed bug 142: The FFI conversion of C string values to Lisp
2867 string values no longer conses excessively. (thanks to Nathan
2868 Froyd porting Raymond Toy's fix to CMU CL)
2869 * began to systematize and improve MOP conformance in PCL (thanks to
2870 Nathan Froyd, Gerd Moellman and Pierre Mai):
2871 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2873 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2874 ** FINALIZE-INHERITANCE is now called on class finalization;
2875 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2876 argument precedence order.
2877 * fixed bug 202: The compiler no longer fails on functions whose
2878 derived types contradict their declared type.
2879 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2880 so it can be non-toplevel.
2881 * The fasl file version number has changed (because of the new
2882 implementation of DEFMACRO).
2883 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2884 safe code, that any length requirement by their type-specifier
2885 argument is valid. The exceptions to this are described in bug
2887 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2888 functions now check the types of their inputs as required by ANSI.
2889 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2890 introduced symbol is DECLAREd to be SPECIAL.
2891 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2892 * fixed bug: :COUNT argument to sequence functions may be negative
2893 * fixed bug: body of DO-SYMBOLS may contain declarations
2894 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2895 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2896 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2897 symbol macro only once
2898 * fixed printing of call frame when argument list is unavailable
2899 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2900 * compiler no longer signals WARNING on unknown keyword
2903 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2904 * A beta-quality port to the mips architecture running Linux,
2905 based on the old CMUCL backend, has been made. It has been tested
2906 on a big-endian kernel, and works sufficiently well to be able to
2907 rebuild itself; it has not been tested in little-endian mode.
2908 * fixed an inconsistency between gencgc.c and purify.c which made
2909 dumping/loading .core files unreliable
2910 * fixed bug 120a: The compiler now deals correctly with IFs where
2911 the consequent is the same as the alternative, instead of
2912 misderiving the return type. (thanks to Alexey Dejneka)
2913 * fixed bug 113: Logical pathnames are now dumpable (the logical
2914 host is resolved at load-time, throwing an error if it is not
2916 * fixed bug 174: FORMAT's error message is slightly clearer when a
2917 non-printing character is used in a format directive.
2918 * fixed several bugs in compiler checking of type declarations, i.e.
2919 violations of the Python "declarations are assertions" principle
2920 (thanks to Alexey Dejneka)
2921 * fixed several bugs in PCL's error checking (thanks to Gerd
2923 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2925 * fixed bug in compilation of functions as first class values
2926 (thanks to Antonio Martinez-Shotton)
2927 * The compiler's handling TYPE-ERRORs which it can prove will
2928 inevitably happen at runtime has been cleaned up and corrected
2929 in several ways. (thanks to Alexey Dejneka)
2930 * improved argument type checking for various basic arithmetic
2931 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2932 lavished on them in the past that they can be compiled in many
2933 ways in different special cases
2934 * fixed bug 181: compiler checks validity of user supplied type
2936 * cleaned up code flushing in optimization: Function calls which
2937 should signal errors for safety purposes (e.g. which ANSI says
2938 should signal errors when their arguments are of incorrect type)
2939 are no longer optimized away.
2940 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2941 * incremented fasl file version number, because changes in the
2942 implementation of sequence functions like COERCE caused
2943 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2944 in old inline expansions) to become undefined. (Actually these
2945 changes were later undone, so we might very well be binary
2946 compatible with 0.7.7 after all, but leaving the version number
2947 incremented seemed like the simplest and most conservative
2950 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2951 * An alpha-quality port to the parisc architecture running Linux,
2952 based on the old CMUCL backend, has been made. This, even more so
2953 than the other backends, should be considered still a work in
2954 progress; known problems include that the Linux kernel in 64-bit
2955 mode does not propagate the correct sigcontext structure to
2956 userspace, and consequently SBCL on a parisc64 kernel will not
2958 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2959 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2960 suppressed.) Also now that the compiler is looking at declarations
2961 in the environment, it checks optimization declarations as well,
2962 and suppresses inlining when (> DEBUG SPEED).
2963 * More fixes have been made to treatment of floating point exception
2964 treatment and other Unix signals. In particular, floating point
2965 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2966 * The detection and handling of control stack exhaustion (infinite
2967 or very deeply nested recursion) has changed. Stack exhaustion
2968 detection is now done by write-protecting pages at the OS level
2969 and applies at all optimization settings; when found, a
2970 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2971 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2972 be caught using IGNORE-ERRORS.
2973 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2974 correctly when transforming recursive LABELS functions to LETs.
2975 (thanks to Alexey Dejneka)
2976 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2977 that are names of constants or global variables.
2978 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2979 alien routines with docstrings.
2980 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2981 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2983 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2984 (thanks to Eric Marsden)
2985 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2986 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2987 * Bug fix: The SPARC backend can now compile functions involving
2988 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2989 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2990 to a functions expecting a PATHNAME-DESIGNATOR.
2991 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2992 lambda lists. (thanks to Alexey Dejneka)
2993 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2994 functions so that EXPORTing the name of the class doesn't cause
2995 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2996 to Antonio Martinez for reporting this)
2997 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2998 constant-folded without failing an assertion. (thanks to Einar
2999 Floystad Dorum for reporting this)
3000 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3001 types are now dealt with more correctly by the compiler.
3002 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3003 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3004 behaviour analogous to RENAME-FILE. This puts its behaviour more
3005 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3007 * The fasl file version number has changed again. (because of the
3008 bug fix involving the names of PCL MAKE-INSTANCE functions)
3010 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3011 * bug fix: Floating point exceptions are treated much more
3012 consistently on the x86/Linux and PPC/Linux platforms.
3013 * Array initialization with :INITIAL-ELEMENT is now much faster for
3014 cases when the compiler cannot open code the array creation, but
3015 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3016 array accesses have also seen a speed increase.
3017 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3018 specified. (This was at the root of some bad interactions between
3019 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3021 * bug fix: Internal error arguments for undefined functions are now
3022 computed correctly on the PPC/Linux platform.
3023 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3024 Raymond Toy's patch for CMU CL)
3025 * Support for the Solaris 9 operating environment has been included
3026 (thanks to Daniel Merritt)
3027 * A very ugly but hopefully complete draft of the missing FFI chapter
3028 of the manual has been created by reformatting the corresponding
3029 CMU CL manual chapter into (currently very ugly and incoherent)
3030 DocBook and bringing it up to date for SBCL behavior. Thus, the
3031 manual is now essentially complete, at least by my extreme
3032 once-and-only-once standards, whereby it's acceptable to refer to
3033 the doc strings of SB-EXT functions as the primary documentation.
3034 * The fasl file version number has changed again, due to cleanup of
3035 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3036 is no longer a static symbol.)
3038 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3039 * SBCL now builds with OpenMCL (version 0.12) as the
3040 cross-compilation host; also, more progress has been made toward
3041 bootstrapping under CLISP.
3042 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3044 * bug 158 fixed: The compiler can now deal with integer loop
3045 increments different from 1; fixing this turned out also to fix
3047 * bug 169 fixed: no more bogus warnings about using lexical bindings
3048 despite the presence of perfectly good SPECIAL declarations (thanks
3049 to David Lichteblau)
3050 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3051 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3052 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3054 * bug 180 fixed: Method combination specifications no longer ignore
3055 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3056 * bug fix: Structure type predicate functions now check their argument
3057 count as they should.
3058 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3059 correctly. (thanks to Pierre Mai)
3060 * minor incompatible change: The --noprogrammer option is deprecated
3061 in favor of the new --disable-debugger option, which is very similar.
3062 (The major difference is that it takes effect at a slightly different
3063 time at startup, causing handling of errors in --sysinit and
3064 --userinit files will be affected differently.) The
3065 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3066 been added to allow this functionality to be controlled from ordinary
3067 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3068 maintainers, who might want to run non-interactive scripts to
3069 build SBCL cores which will later be used interactively.)
3070 * minor incompatible change: The LOAD function no longer, when given
3071 a wild pathname to load, loads all files matching that pathname.
3072 Instead, an error of type FILE-ERROR is signalled.
3074 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3075 * bug 147 fixed: The compiler preserves its block link/count
3076 invariants more correctly now so that it doesn't crash. (thanks
3078 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3080 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3081 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3082 no-value-for-this-character as NIL (as specified by ANSI).
3083 * HOST-NAMESTRING on physical pathnames now returns a string that is
3084 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3085 (thanks to Christophe Rhodes)
3086 * The Alpha port handles icache flushing more correctly. (thanks to
3088 * More progress has been made toward bootstrapping under CLISP. (thanks
3089 to Christophe Rhodes)
3090 * The fasl file format has changed again, because dynamic loading
3091 on OpenBSD (which has non-ELF object files) motivated some cleanups
3092 in the way that foreign symbols are transformed and passed around.
3093 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3094 is no longer treated as whitespace by the reader, but instead as
3095 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3096 |A
\7fB|, instead of A as it used to.
3098 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3099 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3100 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3101 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3102 this as of 0.7.1.45, but was left out of the previous news section
3103 (thanks to Dan Barlow)
3104 * SBCL now runs on the Solaris operating system on SPARC architectures
3105 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3106 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3107 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3108 * SPARC backend cleanups, allowing builds of cores optimized for V8
3109 and V9 SPARCS, and also emission of code targeted to a particular
3110 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3112 * SBCL is closer to bootstrapping under CLISP, thanks to various
3113 fixes by Christophe Rhodes.
3114 * The fasl file format has changed again, to allow the compiler's
3115 INFO database to support symbol macros.
3116 * The user manual (in doc/) is formatted into HTML more nicely.
3117 (thanks to coreythomas)
3118 * The system is smarter about SUBTYPEP relationships, especially
3119 those involving NOT types (including types such as ATOM which are
3120 represented internally using NOT types). Thus SUBTYPEP is less
3121 likely to return (VALUES NIL NIL) in general, and in particular
3122 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3124 * The fasl file format has changed again, because the internal
3125 representation of types now includes a new slot to support the new
3126 SUBTYPEP-of-NOT-types logic.
3127 * (not a change in the main branch of SBCL, but a related prototype
3128 which can hopefully be merged into the main branch of SBCL in the
3129 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3130 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3131 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3132 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3133 dispatch macro character combinations. (thanks to Alexey Dejneka)
3134 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3135 that superficially look like logical namestrings correctly.
3136 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3138 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3139 default supplied-p) for &optional and &key arguments. (thanks to
3142 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3143 * incompatible change: The compiler is now less aggressive about
3144 tail call optimization, doing it only when (> SPACE DEBUG) or
3145 (> SPEED DEBUG). (This is an incompatible change because there are
3146 programs which relied on the old CMU-CL-style behavior to optimize
3147 away their unbounded recursion which will now die of stack overflow.)
3148 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3149 for non-GENCGC systems has been increased to 20M (since that
3150 seems much closer to the likely performance optimum for modern
3151 systems than the old 4M value was)
3152 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3153 and *DEBUG-PRINT-LEVEL*
3154 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3155 of CMU CL's support for SPARC, and various endianness and other
3156 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3157 * new syntactic sugar for the Unix command line: --load foo.bar is now
3158 an alternate notation for --eval '(load "foo.bar")'.
3160 ** The system now detects stack overflow and handles it gracefully,
3161 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3162 optimization settings. (This is a good thing in general, and
3163 its introduction in this version should be particularly timely
3164 for anyone whose code fails because of suppression of tail
3166 ** The system now hunts for the C variable "environ" in a more
3167 devious way, to avoid segfaults when the C library version differs
3168 between compile time and run time. (thanks to Christophe Rhodes)
3169 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3170 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3171 ** The compiler no longer issues bogus style warnings for undefined
3172 classes in the same source file as the DEFCLASSes which defined
3173 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3175 ** fixes in CONDITION class precedence list for undefined function
3176 errors (thanks to Alexei Dejneka)
3177 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3178 correctly. (thanks to Dan Barlow)
3179 ** portability fixes aiming at bootstrapping under CLISP (thanks
3180 to Dave McDonald and Christophe Rhodes)
3181 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3182 ** fixes in type translation and and type inference (thanks to
3184 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3185 ** various fixes in the new ports (thanks to Dan Barlow)
3186 * several changes related to debugging:
3187 ** suppression of tail recursion, as noted above
3188 ** stack overflow detection, as noted above
3189 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3190 is now the default. (For some time encapsulation has been more
3191 reliable than the breakpoint-based :ENCAPSULATE NIL
3192 implementation, at least on X86 systems; and I just noticed that
3193 encapsulation also seems closer to the spirit of the ANSI
3196 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3198 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3199 up properly again. (There was a packaging bug in 0.7.0 which
3200 left their definitions in SB-SYS::LOAD-FOREIGN and
3201 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3202 vital for most things which interface to C-level interfaces,
3203 like extensions working with sockets or databases or
3204 Perl-compatible regexes or whatever, and the need to fix
3205 this bug is the main reason that 0.7.1 was released so
3207 ** DEFGENERIC is now choosier about the methods it redefines, so that
3208 reLOADing a previously-LOADed file containing DEFGENERICs does
3209 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3210 cycle now works as it should. (thanks to Alexey Dejneka)
3211 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3212 fixed by Christophe Rhodes. (He actually submitted this patch
3213 months ago, and I delayed until after 0.7.0.)
3214 ** Bug 111 (internal compiler confusion about runtime checks on
3215 FUNCTION types) was fixed by Alexey Dejneka.
3216 * Some internal cleanups (getting rid of variables which aren't
3217 needed now that the byte interpreter is gone) caused the fasl
3218 file format number to change again.
3220 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3221 * major incompatible change: The default fasl file extension, i.e. the
3222 default extension for files produced by COMPILE-FILE, has changed
3223 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3225 ** There are many changes in the implementation of the compiler.
3226 SBCL is now essentially a compiler-only implementation of ANSI
3227 Common Lisp. EVAL still "interprets" a few special cases, but
3228 almost all the interesting cases are handled by creating
3229 a LAMBDA expression, calling COMPILE on it, then calling
3230 FUNCALL on the result.
3231 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3232 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3233 Since the code is newer, there might still be some new bugs
3234 (though not as many as before Martin Atzmueller's fixes:-). But
3235 the new code is substantially simpler and clearer, and hopefully
3236 any remaining bugs will be simpler, less fundamental, and more
3237 fixable then the bugs in the old code.
3238 ** The revised compiler is still a little unsteady on its feet.
3240 *** The debugging information it produces (particularly the names
3241 of FUNCTION objects) is sometimes much less useful than what
3242 the old compiler produced.
3243 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3244 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3245 within a MACROLET) has been temporarily weakened.
3246 ** There are new compiler optimizations for various functions:
3247 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3248 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3249 *** the math functions TRUNCATE, FLOOR, and CEILING
3250 *** the function-of-all-trades COERCE
3251 Mostly these should be transparent, but there's one
3252 potentially-annoying problem (bug 117): when the compiler
3253 inline-expands a function and does type analysis on the result,
3254 it can create control paths which have type mismatches, and
3255 when it can't prove that those control paths aren't taken,
3256 it will issue WARNINGs about the type mismatches. This is
3257 a particular problem in practice for the new sequence functions.
3258 It's not clear how this should be fixed, and for now, a
3259 workaround is given in the entry for 117 in the BUGS file.
3260 ** (Because of the interaction between the two previous items --
3261 occasional inlining problems and new inline expansions -- some
3262 of the new sequence function optimizations won't really kick in
3263 completely until debugging information, and then inlining, are
3264 straightened out in some future version.)
3265 * minor incompatible changes:
3266 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3267 DIRECTORY no longer implicitly promotes NIL slots of its
3268 pathname argument to :WILD. In particular, when you ask for the
3269 contents of a directory (which you used to be able to do without
3270 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3271 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3272 ** changes in behavior that ANSI explicitly defines to be
3273 implementation dependent:
3274 *** The new compiler-only implementation still conforms with ANSI,
3275 but acts a little different than before. Besides the obvious
3276 changes in performance tradeoffs (that the cost per form passed
3277 to EVAL has gone up, and the cost per form executed by EVAL
3278 has gone down), the behavior of the system changes a little
3279 because there are no longer any interpreted function objects.
3280 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3281 e.g. doing COMPILE on the output of interactive DEFUN is
3283 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3285 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3286 has changed from (:CL) to NIL.
3287 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3288 e.g. #\Newline and #\Tab, don't have names specified in the
3289 ANSI Common Lisp standard, is now based on their ASCII symbolic
3290 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3291 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3292 are no longer used for output.
3293 ** changes in internal implementation constants:
3294 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3295 4 million. (If your application spends a lot of time GCing and
3296 you have a lot of RAM, you might want to experiment with
3297 increasing it even more.)
3298 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3299 However, almost all old code should still continue to work without
3300 immediate update, as SB-C-CALL is now a (deprecated) nickname
3302 ** Old operator names in the style DEF-FOO are now deprecated in
3303 favor of new corresponding names DEFINE-FOO, for consistency with
3304 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3305 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3306 internal symbols, but a few supported extensions like
3307 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3308 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3309 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3310 etc. as you get deeper into recursive calls to the debugger
3311 command loop, instead of the old "5]", "5]]", "5]]]"
3312 sequence. (I was motivated to do this when squabbles between
3313 ILISP and SBCL left me very deeply nested in the debugger. In the
3314 short term, this change will probably provoke more ILISP/SBCL
3315 squabbles, but hopefully it will be an improvement in the long run.)
3316 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3317 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3318 function, since declaiming return types involving aliens is
3319 (1) annoyingly messy to do by hand and (2) vital to efficient
3320 compilation of code which calls such functions.
3321 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3322 longer reexported by the SB-EXT package. They're solely useful
3323 for alien code, so it seems more logical that you should get
3324 them from the SB-ALIEN package, not in SB-EXT.
3325 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3326 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3327 features. Instead, the code that they used to control is always
3328 built into the system.
3329 * many other bug fixes
3330 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3331 advantage of the new EVAL-WHEN stuff and to clean them up in
3332 general, and they are now more ANSI-compliant in a number of
3333 ways. Martin Atzmueller is responsible for a lot of this.
3334 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3336 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3337 *** correct ERROR type for various file operations
3338 *** some fixes for Lisp streams
3339 *** DEFMETHOD syntax checking
3340 *** changing old weird representation of debug information as
3341 strings (which, among their other deficiencies, don't transform
3342 correctly when you rename packages, and don't change their
3343 print representation when you change things like *PACKAGE*
3344 and *PRINT-LENGTH*) to symbols and lists of symbols
3345 He also made several improvements and fixed several bugs in DESCRIBE.
3346 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3348 *** misbehavior of WRITE-STRING/WRITE-LINE
3349 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3350 and several other LOOP problems as well
3351 *** DIRECTORY when similar filenames are present
3352 *** DEFGENERIC with :METHOD options
3353 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3354 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3355 *** argument ordering in FIND with :TEST option
3356 *** mishandled package designator argument in APROPOS-LIST
3357 *** various problems in the backquote readmacro
3358 *** a bug in APROPOS
3359 *** probably some others that I'm not describing very well here,
3360 since the CVS log documents them by reference to sbcl-devel
3361 messages, and the SourceForge archives aren't working well.:-(
3362 ** Dan Barlow improved the Alpha port (and is making progress on the
3363 PPC port, for those of you who think different).
3364 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3365 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3366 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3367 continues to work on the SPARC port (for those of you in a position
3368 to look down upon our little PC-compatible boxes from a great height).
3369 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3370 and uses that copy, rather than the current dynamic value, when
3371 it's trying to decide whether to truncate output. Thus e.g.
3372 (let ((*print-lines* 50))
3373 (pprint-logical-block (stream nil)
3375 (let ((*print-lines* 8))
3376 (print (aref possiblybigthings i) stream)))))
3377 should now truncate the logical block only at 50 lines, instead of
3378 often truncating it at 8 lines, as it did before.
3379 * The doc/cmucl/ directory, containing old CMU CL documentation
3380 from the time of the fork, is no longer part of the base system.
3381 SourceForge has shut down its anonymous FTP service, and with it
3382 my original plan for distributing the old CMU CL documentation
3383 there. For now, if you need these files you can download an old
3384 SBCL source release and extract them from it.
3385 * The fasl file version number changed again, for dozens of reasons,
3386 some of which are apparent above.
3388 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3389 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3390 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3391 patch, so that DISASSEMBLE output is much nicer.
3392 * The code in the SB-PROFILE package now seems reasonably stable.
3393 I still haven't decided what the final interface should look like
3394 (I'd like PROFILE to interact cleanly with TRACE, since both
3395 facilities use function encapsulation) but if you have a need
3396 for profiling now, you can probably use it successfully with
3397 the current CMU-CL-style interface.
3398 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3399 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3400 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3401 link now returns the dangling link itself, and for similar
3402 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3403 link itself. (In these cases the old code signalled an error and
3404 looped endlessly, respectively.) Thus, DIRECTORY now works even
3405 in the presence of dangling and cyclic symbolic links.
3406 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3407 is now a supported extension again, since the consensus on
3408 sbcl-devel was that it can be useful for ordinary development
3409 work, not just for debugging SBCL itself.
3410 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3411 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3412 that currently-defined functions might be redefined later with
3413 different return types.
3414 * Hash tables can be printed readably, as inspired by CMU CL code
3415 of Eric Marsden and SBCL code of Martin Atzmueller.
3416 * better error handling in CLOS method combination, thanks to
3417 Martin Atzmueller porting Pierre Mai's CMU CL patches
3418 * more overflow fixes for >16Mbyte I/O buffers
3419 * A bug in READ has been fixed, so that now a single Ctrl-D
3420 character suffices to cause end-of-file on character streams.
3421 In particular, now you only need one Ctrl-D at the command
3422 line (not two) to exit SBCL.
3423 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3425 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3426 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3427 does the right thing.
3428 * The implementation of some type tests, especially for CONDITION
3429 types, is now tidier and maybe faster, due to CMU CL code
3430 originally by Douglas Crosher, ported by Martin Atzmueller.
3431 * Some math functions have been fixed, and there are new
3432 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3433 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3434 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3435 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3436 likely to be built into sbcl-0.7.0.)
3437 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3438 CL, ported to SBCL by Martin Atzmueller)
3439 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3440 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3441 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3442 * There's a new slam.sh hack to shorten the edit/compile/debug
3443 cycle for low-level changes to SBCL itself, and a new
3444 :SB-AFTER-XC-CORE target feature to control the generation of
3445 the after-xc.core file needed by slam.sh.
3446 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3447 COMPILE-FILE is no longer supported, so that now every function
3448 gets an entry point, so that block compilation looks a little
3449 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3450 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3451 returns the number of bytes consed since the system started,
3452 rather than the number consed since the first time the function
3453 was called. (The new definition parallels ANSI functions like
3454 CL:GET-INTERNAL-RUN-TIME.)
3455 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3456 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3457 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3458 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3459 * Fasl file version numbers are now independent of the target CPU,
3460 since historically most system changes which required version
3461 number changes have affected all CPUs equally. Similarly,
3462 the byte fasl file version is now equal to the ordinary
3465 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3466 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3467 is no longer recognized. I apologize for this, because it was
3468 listed in SB-EXT as a supported extension, but I found that
3469 its existing behavior was poorly specified, as well as incorrectly
3470 specified, and it looked like too much of a mess to straighten it
3471 out. I have enough on my hands trying to get ANSI stuff to work..
3472 * many patches ported from CMU CL by Martin Atzmueller, with
3473 half a dozen bug fixes in pretty-printing and the debugger, and
3474 half a dozen others elsewhere
3475 * fixed bug 13: Floating point infinities are now supported again.
3476 They might still be a little bit flaky, but thanks to bug reports
3477 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3478 as flaky as they were.
3479 * The --noprogrammer command line option is now supported. (Its
3480 behavior is slightly different in detail from what the old man
3481 page claimed it would do, but it's still appropriate under the
3482 same circumstances that the man page talks about.)
3483 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3484 are now supported, and enabled by default. Thus, the compiler can
3485 handle many floating point and complex operations much less
3486 inefficiently. (Thus e.g. you can implement a complex FFT
3488 * The compiler now detects type mismatches between DECLAIM FTYPE
3489 and DEFUN better, and implements CHECK-TYPE more correctly, and
3490 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3491 thanks to patches from Martin Atzmueller.
3492 * various fixes to make the cross-compiler more portable to
3493 ANSI-conforming-but-different cross-compilation hosts (notably
3494 Lispworks for Windows, following bug reports from Arthur Lemmens)
3495 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3496 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3497 fixed thanks to Pierre Mai's CMU CL patches.
3498 * fixes to keep the system from overflowing internal counters when
3499 it tries to use i/o buffers larger than 16M bytes
3500 * fixed bug 45a: Various internal functions required to support
3501 complex special functions have been merged from CMU CL sources.
3502 (When I was first setting up SBCL, I misunderstood a compile-time
3503 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3504 * improved support for type intersection and union, fixing bug 12
3505 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3506 more obscure bugs as well
3507 * some steps toward byte-compiling non-performance-critical
3508 parts of the system, courtesy of patches from Martin Atzmueller
3509 * Christophe Rhodes has made some debian packages of sbcl at
3510 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3511 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3512 stable, but are nonetheless usable. When he's ready, I'd be happy
3513 to add them to the SourceForge "File Releases" section. (And if
3514 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3515 * new fasl file format version number (because of changes in
3516 internal representation of (OR ..) types to accommodate the new
3517 support for (AND ..) types, among other things)
3519 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3520 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3522 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3524 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3525 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3526 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3527 * DESCRIBE now works on structure objects again.
3528 * Most function call argument type mismatches are now handled as
3529 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3530 know whether the function will be redefined before the call is
3531 executed. (The compiler could flag local calls with full WARNINGs,
3532 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3533 it doesn't keep track of enough information to know whether calls
3534 are local in this sense.)
3535 * Compiler output is now more verbose, with messages truncated
3536 later than before. (There should be some supported way for users
3537 to override the default verbosity, but I haven't decided how to
3538 provide it yet, so this behavior is still controlled by the internal
3539 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3540 src/compiler/ir1util.lisp.)
3541 * Fasl file format version numbers have increased again, because
3542 support for the Gray streams extension changes the layout of the
3543 system's STREAM objects.
3544 * The Gray subclassable streams extension now works, thanks to a
3545 patch from Martin Atzmueller.
3546 * The full LOAD-FOREIGN extension (not just the primitive
3547 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3548 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3549 but like most other programs, it defaults to copying the Unix
3550 environment from the original process instead of starting the
3551 new process in an empty environment.
3552 * Extensions which manipulate the Unix environment now support
3553 an :ENVIRONMENT keyword option which doesn't smash case or
3554 do other bad things. The CMU-CL-style :ENV option is retained
3555 for porting convenience.
3556 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3557 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3559 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3561 * A patch from Martin Atzmueller seems to have solved the SIGINT
3562 problem, and as far as we know, signal-handling now works cleanly.
3563 (If you find any new bugs, please report them!)
3564 * The system no longer defaults Lisp source file names to types
3565 ".l", ".cl", or ".lsp", but only to ".lisp".
3566 * The compiler no longer uses special default file extensions for
3567 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3568 seems to expect a single default extension for all compiled code,
3569 and there's no compelling reason to try to stretch the standard
3570 to allow two different extensions.) Instead, byte-compiled files
3571 default to the same extension as native-compiled files.
3572 * Fasl file format version numbers have increased again, because
3573 a rearrangement of internal implementation packages made some
3574 dumped symbols in old fasl files unreadable in new cores.
3575 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3576 many fewer weird special cases.
3577 * Bug #17 (differing COMPILE-FILE behavior between logical and
3578 physical pathnames) has been fixed, and some related misbehavior too,
3579 thanks to a patch from Martin Atzmueller.
3580 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3581 by Tim Moore, ported to SBCL by Martin Atzmueller.
3582 * Martin Atzmueller fixed several filesystem-related problems,
3583 including bug #36, in part by porting CMU CL patches, which were
3584 written in part by Paul Werkowski.
3585 * More compiler warnings in src/runtime/ are gone, thanks to
3586 more patches from Martin Atzmueller.
3587 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3590 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3592 * DESCRIBE now works on CONDITION objects.
3593 * The debugger now handles errors which arise when trying to print
3594 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3596 * The build system now uses an additional file, customize-target-features.lisp,
3597 to allow local modifications to the target *FEATURES* list. (The point of
3598 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3599 debugging features enabled, without having to worry about propagating it
3600 into everyone's system when I do a "cvs update".) When no
3601 customize-target-features.lisp file exists, the target *FEATURES* list
3602 should be constructed the same way as before.
3603 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3604 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3605 warn about behavior which is undefined under the ANSI spec. Note: This
3606 is specified by ANSI, but it's not very popular with programmers.
3607 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3608 macro in the SBCL sources for an example of a workaround which you
3609 might use to make portable ANSI-standard code which does what you want.
3610 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3611 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3612 when it's used as a non-toplevel form. (This is required in order
3613 to implement the DEFCONSTANT-EQX macro.)
3614 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3615 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3616 funny things, due to bugs in the implementation of EVAL-WHEN
3617 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3618 non-toplevel DEFCONSTANTs are uncommon.)
3619 * The core file version number and fasl file version number have been
3620 incremented, because the old noncompliant DEFCONSTANT behavior involved
3621 calling functions which no longer exist, and because I also took the
3622 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3623 * fixed bug 1 (error handling before read-eval-print loop starts), and
3624 redid debugger restarts and related debugger commands somewhat while
3626 ** The QUIT debugger command is gone, since it did something
3627 rather different than the SB-EXT:QUIT command, and since it never
3628 worked properly outside the main toplevel read/eval/print loop.
3629 Invoking the new TOPLEVEL restart provides the same functionality.
3630 ** The GO debugger command is also gone, since you can just invoke
3631 the CONTINUE restart directly instead.
3632 ** The TOP debugger command is also gone, since it's redundant with the
3633 FRAME 0 command, and since it interfered with abbreviations for the
3635 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3637 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3638 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3639 some porting work by Martin Atzmueller)
3640 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3641 with a boatload of patches which he ported to SBCL. Now that those
3643 ** The system tries to make sure that its low-priority messages
3644 are prefixed by semicolons, to help people who like to use
3645 syntax highlighting in their ILISP buffer. (This patch
3646 was originally due to Raymond Toy.)
3647 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3648 patches originally written by Raymond Toy.
3649 ** The compiler understands coercion between single-value and
3650 multiple-VALUES type expressions better, getting rid of some very
3651 weird behavior, thanks to patches originally by Robert MacLachlan
3652 and Douglas Crosher.
3653 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3654 lambda lists, thanks to a patch originally by Pierre Mai.
3655 ** The system no longer bogusly warns about "abbreviated type
3657 ** The compiler gets less confused by inlining and RETURN-FROM,
3658 thanks to some patches originally by Tim Moore.
3659 ** The system no longer hangs when dumping circular lists to fasl
3660 files, thanks to a patch originally from Douglas Crosher.
3661 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3662 undefined function error.
3663 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3664 in structure slot initforms) for now, documented workaround instead:-|
3665 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3666 * fixed bug 5: added stubs for various Gray stream functions called
3667 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3668 installed, at least appropriate type errors are generated
3669 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3670 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3671 consistently in DEFMETHOD forms.
3672 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3673 it doesn't seem to affect SBCL after all
3674 * The C runtime system now builds with better optimization and many
3675 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3677 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3679 * The system is now under CVS at SourceForge (instead of the
3680 CVS repository on my home machine).
3681 * The new signal handling code has been tweaked to treat register
3682 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3683 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3684 downstream, e.g. in debug-int.lisp, has implicit dependencies
3685 on the unsignedness of integer representation of machine words,
3686 and that caused the system to bomb out with infinite regress
3687 when trying to recover from type errors involving signed values,
3688 e.g. (BUTLAST '(1 2 3) -1).)
3689 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3690 (This was one of the bugs Peter Van Eynde reported back in July.)
3691 * The system now uses code inspired by Colin Walters' O(N)
3692 implementation of MAP (from the cmucl-imp@cons.org mailing
3693 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3694 inline the MAP operation, and there is more than one
3695 sequence argument to the MAP call (so that it can't just
3696 do ETYPECASE once and for all based on the type of the
3697 single sequence argument). (The old non-inline implementation
3698 of the general M-argument sequence-of-length-N case required
3699 O(M*N*N) time when any of the sequence arguments were LISTs.)
3700 * The QUIT :UNIX-CODE keyword argument has been renamed to
3701 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3702 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3704 * Raymond Wiker's patches to port dynamic loading from Linux to
3705 FreeBSD have been added.
3706 * The BUGS file is now more nearly up to date, thanks in large part
3707 to Martin Atzmueller's review of it.
3708 * The debugger now flushes standard output streams before it begins
3709 its output ("debugger invoked" and so forth).
3710 * The core version number and fasl file version number have both
3711 been incremented, because of incompatible changes in the layout
3713 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3714 used to not be called for a saved Lisp image.)
3715 * Martin Atzmueller's version of a patch to fix a compiler crash,
3716 as posted on sbcl-devel 13 September 2000, has been installed.
3717 * Instead of installing Martin Atzmueller's patch for the
3718 compiler transform for SUBSEQ, I deleted the compiler transform,
3719 and transforms for some similar consing operations.
3720 * A bug in signal handling which kept TRACE from working on OpenBSD
3722 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3723 away by constant folding
3724 * The system now defines its address space constants in one place
3725 (in the Lisp sources), and propagates them automatically elsewhere
3726 (through GENESIS and the sbcl.h file). Therefore, patching the
3727 address map is less unnecessarily tedious and error-prone. The
3728 Lisp names of address space constants have also been systematized.
3729 * CVS tags like dollar-Header-dollar have been removed from
3730 the sources, because they have never saved me trouble and
3731 they've been source of trouble working with patches and other
3732 diff-related operations.
3733 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3734 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3736 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3738 * The system has been ported to OpenBSD.
3739 * The system now compiles with a simple "sh make.sh" on the systems
3740 that it's supported on. I.e., now you no longer need to tweak
3741 text in the target-features.lisp-expr and symlinks in src/runtime/
3742 by hand, the make.sh takes care of it for you.
3743 * The system is no longer so grossly inefficient when compiling code
3744 involving vectors implemented as general (not simple) vectors (VECTOR T),
3745 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3746 which dares to use the various sequence functions on non-simple
3747 vectors, takes less of a performance hit.
3748 * There is now a primitive type predicate VECTOR-T-P
3749 to test for the (VECTOR T) type, so that e.g.
3750 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3751 can now be compiled with some semblance of efficiency. (The old code
3752 turned the type declaration into a full call to %TYPEP at runtime!)
3753 * AREF on (VECTOR T) is still not fast, since it's still compiled
3754 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3755 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3756 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3757 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3758 without making full calls to SUBTYPEP at runtime.
3759 (Some analogous efficiency issues for non-simple vectors specialized to
3760 element types other than T, or for non-simple multidimensional arrays,
3761 have not been addressed. They could almost certainly be handled the
3762 same way if anyone is motivated to do so.)
3763 * The changes in array handling break binary compatibility, so
3764 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3765 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3766 instead of (VALUES T T).
3767 * By following the instructions that Dan Barlow posted to sbcl-devel
3768 on 2 July 2000, I was able to enable primitive dynamic object
3769 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3770 functionality is not implemented (since it calls ld to resolve
3771 library references automatically, requiring RUN-PROGRAM for its
3772 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3773 try to resolve library references) is now supported.
3774 * The system now flushes the standard output streams when it terminates,
3775 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3776 them at several other probably-convenient times, e.g. in each pass of
3777 the toplevel read-eval-print loop, and after evaluating a form given
3778 as an "--eval" command-line option. (These changes were motivated by a
3779 discussion of stream flushing issues on cmucl-imp in August 2000.)
3780 * The source transform for TYPEP of array types no longer assumes
3781 that an array whose element type is a not-yet-defined type
3782 is implemented as an array of T, but instead punts, so that the
3783 type will be interpreted at runtime.
3784 * There is now some support for cross-compiling in make.sh: each of
3785 the phases of make.sh has its own script. (This should be transparent
3786 to people doing ordinary, non-cross-compile builds.)
3787 * Since my laptop doesn't have hundreds of megabytes of memory like
3788 my desktop machine, I became more motivated to do some items on
3789 my to-do list in order to reduce the size of the system a little:
3790 ** Arrange for various needed-only-at-cold-init things to be
3791 uninterned after cold init. To support this, those things have
3792 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3793 symbols with such names are now uninterned after cold init).
3794 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3795 fasl files for cold load.
3796 ** Remove the old compiler structure pooling code (which used to
3797 be conditional on the target feature :SB-ALLOC) completely.
3798 ** Redo the representation of some data in cold init to be more compact.
3799 (I also looked into supporting byte compiled code at bootstrap time,
3800 which would probably reduce the size of the system a lot, but that
3801 looked too complicated, so I punted for now.)
3802 * The maximum signal nesting depth in the src/runtime/ support code has
3803 been reduced from 4096 to 256. (I don't know any reason for the very
3804 large old value. If the new smaller value turns out to break something,
3805 I'll probably just bump it back up.)
3806 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3808 * Many, many bugs reported by Peter Van Eynde have been added to
3809 the BUGS list; some have even been fixed.
3810 * While enabling dynamic object file loading, I tried to make the
3811 code easier to understand, renaming various functions and variables
3812 with less ambiguous names, and changing some function calling
3813 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3814 * While trying to figure out how to do the OpenBSD port, I tried to
3815 clean up some of the code in src/runtime/. In particular, I dropped
3816 support for non-POSIX signal handling, added various comments,
3817 tweaked the code to reduce the number of compilation warnings, and
3818 renamed some files to increase consistency.
3819 * To support the new automatic configuration functionality in make.sh,
3820 the source file target-features.lisp-expr has been replaced with the
3821 source file base-target-features.lisp-expr and the machine-generated
3822 file local-target-features.lisp-expr.
3823 * fixed a stupid quoting error in make.sh so that using CMU CL
3824 "lisp -batch" as cross-compilation host works again
3826 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3828 * DESCRIBE no longer tries to call itself recursively to describe
3829 bound/fbound values, so that it no longer fails on symbols which are
3830 bound to themselves (like keywords, T, and NIL).
3831 * DESCRIBE now works on generic functions.
3832 * The printer now prints less-screwed-up representations of closures
3833 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3834 * A private symbol is used instead of the :EMPTY keyword previously
3835 used to mark empty slots in hash tables. Thus
3836 (DEFVAR *HT* (MAKE-HASH-TABLE))
3837 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3838 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3839 now does what ANSI says that it should. (You can still get
3840 similar noncompliant behavior if bang on the hash table
3841 implementation with all the symbols you get back from
3842 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3843 This breaks binary compatibility, since tests for equality to
3844 :EMPTY are wired into things like the macroexpansion of
3845 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3847 * There's now a minimal placeholder implementation for CL:STEP,
3848 as required by ANSI.
3849 * An obscure bug in the interaction of the normal compiler, the byte
3850 compiler, inlining, and structure predicates has been patched
3851 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3852 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3854 * Missing ordinary arguments in a macro call are now detected even
3855 when the macro lambda list contains &KEY or &REST.
3856 * The debugger no longer complains about encountering the top of the
3857 stack when you type "FRAME 0" to explicitly instruct it to go to
3858 the top of the stack. And it now prints the frame you request even
3859 if it's the current frame (instead of saying "You are here.").
3860 * As specified by ANSI, the system now always prints keywords
3861 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3863 * The default initial SIZE of HASH-TABLEs is now smaller.
3864 * Type information from CLOS class dispatch is now propagated
3865 into DEFMETHOD bodies, so that e.g.
3866 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3868 is now basically equivalent to
3869 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3870 (DECLARE (TYPE SINGLE-FLOAT X))
3872 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3873 operation, without having to do run-time type dispatch.
3874 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3875 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3876 * Since I don't seem to be making big file reorganizations very often
3877 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3878 is overflowing my ability to conveniently back them up), I've finally
3879 checked the system into CVS. (The CVS repository is on my home system,
3880 not at SourceForge -- putting it on SourceForge might come later.)
3881 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3882 high-level GC-NOTIFY-FOO functions send their output. (There's
3883 still very little control of where low-level verbose GC functions
3884 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3885 less than it used to -- the GC-NOTIFY-FOO functions are now under
3886 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3887 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3888 in only one place in the source code, and propagates it automatically
3889 everywhere that it's needed. Thus e.g. when I bump the version from
3890 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3891 * The C source files now include boilerplate legalese and documentation
3892 at the head of each file (just as the Lisp source files already did).
3893 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3894 to his page will be replaced with a link to his new CLiki service.
3896 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3898 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3899 * The build process now looks for GNU make under the default name "gmake",
3900 instead of "make" as it used to. If GNU make is not available as "gmake"
3901 on your system, you can change this default behavior by setting the
3902 GNUMAKE environment variable.
3903 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3904 can build without error under CMU CL.
3906 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3908 * There is now a partial SBCL user manual (with some new text and some
3909 text cribbed from the CMU CL manual).
3910 * The beginnings of a profiler have been added (starting with the
3911 CMU CL profiler and simplifying and cleaning up). Eventually the
3912 main interface should be through the TRACE macro, but for now,
3913 it's still accessed through vaguely CMU-CL-style functions and macros
3914 exported from the package SB-PROFILE.
3915 * Some problems left over from porting CMU CL to the new
3916 cross-compilation bootstrap process have been cleaned up:
3917 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3918 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3919 part when getting rid of the compiler *BACKEND* stuff.)
3920 ** The value of *NULL-TYPE* was screwed up, because it was
3921 being initialized before the type system knew the final
3922 definition of the 'NULL type. This screwed up several key
3923 optimizations in the compiler, causing inefficiency in all sorts
3924 of places. (I found it because I wanted to understand why
3925 GET-INTERNAL-RUN-TIME was consing.)
3926 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3927 PROCLAIM FTYPE information. Unfortunately this broke binary
3928 compatibility again, since now the forms output by DEFGENERIC
3929 to refer to functions which didn't exist in 0.6.3.
3930 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3931 can use the new (as of 0.6.3) transform for SOME into MAP into
3933 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3934 values to (INTEGER 0 1000000), so that the time code will no longer
3935 occasionally get blown up by Linux returning 1000000 microseconds
3936 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3937 its output conform to the ANSI spec. (Alas, this makes its output
3938 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3939 * A full call to MAP NIL with a single sequence argument no longer conses.
3940 * fixes to problems pointed out by Martin Atzmueller:
3941 * The manual page no longer talks about multiprocessing as though
3942 it were currently supported.
3943 * The ILISP support patches have been removed from the distribution,
3944 because as of version 5.10.1, ILISP now supports SBCL without us
3945 having to maintain patches.
3946 * added a modified version of Raymond Toy's recent CMU CL patch for
3947 EQUALP comparison of HASH-TABLE
3949 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3951 * The system still can't cross-compile itself with
3952 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3953 entails), but at least it can compile more of itself that way
3954 than it used to be able to, and various buglets which were uncovered
3955 by trying to cross-compile itself that way have now been fixed.
3956 * This release breaks binary compatibility again. This time
3957 at least I've incremented the FASL file format version to 2, so that the
3958 problem can be detected reliably instead of just causing weird errors.
3959 * various new style warnings:
3960 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3961 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3962 ** using the lexical binding of a variable named in the *FOO* style
3963 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3964 as specified by ANSI.
3965 * *RANDOM-STATE* is no longer automatically initialized from
3966 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3967 default behavior of the system is to repeat its behavior every time
3968 it's run. If you'd like to change this behavior, you can always
3969 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3970 old convention there was no comparably easy way to get the system to
3971 repeat its behavior every time it was run.
3972 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3973 FTYPE declarations has been removed, in favor of their ANSI
3974 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3975 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3976 the types of their sequence arguments can be determined at compile time.
3977 This is done through a new open code expansion for MAP which eliminates
3978 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3979 argument types can be determined at compile time.
3980 * The optimizer now transforms COERCE into an identity operation when it
3981 can prove that the coerced object is already of the correct type. (This
3982 can be a win for machine generated code, including the output of other
3983 optimization transforms, such as the MAP transform above.)
3984 * Credit information has been moved from source file headers into CREDITS.
3985 * Source file headers have been made more standard.
3986 * The CASE macro now compiles without complaining even when it has
3989 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3991 * (Note that the way that the PCL macroexpansions were rewritten
3992 to accommodate the change in DEFGENERIC below breaks binary
3993 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3994 not run under sbcl-0.6.2. Once we get out of alpha releases,
3995 i.e. hit release 1.0.0, we'll probably try to maintain binary
3996 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3997 and sbcl-1.4.4. Until then, however, it might be fairly common
3998 for maintenance releases to break binary compatibility.)
3999 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4001 * The reporting of unhandled signals has been changed to print some
4002 explanatory text as well as the report form. (Previously only
4003 the report form was printed.)
4004 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4005 it defines, so that the compiler no longer issues undefined function
4006 warnings for compiled-but-not-yet-loaded generic functions.
4007 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4008 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4009 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4010 section of the ANSI spec are supported.
4011 * The "" nickname for the "KEYWORD" package has been removed.
4012 The reader still handles symbol tokens which begin with a package marker
4013 as keywords, but it doesn't expose its mechanism for doing so in the
4014 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4015 * The system now issues STYLE-WARNINGs for contradictory TYPE
4016 proclamations. (Warnings for contradictory FTYPE proclamations would
4017 be nice too, but those can't be done usefully unless the type system
4018 is made smarter about FUNCTION types.)
4019 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4020 systematized, so that "*target-*.lisp is supposed to exist only on the
4021 target and imply that there's a related file which exists on the
4022 host, and *host-*.lisp is supposed to exist only on the host and imply
4023 that there's a related file which exists on the target. This involves a
4024 lot of renaming. Hopefully the acute confusion caused by the renaming
4025 will be justified by the reduction in chronic confusion..
4026 ** runtime-type.lisp -> early-target-type.lisp
4027 ** target-type.lisp -> late-target-type.lisp
4028 ** early-host-format.lisp -> early-format.lisp
4029 ** late-host-format.lisp -> late-format.lisp
4030 ** host-error.lisp -> misc-error.lisp
4031 ** early-error.lisp -> early-target-error.lisp
4032 ** late-error.lisp -> late-target-error.lisp
4033 ** host-defboot.lisp -> early-defboot.lisp
4034 ** code/misc.lisp -> code/target-misc.lisp
4035 ** code/host-misc.lisp -> code/misc.lisp
4036 ** code/numbers.lisp -> code/target-numbers.lisp
4037 ** code/early-numbers.lisp -> numbers.lisp
4038 ** early-host-type.lisp -> early-type.lisp
4039 ** late-host-type.lisp -> late-type.lisp
4040 ** host-typep.lisp -> typep.lisp
4041 ** load.lisp -> target-load.lisp
4042 ** host-load.lisp -> load.lisp
4043 ** host-disassem.lisp -> disassem.lisp
4044 ** host-insts.lisp -> insts.lisp
4045 ** byte-comp.lisp -> target-byte-comp.lisp
4046 ** host-byte-comp.lisp -> byte-comp.lisp
4047 ** host-signal.lisp -> signal.lisp
4048 ** host-defstruct.lisp -> defstruct.lisp
4049 ** late-target-type.lisp -> deftypes-for-target.lisp
4050 Furthermore, several other previously target-only files foo.lisp (e.g.
4051 hash-table.lisp and random.lisp) have been split into a target-and-host
4052 foo.lisp file and a target-only target-foo.lisp file, with their key type
4053 definitions in the target-and-host part, so that the cross-compiler will
4054 know more about target types.
4055 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4056 gone away. In their place are various *BACKEND-FOO* variables
4057 corresponding to the slots of the old structure.
4058 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4059 into the target SBCL has been fixed.
4060 * The chill.lisp system for loading cold code into a running SBCL
4062 * Support for the CMU CL "scavenger hook" extension has been removed.
4063 (It was undocumented and unused in the CMU CL sources that SBCL was
4064 derived from, and stale in sbcl-0.6.1.)
4065 * Various errors in the cross-compiler type system were detected
4066 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4067 (enabling various consistency checks). Many of them were fixed,
4068 but some hard problems remain, so the compiler is back to
4069 running without *TYPE-SYSTEM-INITIALIZED* for now.
4070 * As part of the cross-compiler type system cleanup, I implemented
4071 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4072 * I have started adding UNCROSS calls throughout the type system
4073 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4074 on cross-compiler input files will be able to go away, and various
4076 * CONSTANTP now returns true for quoted forms (as explicitly required
4079 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4081 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4082 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4083 (MAKE-LIST -1). In the longer term, it ought to become true
4084 that declarations are assertions even at SAFETY 1. For now, it's not
4085 quite true even at SAFETY 3, but it's at least more nearly true..
4086 (Note that this change seems to increases the size of the system by
4087 O(5%) and to decrease the speed of the compiler by 20% or more.)
4088 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4089 for the problem of printing dozens of lines of distracting information
4090 about low-level system machinery as part of the top stack frame
4091 on entry to the debugger when an undefined function was called.
4092 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4093 is not reset to COMMON-LISP-USER.
4094 * Compilation of stuff related to dyncount.lisp has been made conditional
4095 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4096 smaller. The various dyncount-related symbols have been moved into
4097 a new "SB-DYNCOUNT" package.
4098 * tty-inspect.lisp has been renamed to inspect.lisp.
4099 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4100 feature has gone away. (When we eventually port to other flavors of
4101 libc and/or Unix, we'll try to make the differences between flavors
4102 invisible at the user level.)
4103 * Various other *FEATURES* tags, and/or their associated conditionals,
4104 have been removed if obsolescent, or given better documentation, or
4105 sometimes given more-mnemonic names.
4107 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4109 * tidied up "make.sh" script
4110 * tidied up system directory structure
4111 * better "clean.sh" behavior
4112 * added doc/FOR-CMUCL-DEVELOPERS
4113 * many many small tweaks to output format, e.g. removing possibly-confusing
4114 trailing #\. character in DESCRIBE-INSTANCE
4115 * (EQUALP #\A 'A) no longer signals an error.
4116 * new hashing code, including EQUALP hashing
4117 * tidied up Lisp initialization and toplevel
4118 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4119 * command line argument processing
4120 * added POSIX-GETENV function to deal with Unix-ish environment variables
4121 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4122 terminating SBCL on EOF
4123 * non-verbose GC by default
4124 * There is no more "sbcl" shell script; the sbcl file is now the C
4125 runtime executable (just like CMU CL).
4126 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4128 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4129 symbol and package information as Lisp native symbol and package objects
4130 instead of strings naming symbols and strings naming packages. This way,
4131 whenever packages are renamed (as in warm init), debug information is
4132 transformed along with everything else.
4133 * tweaked the optimization policy declarations which control the building
4134 of SBCL itself. Now, among other things, the system no longer saves
4135 source location debugging information. (This helps two problems at once
4136 by reducing SBCL size and by keeping SBCL from trying to look for its
4137 sources -- which may not exist -- when reporting errors.)
4138 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4139 debugging and testing purposes
4140 * cleaned up printing, making the printer call PRINT-OBJECT for
4141 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4142 methods, giving nearly-ANSI behavior
4143 * converted almost all special variables to use *FOO* naming convention
4144 * deleted PARSE-TIME functionality, since it can be done portably
4145 * moved some files out of cold init into warm init
4146 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4148 * regularized formatting of source files
4149 * added an install.sh script
4150 * fixed ridiculous memory usage of cross-compiler by making
4151 compiler/alloc.lisp not try to do pooling unless it can hook
4152 itself into the GC of the cross-compilation host. Now the system
4153 builds nicely on my old laptop.
4154 * added :SB-ALLOC in target-features.lisp-expr
4155 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4156 was not implemented)
4157 * re-did condition handling and note reporting in the compiler. Notes
4158 are no longer handled by signalling conditions. Style warnings
4159 and warnings are handled more correctly and reported in such a way
4160 that it's easy to find one or the other in your output (so that you
4161 can e.g. figure out which of many problems caused COMPILE-FILE to
4163 * changed the severity of several compiler warnings from full WARNING
4164 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4165 compiler note reporting so that it doesn't use the condition system
4166 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4167 COMPILE-FILE command)
4168 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4169 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4170 no longer supported, since their implementation was deeply intertwingled
4171 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4172 * removed bogus "support" for compiler macros named (SETF FOO), and
4173 removed the compiler macro for SETF INFO (but only after making a fool
4174 of myself on the cmucl-imp mailing list by posting a bogus patch for
4175 DEFINE-COMPILER-MACRO..)
4176 * Compiled files containing forms which have side effects on the Lisp
4177 reader (such as DEFPACKAGE forms) are now handled more correctly.
4178 (Compiler queuing of top level lambdas has been suppressed by setting
4179 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4180 * deleted various currently-unused source files, e.g. gengc.lisp. They
4181 may be added back at some point e.g. when porting to other architectures,
4182 but until they are it's distracting to distribute them and to try to
4184 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4185 there were so many of them they're just distractions, and UNCROSS is
4186 known to be able to handle the current sources
4187 * moved PROFILE functionality into TRACE, so that it will be clear
4188 how the wrapping and unwrapping of functions when you profile them
4189 interacts with the wrapping and unwrapping of functions when you
4190 trace them. (Actually, the functionality isn't there yet, but at least
4191 the interface specification is there. Hopefully, the functionality will
4192 arrive with some future maintenance release.)
4193 * removed host-oops.lisp
4194 * changed signature of QUIT function to allow UNIX-CODE argument
4195 * fixed READ-SEQUENCE bug
4196 * tweaked verbose GC output so that it looks more like the progress
4197 output that ANSI specifies for functions like LOAD
4198 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4199 * added <http://sbcl.sourceforge.com> to the banner information printed by