1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
3 * incompatible change: removed writer methods for host-ent-name,
4 host-ent-addresses -- changing the values did not update the DNS
6 * minor incompatible change: changed experimental JOIN-THREAD interface
7 * documentation: the manual now lists reader and writer methods
8 in class slot documentation sections. (thanks to Richard M Kreuter)
9 * documentation: unwinding from asyncronous events has been
11 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
12 in multithreaded application code.
13 * bug fix: GC deadlocks from asynchronous interrupts has been fixed
14 by disabling interrupts for the duration of any
15 SB-SYS:WITHOUT-GCING section.
17 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
18 * new platform: experimental support for x86-64/darwin (MacOS).
19 * incompatible change: the thread-safe (on most platforms) getaddrinfo
20 and getnameinfo sockets functions are used instead of gethostbyaddr
21 and gethostbyname, on platforms where the newer functions are available.
22 As a result, the ALIASES field of HOST-ENT will always be NIL on these
24 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
25 don't print the startup banner, but behave as if --noinform was passed
26 as a command line argument. (thanks to Kevin Reid)
27 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
28 * optimization: code using alien values with undeclared types is much faster.
29 * optimization: the compiler is now able to open code SEARCH in more cases.
30 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
31 * bug fix: using standardized COMMON-LISP special variables as loop
32 variables no longer signals bogus package lock violations. (reported
34 * bug fix: declaring local loop variables to be of a range-limited type
35 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
36 (reported by Andras Simon)
37 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
38 bugs remain on x86-64.)
39 * bug fix: the #= and ## reader macros now interact reasonably with
40 funcallable instances.
41 * bug fix: type-checks for function arguments were compiled using the
42 compiler policy settings of the wrong lexical environment (in 1.0.2
44 * bug fix: SHADOW accepts characters as string designators, as required
45 by the spec (thanks to Eric Marsden)
46 * bug fix: fixed GC safety issues when foreign functions are called with
47 non-base strings as arguments
48 * bug fix: more consistent error output for fatal-compiler-warnings, like
50 * bug fix: functions with proper names were showing up as NIL in
51 backtraces (reported by Edi Weitz, regression in 1.0.2)
53 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
54 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
55 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
56 produce more efficient code on the x86 if the compiler can determine
57 that (+ INDEX <constant>) does not require a bounds check and FOO
58 has an element type at least 8 bits wide.
59 * bug fix: references to undefined variables are handled the same way
60 in toplevel forms as in the normal compiler
61 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
63 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
64 (thanks to Jon Buller)
65 * improvement: faster compilation times for complex functions
66 * improvement: added readlink support to SB-POSIX (thanks to Richard
69 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
70 * improvement: experimental support for mach exception handling on
71 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
73 * improvement: support for GBK external format.
74 (thanks to Chun Tian (binghe))
75 * improvement: the debugger now displays variables that have been closed
76 over, in code compiled with (DEBUG 2) or higher
77 * improvement: support for executable cores on NetBSD (thanks to
79 * new feature: added a RESTART-FRAME debugger command
80 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
81 be used to provide an implementation for FILE-POSITION on Gray streams
82 (thanks to Eric Marsden)
83 * new feature: add syslog, openlog and closelog support to SB-POSIX
84 (thanks to Richard Kreuter)
85 * optimization: the function call overhead in code compiled with
86 a high DEBUG optimization setting is significantly
87 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
88 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
89 (thanks to Tony Martinez)
90 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
91 evaluated once (reported by Kevin Reid)
92 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
93 works again (thanks to Kilian Sprotte)
94 * bug fix: an error is signaled for tagbodies with duplicate tags
95 (thanks to Stephen Wilson)
96 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
97 * bug fix: Win32 port can now handle foreign code unwinding Lisp
98 stack frames from alien callbacks.
99 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
100 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
101 65k pages (thanks to David Woodhouse)
102 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
104 changes in sbcl-1.0.1 relative to sbcl-1.0:
105 * new platform: FreeBSD/x86-64, including support for threading.
106 * new feature: the compiler stores cross-referencing information
107 abount function calls (who-calls), macroexpansion (who-macroexpands)
108 and special variables (who-binds, who-sets, who-references) for code
109 compiled with (< SPACE 3). This information is available through the
110 sb-introspect contrib.
111 * new feature: users may subclass SEQUENCE, and have instances of
112 these classes interoperate with standard Common Lisp functions if
113 a number of methods are defined. (This feature is experimental
114 and the interface subject to change based on feedback from SBCL
115 users and the general community)
116 * improvement: sb-sprof traces call stacks to an arbitrary depth on
117 x86/x86-64, rather than the previous fixed depth of 8
118 * improvement: another pthread back-end of mutex "pthread-futex".
119 * bug fix: non-ascii command-line arguments are processed correctly
120 (thanks to Yaroslav Kavenchuk)
121 * bug fix: non-required arguments were not passed correctly when a method
122 defined using DEFMETHOD was called from a mop-generated method using
123 CALL-NEXT-METHOD (reported by Pascal Costanza)
124 * bug fix: recursion is now permitted in accessors through
125 SLOT-UNBOUND. (reported by Pascal Costanza)
126 * bug fix: an error was signaled at startup if the HOME environment
127 variable was defined, but had an empty value (reported by Peter Van Eynde)
128 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
129 signal a PROGRAM-ERROR, not a BUG.
130 * bug fix: SB-POSIX:READDIR works when built with large file support.
131 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
133 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
134 for better automatic indentation support. (thanks to Matt Pillsbury)
135 * optimization: loading generic functions no longer takes O(n^2) time,
136 proportional to the amount of methods in the generic function
137 (reported by Todd Sabin and Jeremy Brown)
138 * optimization: the FIND and POSITION family of sequence functions
139 are significantly faster on arrays whose element types have been
141 * improvements to the Windows port:
142 ** Intermittent heap corruption problems have been fixed. (thanks
143 to Alastair Bridgewater)
144 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
146 ** Lisp is able to unwind foreign exception frames from alien
147 callbacks. (thanks to Alastair Bridgewater)
149 changes in sbcl-1.0 relative to sbcl-0.9.18:
150 * improvement: experimental support for threading on FreeBSD/x86.
151 (thanks to NIIMI Satoshi)
152 * improvement: runtime option --dynamic-space-size can be used
153 to set the size of the dynamic space reserved on startup.
154 * improvement: floating point modes in effect are now saved in
155 core, and restored on startup.
156 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
157 startup, not time since first call to GET-INTERNAL-REAL-TIME.
158 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
159 threads are not running after *SAVE-HOOKS* have run.
160 * improvement: writes to CLOS instance slots are type-checked in code
161 compiled with (SAFETY 3)
162 * improvement: floating-point exception handling on FreeBSD (thanks to
164 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
165 (thanks to Zach Beane)
166 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
168 * improvement: added support for the Shift-JIS external format.
169 (contributed by NIIMI Satoshi)
170 * improvement: callbacks are supported on Linux/PPC. (thanks to
172 * bug fix: compiler bug triggered by a (non-standard) VALUES
173 declaration in a LET* was fixed. (reported by Kaersten Poeck)
174 * bug fix: file compiler no longer confuses validated and already
175 dumped structurres. (reported by Kaersten Poeck)
176 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
177 fill-pointer signals a type-error as required. (thanks to
179 * bug fix: disassemly of funcallable instances works.
180 * bug fix: single stepping on PPC.
181 * bug fix: fix thread-safety problems in the type system (generally
182 manifesting as nonsensical errors like "STRING is a bad type specifier
183 for sequences" or "The value 1 is not of type FIXNUM")
184 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
185 (reported by Josip Gracin)
186 * bug fix: an error is signaled for attempts to displace arrays with
187 incompatible element types (thanks to Mario Mommer)
188 * bug fix: more correct handling of wide characters in debug info
189 (bug reported by Attila Lendvai and fixed by Juho Snellman)
190 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
191 and don't cause extra consing
192 * optimization: MAP and MAP-INTO are significantly faster on vectors
193 whose elements types have been declared.
194 * Improvements to SB-SPROF:
195 ** Support for allocation profiling
196 ** Reduced profiling overhead, especially for long profiling runs
197 * Improvements to the Windows port:
198 ** floating point exceptions are now reported correctly.
199 ** stack exhaustion detection works partially.
200 ** more accurate GET-INTERNAL-REAL-TIME.
201 ** better RUN-PROGRAM behaviour in case of errors in the child
203 ** PROBE-FILE now simplifies pathnames correctly.
204 ** DIRECTORY now works correctly with :WILD-INFERIORS.
205 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
206 1st 1970 (although time timezone data will be unreliable).
207 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
209 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
210 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
211 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
213 * bug fix: two potential GC deadlocks affecting threaded builds.
214 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
215 returns T (reported by Anton Kazennikov)
216 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
217 with non-variable places
218 * bug fix: remove a race condition in the setting of
219 funcallable-instance functions, this should make threaded CLOS
220 code more stable against memory faults.
221 * bug fix: corruption of specials when unbinding is interrupted by an
222 asynchronous unwind (reported by Hannu Koivisto)
223 * improvement: the debugger will now also display local variables that
224 are only used once, for code compiled with a DEBUG optimization quality
227 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
228 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
229 * incompatible change: External-format support for FFI calls. The
230 SB-ALIEN:C-STRING no longer implies an ASCII
231 external-format. Instead the string is subject to external-format
232 conversion. Additionally return values of type SB-ALIEN:C-STRING
233 are by default of type (SIMPLE-ARRAY CHARACTER), not
234 SIMPLE-BASE-STRING. For an alien type that behaves like the old
235 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
236 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
237 * incompatible change: SB-EXT package no longer contains the
238 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
239 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
241 * incompatible change: the single-stepper is no longer available
242 on Alpha, Mips and Sparc platforms.
243 * minor incompatible change: the direct superclasses of
244 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
245 not (STANDARD-OBJECT FUNCTION). This makes the
246 class-precedence-lists of GENERIC-FUNCTION and
247 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
249 * minor incompatible change: the default stream external format on
250 non-unicode SBCL is no longer detected from the locale, but is always
252 * new feature: Add a version of evaluator that uses an interpreter instead
253 of the compiler. EVAL still uses the compiler by default, to switch it
254 to use the interpreter, set the value of the variable
255 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
256 * minor incompatible change: the single-stepper REPL has been merged
257 with the normal debugger (see the "Stepping" heading of the debugger help
258 for more details). The debugger command STEP will no longer switch
259 to the single-stepper REPL.
260 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
261 for a type now works.
262 * bug fix: loading of default sysinit file works. (thanks to Leonid
264 * bug fix: better detection of circularities in the file-compiler.
265 (reported by Marco Monteiro)
266 * bug fix: the CL pathname functions now work with files that have
267 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
268 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
269 systems with Unix98 pty semantics.
270 * bug fix: ASDF-INSTALL will now work with bsd tar.
271 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
273 * bug fix: timers expiring in dead threads no longer cause a
274 type-error (reported by Paul "Nonny Mouse").
275 * bug fix: thanks to more lightweight single-stepper instrumentation,
276 code compiled with (DEBUG 3) will compile and execute significantly faster,
277 and will have more accurate type-inferencing than before
278 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
279 whose bindings are modified
280 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
281 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
282 ** replace dummy implementations of CL:MACHINE-INSTANCE and
283 CL:SOFTWARE-VERSION with proper version
285 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
286 * feature: implemented the READER-METHOD-CLASS and
287 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
288 as specified by AMOP.
289 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
291 * optimization: faster LOGCOUNT implementation on x86 and x86-64
292 (thanks to Lutz Euler)
293 * optimization: hashing of general arrays and vectors has been
294 improved. (reported by Any Fingerhut)
295 * enhancement: SB-INTROSPECT is now able to find definitions of
296 profiled functions. (thanks to Troels Henriksen)
297 * enhancement: compiler-macro expansion applies now to FUNCALL forms
298 as well, allowing compiler-macros for SETF-functions to expand.
299 * enhancement: step-instrumentation no longer wraps known
300 single-value functions with multiple-value context, allowing
301 better type inference.
302 * fixed bug #337: use of MAKE-METHOD in method combination now works
303 even in the presence of user-defined method classes. (reported by
304 Bruno Haible and Pascal Costanza)
305 * fixed bug #339(c): if there are applicable methods not part of any
306 long-form method-combination group, call INVALID-METHOD-ERROR.
307 (reported by Bruno Haible)
308 * fixed bug #361: the :FUNCTION initarg in the protocol for
309 initialization of methods can now be used to override
310 internally-produced optimized functions. (reported by Bruno
312 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
313 system-provided lambda expression no longer cause warnings about
314 unbound #:|pv-table| symbols.
315 * bug fix: improved the handling of type declarations and the
316 detection of violations for keyword arguments with non-constant
318 * bug fix: potentially erronous calls to PATHNAME and
319 MERGE-PATHNAMES were being flushed in some cases.
320 (reported by Richard Kreuter)
321 * bug fix: compiled calls to TYPEP were mishandling obsolete
322 instances. (reported by James Bielman and Attila Lendvai)
323 * bug fix: format strings with ~> without matching ~< no longer
324 trigger an AVER, but signal an understandable error instead.
325 (reported by Antonio Martinez)
326 * bug fix: specifying an output-file without a pathname-type for
327 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
328 of input-file instead of "fasl". (reported by Robert Dodier)
329 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
330 (reported by James Y Knight).
331 * bug fix: compiler-macros-function did not consider the environment
332 argument for shadowing by local functions.
333 * bug fix: compiler-macros expansion was inhibited by local INLINE
335 * bug fix: inline expansions of known functions were subject to
336 step-instrumentation in high debug policies, leading to problems
338 * bug fix: compiler failed to differentiate between different CONS
340 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
341 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
342 * bug fix: #368: incorrect use of expressed vs. upgraded array
344 * bug fix: #306a: more precise unions of array types.
345 * thread-safety improvements:
346 ** CONDITION-WAIT could return early on Linux, if the thread was
347 interrupted and subsequently continued with SIGCONT.
348 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
350 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
351 * added support for the ucs-2 external format. (contributed by Ivan
353 * minor incompatible change: pretty printing of objects of type
354 (cons symbol) is, in the default pprint-dispatch-table, now
355 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
357 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
358 called later in a class's lifetime, possibly as late as when the
359 first instance of the class is created. Previously,
360 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
361 class became finalizeable.
362 * fixed bug: FILE-POSITION sometimes returned inconsistent results
363 for multibyte external-format streams. (thanks to "vbzoli")
364 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
365 with :ALLOCATION :CLASS inherited from superclasses of the
367 * fixed bug: anonymous classes can now be created using the :NAME
368 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
369 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
370 * fixed bug: core-files saved with :EXECUTABLE T can again be
371 executed when SBCL_HOME isn't set. (reported by James Knight)
372 * fixed bug: toplevel LOCALLY forms with declarations could
373 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
374 * fixed bug: printing from several different threads using different
375 values of *print-case* could cause invalid output, due to
376 some internal special variables of the printer not being bound
377 thread-locally (reported by Max Mikhanosha)
378 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
379 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
380 (reported by Pascal Costanza)
381 * fixed bug: FINALIZE-INHERITANCE is called from
382 REINITIALIZE-INSTANCE on classes when the class has previously
383 been finalized, as required by AMOP.
384 * minor code generation optimizations:
385 ** better register allocation in CLOS dispatching functions
386 ** overflow detection when coercing signed bytes to fixnums on x86-64
387 ** is now implemented with one IMUL instruction instead of three shifts
388 ** more efficient bit-vector access on x86 and x86-64
389 ** more efficient access to raw structure slots on x86 and x86-64
390 * fixed some bugs revealed by Paul Dietz' test suite:
391 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
394 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
395 * feature: thread support on Solaris/x86, and experimental thread support
397 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
399 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
400 default method for SLOT-UNBOUND.
401 * minor incompatible change: prevent the user from specializing the
402 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
403 somewhat counter to the intent of the protocol, I (CSR) think, and
404 additionally it just doesn't work in SBCL as currently
405 implemented, thanks to optimizations (that are always valid for
406 the other three SLOT-VALUEish functions, but not for the setter).
407 * minor incompatible change: the :SB-LDB feature is now enabled by
408 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
409 the low-level debugger.
410 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
411 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
412 down to an SB-SYS:FD-STREAM, allowing direct communication in
413 more cases. Temporary files and pipes are still used as a fallback
415 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
416 on the non-reentrant C functions localtime(3) and gmtime(3).
417 * bug fix: global optimization policy was not visible in LOCALLY and
419 * bug fix: class objects can be used as specializers in methods.
420 (reported by Pascal Costanza)
421 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
422 with a type error. (reported by blitz_ on #lisp)
423 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
424 failed with a type error. (reported by Pascal Bourguignon)
425 * bug fix: merging pathnames against defaults with :DIRECTORY
426 starting with '(:RELATIVE :BACK) should preserve the :BACK.
427 (reported by James Y Knight)
428 * bug fix: saving large (>2GB) cores on x86-64 now works
429 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
430 constant index and a (simple-array (signed-byte 32)) array
431 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
432 error (patch by Robert J. Macomber)
433 * feature: TIME now displays the amount of run-time spent in GC
434 * bug fix: The debugger now does a better job of respecting
435 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
436 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
437 workaround for bug 403.)
438 * bug fix: floating point exception handling now works on FreeBSD
439 (thanks to NIIMI Satoshi)
440 * fixed some bugs revealed by Paul Dietz' test suite:
441 ** MISC.641: LET-conversion were not supposed to work in late
443 * improvements to the Win32/x86 port:
444 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
445 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
447 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
448 * new feature: source path information is generated for macro-expansion
449 errors for use in IDE's like Slime (thanks to Helmut Eller)
450 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
451 * bug fix: compiler does not loop forever on an invalid type in
453 * improvement: compilation of most CLOS applications is significantly
455 * optimization: added a limited bytecode compiler for simple toplevel
456 forms, speeding up compilation and FASL loading
457 * bug fix: the statistical profiler now properly distinguishes anonymous
460 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
461 * minor incompatible change: in sbcl-0.9.11 (but not earlier
462 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
463 into subclasses of the finalized class. Now user calls to
464 FINALIZE-INHERITANCE finalize just the one class, and calls by the
465 system return before any subclasses are finalized.
466 * minor incompatible change: The reader no longer ignores errors
467 regarding non-existent packages in #+ and #- feature tests.
468 * new feature: command line options --no-sysinit, --no-userinit to
469 inhibit loading the corresponding init files
470 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
471 for sending data through UDP sockets (thanks to François-René Rideau)
472 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
473 error is signalled instead (thanks to François-René Rideau)
474 * minor incompatible change: Error signalling behaviour of lexical
475 operations violating package locks has changed slightly. Refer to
476 documentation on package locks for details.
477 * enhancement: EVAL can process IF-expressions without resorting to the
479 * bug fix: Some operations involving SIN, COS, or TAN weren't being
480 constant-folded properly due to missing out-of-line functions.
481 (reported by Mika Pihlajamäki)
482 * bug fix: LISTEN sometimes returned T even in cases where no data was
483 immediately available from the stream
484 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
485 were reversed. (reported by Levente Mészáros)
486 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
487 when given a vector displaced to an adjustable array. (reported
489 * bug fix: derivation of float boundaries from numbers outside the
490 appropriate float range (reported by John Wiseman)
491 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
493 * bug fix: precomputing cache entries for generic functions with
494 some subclasses of specializers as yet invalid does not attempt to
495 fill a cache line with a negative offset. (reported by Levente
497 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
498 allow condition class objects as arguments to DOCUMENTATION and
499 (SETF DOCUMENTATION); only find and set documentation for
500 structure names for the STRUCTURE doc-type. (suggested by Gary
502 * improvements to the Win32/x86 port:
503 ** (user-homedir-pathname) and default initialization file
504 locations now know about the user's "Documents and Settings"
505 directory (thanks to Yaroslav Kavenchuk)
506 ** run-program is implemented (thanks to Mike Thomas)
507 ** sockets support (thanks to Timothy Ritchey)
508 ** better backtrace support (thanks to Alastair Bridgewater)
509 ** sb-grovel supported
510 ** asdf-install and sb-posix work somewhat
511 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
512 * improvements to the Solaris/x86 port:
513 ** works on Solaris 11/Solaris Express
514 ** floating-point exception handling support
515 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
516 * fixed some bugs revealed by Paul Dietz' test suite:
517 ** REMOVE-METHOD returns its generic function argument even when
518 the method is not one of the generic functions' methods.
519 ** objects declared as MEMBER types can be admissible arguments to
521 ** printing characters should simply be printed by the FORMAT ~:C
523 ** compiler failure when compiling functions with hairy constant
524 defaults for optional parameters.
525 ** compiler produces wrong code when MAYBE-INLINE-expanding a
526 function, which is already optimized.
528 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
529 * new platform: experimental support for SBCL x86/Darwin, including
530 MacOS X 10.4.5 on Intel.
531 * new feature: Unicode character names are now known to the system
532 (through CHAR-NAME and NAME-CHAR).
533 * new feature: the filesystem location of SBCL's core file is
534 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
535 * minor incompatible change: the contrib modules SB-POSIX and
536 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
537 this change is to make it easier to distribute
538 location-independent binaries.
539 * bug fix: as implied by AMOP, standardized classes no longer have
540 slots named by external symbols of public packages. (reported by
542 * optimization: calling functions via constant symbols -- as in
543 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
544 via the function object as in (FUNCALL #'FOO).
545 * optimization: on x86 and x86-64, the calling convention for the
546 case of dealing with an unknown number of values has been altered
547 to be friendlier to the prediction heuristics implemented,
548 particularly with reference to CALL and RET pairing. (thanks to
549 Alastair Bridgewater)
550 * optimization: on x86, the code for access to thread-local symbol
551 values and binding/unbinding of thread-local symbols is smaller.
552 (thanks to Alastair Bridgewater)
553 * enhancement: CONSTANTP is now able to determine constantness of
554 more complex forms, including calls to constant-foldable standardized
555 functions and some special forms beyond QUOTE.
556 * fixed bug: occasional GC crashes on Solaris/x86
557 * optimization: x86-64 supports stack allocation of results of simple
558 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
559 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
560 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
561 (thanks to James Knight)
562 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
563 by the standard (thanks to Stephen Compall)
565 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
566 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
567 be used for bundling the runtime and the core file into one
568 executable binary. This feature is not currently supported on all SBCL
569 platforms. (thanks to James Bielman and NIIMI Satoshi)
570 * new feature: a generational or ephemeral garbage collector is now
571 the default on the PowerPC platform (both Linux and Darwin). The
572 old Cheney (stop and copy) collector is a build-time option.
573 (thanks to Cyrus Harmon, after Raymond Toy)
574 * minor incompatible change: the method by which SBCL finds its
575 contributed modules has changed; it no longer relies on symbolic
576 links from an $SBCL_HOME/systems directory, but searches directly
577 in the subdirectories of $SBCL_HOME.
578 * enhancement: the dynamic heap size on the Linux/PPC platform is
579 markedly larger, even using the older Cheney garbage collector.
580 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
581 certain complicated string types. (reported by Gary King)
582 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
583 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
584 * fixed bug: DOCUMENTATION on structures no longer signals an
585 error if no structure type of the right name exists. (reported by
587 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
588 applicable non-standard primary or :AROUND method on
589 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
590 no longer cause unbound variable errors. (reported by Kevin Reid)
591 * fixed bug: metacircle resolution in cases where methods have slots
592 added before the slots from STANDARD-METHOD. (reported by Jean
594 * fixed bug: the Power PC assembly code for calling into lisp
595 sometimes computed its return address wrongly (depending on the
596 15th bit of the address, fixed at link time). This used to
597 manifest itself in a segmentation violation while building PCL.
598 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
599 many others over the years)
600 * fixed bug: ROOM no longer reports silly numbers for stack usage if
601 the stack pointer (treated as a signed integer) is negative.
602 (thanks to Peter van Eynde)
604 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
605 * new platform: experimental support for the Windows operating
606 system has been added. (thanks to Alastair Bridgewater)
607 * fixed several bugs in and robustified the PPC FFI (including
608 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
609 * bug fix: allow non-simple string symbol names (reported by Paul
611 * bug fix: interrupt handling on NetBSD (thanks to Richard M
613 * bug fix: saving a core corrupted callbacks on x86/x86-64
614 * bug fix: closed a loophole in metacircularity detection and
615 grounding in the PCL implementation of CLOS.
616 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
617 (reported by tomppa on #lisp)
618 * bug fix: clear the direction flag on Lisp -> C transitions, as
619 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
620 (reported by Andrej Grozin and Hendrik Maryns)
621 * optimization: major improvements to GC efficiency on GENCGC platforms
622 * optimization: faster implementation of EQUAL
623 * optimization: emit more efficient opcodes for some common
624 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
625 * optimization: several other minor code-generation improvements on x86-64
626 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
628 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
629 * minor incompatible change: (SETF CLASS-NAME) and (SETF
630 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
631 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
632 by ANSI as a generic function, and for consistency (SETF
633 GENERIC-FUNCTION-NAME) is treated likewise.
634 * fixed bug #233.b: make constraint propagation notice when a variable
635 value is changed after it is referenced but before it is used
636 * fixed bug #296: no more arbitrary behaviour from filenames with
637 odd characters as --load arguments. (reported by Adam Warner)
638 * fixed bug #390: :CHARACTER-SET pathname components now work as
639 expected. (reported by Tim Daly Jr)
640 * fixed bug #391: complicated :TYPE intersections in slot
641 definitions no longer cause an error in PCL internals.
642 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
643 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
644 returning the number of octets which would be written to the
645 file-stream. (thanks to Robert J. Macomber)
646 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
647 arguments to the internal stream functions. (thanks to David
649 * bug fix: stack exhaustion is now handled on the x86/SunOS
650 platform. (thanks to Daisuke Homma)
651 * bug fix: more accurate ROOM results on GENCGC platforms
652 * optimization: improved type inference for arithmetic-for
653 index variables in LOOP
654 * optimization: faster floating-point SQRT on x86-64
655 * fixed some bugs revealed by Paul Dietz' test suite:
656 ** DOCUMENTATION returns NIL instead of "" for method combinations
657 that don't have a docstring
659 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
660 * minor incompatible change: (SETF CLASS-NAME) and (SETF
661 GENERIC-FUNCTION-NAME) are no longer generic functions, and
662 therefore are not customizeable by user code (as seems to be at
663 least permitted and maybe required by AMOP). As a consolation,
664 however, the SBCL implementation of these functions now calls
665 REINITIALIZE-INSTANCE as specified by AMOP.
666 * bug fix: slot-definition documentation is propagated as per ANSI
667 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
668 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
669 argument, as do eslotd initialization methods. (from Pascal
670 Costanza's "Closer" project)
671 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
672 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
674 * bug fix: it is now possible to have more than one subclass of
675 STANDARD-GENERIC-FUNCTION without causing stack overflow.
676 (reported by Bruno Haible, Pascal Costanza and others)
677 * bug fix: the dependent update protocol now works for generic
678 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
680 * bug fix: condition-class instances corresponding to
681 DEFINE-CONDITION forms are now created eagerly. (reported by
682 Kalle Olavi Niemitalo on comp.lang.lisp)
683 * bug fix: floating point printing is more accurate in some
684 circumstances. (thanks to Simon Alexander)
685 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
686 merged with *DEFAULT-PATHNAME-DEFAULTS*.
687 * bug fix: callbacks on OS X now preserve stack-alignment.
688 (thanks to Cyrus Harmon)
689 * enhancement: the x86-64 disassembler is much better at
690 disassembling SSE instructions. (thanks to Lutz Euler)
691 * enhancement: CHECK-TYPE on a variable now allows type-inference
692 (thanks to Vincent Arkesteijn)
693 * optimization: improved performance of EUC-JP external format.
694 (thanks to NIIMI Satoshi)
695 * optimization: performance improvements to IO on file streams of
696 :ELEMENT-TYPE CHARACTER
697 * optimization: much faster memory allocation on x86-64
698 * optimization: faster garbage collections (latency and throughput)
700 * optimization: faster bignum subtraction and division on x86 and x86-64
701 (thanks to Lutz Euler)
702 * optimization: more accurate type-inference for loops that use a
703 floating point index variable or a negative step.
705 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
706 * bug fix: add a workaround to SBCL looping infinitely at startup on
707 Linux kernels with apparently buggy implementations of personality().
708 (thanks to Svein Ove Aas)
709 * bug fix: Unicode symbols are correctly printed in LDB backtraces
710 (thanks to David Lichteblau)
711 * bug fix: local bindings shadow global symbol macros for
712 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
713 * optimization: non-open coded uses of numeric comparison operators
714 (e.g. >) no longer cons when called with more than one parameter
715 on platforms supporting dynamic-extent allocation.
716 * enhancement: saving cores with foreign code loaded is now supported
717 on MIPS/Linux in addition to the previously supported platforms.
718 * bug fix: FIND-RESTART now tests for activity, not applicability when given
719 a restart object as identifier. (reported by Helmut Eller for CMUCL)
720 * bug fix: division by zero in sb-sprof when no samples were collected
721 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
722 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
723 (thanks to Alexander Kjeldaas)
724 * new feature: ignoring the return values of destructive functions
725 like NREVERSE signals a compile-time style-warning.
726 (thanks to Kevin Reid)
727 * new feature: the alignment of alien structure fields can be
728 explicitly specified. (thanks to Cyrus Harmon)
729 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
730 and dump core on SIGQUIT
732 ** incompatible change: threads do not inherit values of specials
733 from their parents (see manual)
734 ** bug fix: threads stacks belonging to dead threads are freed by the
735 next exiting thread, no need to gc to collect thread stacks anymore
736 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
737 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
738 ** bug fix: dynamic variable and thread start related gc lossage
739 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
741 * fixed some bugs revealed by Paul Dietz' test suite:
742 ** SUBTYPEP is slightly more accurate on heinously complicated
743 CONS types where some of the members have uncertain (in the
744 NIL, NIL sense) type relationships to each other.
746 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
747 no longer purified unless :PURIFY T is explicitly specified.
748 ** Non-purified cores are significantly smaller than before
750 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
751 * new feature: timers based on Zach Beane's excellent timer package
752 * added support for the following external formats: koi8-u,
753 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
754 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
755 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
756 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
757 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
758 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
759 * incompatible change: a threaded SBCL will no longer revert to
760 non-threaded mode on non-NPTL systems, but refuse to start entirely.
761 * bug fix: interrupts are disabled until startup is complete; no
762 more sigsegvs when receiving a signal to soon
763 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
765 * bug fix: add a workaround for the memory randomization features in
766 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
767 workaround will only be in effect on systems with the proc filesystem
769 * bug fix: printing objects of type HASH-TABLE signals a
770 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
772 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
773 have been running for over 50 days. (reported by Gilbert Baumann)
774 * bug fix: the logic for getting names of functions gets less
775 confused when confronded with alternate-metaclass
776 funcallable-instances. (reported by Cyrus Harmon)
777 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
778 consistent, even on internal alternate-metaclass objects.
779 * bug fix: flush closure information collected by physical
780 environment analysis prepass before the main pass. (bug reported
782 * bug fix: compiler pack phase does not modify a hashtable, which is
783 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
784 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
785 may be expanded. (reported by Alexander <malishev> on c.l.l)
786 * bug fix: no more style-warnings when compiling code using LAST
787 under high-SPEED low-SPACE settings. (reported by David Wallin
789 * fixed some incompatibilities between SBCL's MOP and the MOP
791 ** the METAOBJECT class is now implemented;
792 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
793 STANDARD-OBJECT, as required;
794 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
795 now compatible (as required by VALIDATE-SUPERCLASS); there
796 remains a consistency requirement over the presence or absence
797 of the FUNCTION class in the superclass list of finalized
798 classes; see the manual for more details;
799 ** the :AROUND method for COMPUTE-SLOTS on
800 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
801 requested slot ordering.
803 ** bug fix: parent thread now can be gc'ed even with a live
805 ** bug fix: binding a special with PROGV to no value is not
806 the same as that symbol not having been bound (thanks to
808 ** bug fix: binding specials is thread safe (thanks to
810 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
811 and SB-SPROF do not die with 'no handler for signal XX in
812 interrupt_handle_now(..)' anymore
813 ** bug fix: WITH-TIMEOUT works with multiple running threads
814 * fixed some bugs revealed by Paul Dietz' test suite:
815 ** ENSURE-GENERIC-FUNCTION should take a method class object for
816 the :method-class keyword argument.
818 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
819 * new port: the Solaris operating system on x86 processors is now
820 mostly supported, though some rough edges in the environment
821 remain. (thanks to Daisuke Homma)
822 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
823 space for its heap. (thanks to Thiemo Seufer)
824 * optimization: SBCL on MIPS platforms now supports dynamic-extent
825 closures. (thanks to Thiemo Seufer)
826 * minor incompatible change: eof selects abort in the debugger.
827 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
828 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
830 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
831 deprecated, and will go away in a future revision of SBCL.
832 * minor incompatible change: GC-ON and GC-OFF are no longer
833 implemented with a counter, it does not matter how many times gc
834 is switched on or off
835 * bug fix: discriminating functions for generic function classes
836 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
837 make invalid assumptions about method precedence order. (reported
839 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
840 platforms. (thanks to Thiemo Seufer)
841 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
842 * bug fix: correct pathname printing with printer escaping is on.
843 (thanks to Kevin Reid)
844 * bug fix: complex VOP definitions in "user-space" no longer trigger
845 package locks. (reported by Zach Beane)
846 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
847 for funcallable-instance functions, and is no different from
849 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
850 characters in the current readtable. (reported by Nicholas Neuss)
851 * bug fix: SUBTYPEP on various CONS types returns more a more
852 accurate acknowledgment of its certainty.
853 * optimizations: REMOVE-DUPLICATES now runs in linear time on
854 lists in some cases. This partially fixes bug 384.
855 * flush all standard streams before prompting in the REPL and the
857 * bug fix: signal handling and triggering gc do not conflict
858 directly anymore, in particular a high frequency sb-sprof does
859 not prevent gc from running
860 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
861 approximation for timezone and DST information between the
862 universal time epoch and the smallest negative 32-bit time_t.
863 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
864 year 1899 so long as the encoded time is non-negative.
865 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
866 the correct answer for non-constant rotate amounts when performing
867 an inline 32-bit rotation.
869 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
870 there is only one thread in the session
871 ** bug fix: memory leak for streams created in one thread and
872 written to in another
873 ** bug fix: lockup when compiled with gcc4
874 ** bug fix: race that allows the gc to be triggered when gc is
876 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
877 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
879 ** bug fix: finished threads are gc'ed properly
880 * fixed some bugs revealed by Paul Dietz' test suite:
881 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
882 has arguments with a different set of applicable methods from
883 the orignal arguments.
884 ** The type error thrown by MAP now has a correct expected-type
886 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
887 name a compiled function.
888 ** Three MISC tests where a large bignum was improperly coerced to
889 a float that couldn't represent that bignum during type
890 derivation were fixed.
891 ** SUBTYPEP can now handle types involving the negation of a
892 list-form FUNCTION type.
893 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
894 as (COMPLEX (AND RATIO (NOT FIXNUM))).
895 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
897 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
898 * New feature: Experimental support for bivalent streams: streams
899 opened with :element-type :default now allow character and binary
900 (unsigned-byte 8) I/O
901 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
902 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
904 * bug fix: no more highly sporadic "couldn't check whether ~S is
905 readable" when reading a stream and an interrupt hits in the middle
906 of a select system call
907 * compiler better recognizes complex arrays (reported by Eduardo
909 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
910 for variables on SBCL built with linkage-tables. (reported by Luis
912 * various error reporting improvements.
913 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
914 (Thanks to Thiemo Seufer)
915 * bug fix: nested reader invokations work correctly
916 * bug fix: it is possible to have simultaneous references to foreign
917 code and foreign data with the same name.
919 ** added x86-64 support
920 ** incompatible change: the threading api now works with thread
921 objects instead of thread ids
922 ** bug fix: threads are protected from signals and interruption when
923 starting up or going down
924 ** bug fix: a race where an exiting thread could lose its stack to gc
925 ** bug fix: don't halt on infinite error in threads if possible
926 ** fixed numerous gc deadlocks introduced in the pthread merge
927 ** bug fix: fixed thread safety issues in read and print
928 ** bug fix: debugger doesn't hang on session lock if interrupted at
929 an inappropriate moment
930 ** bug fix: run-program is now thread safe(r)
931 ** bug fix: inner with-recursive-lock no longer releases the mutex
932 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
933 * fixed some bugs revealed by Paul Dietz' test suite:
934 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
935 EXPECTED-TYPE slots filled.
936 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
938 ** MAKE-CONDITION accepts classes as type-designators.
939 ** COMPILE may never return NIL.
940 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
941 range before calling Unix time functions
943 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
944 * numerous signal handling fixes to increase stability
945 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
946 * minor incompatible change: we now correctly canonize default
947 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
948 as per the MOP, rather than the historical (INITARG INITFUNCTION
949 INITFORM). (reported by Bruno Haible)
950 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
951 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
952 for more information.
953 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
954 pathname is a directory pathname.
955 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
956 * dynamic space size on PPC has been increased to 768Mb. (thanks to
958 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
959 :METACLASS argument in addition to a class name. (reported by
960 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
961 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
962 (reported by Bruno Haible; patch from Gerd Moellmann)
963 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
965 * bug fix: more cleanups to the floating point exception handling on
966 x86-64 (thanks to James Knight)
967 * bug fix: the compiler does not try to constant fold calls of
968 COERCE and BIT-* functions when they return freshly constructed
969 objects. (reported by Nikodemus Siivola and Paul Dietz)
970 * optimization: improved the MIPS versions of generic (in the
971 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
972 * optimization: direct conversion of (unsigned-byte 32) to floats on
973 the PowerPC platform.
974 * optimization: structure instances with raw slots now use less
975 memory, and probably show better memory locality. (thanks to
977 * optimization: DYNAMIC-EXTENT declarations for lists and closures
978 are treated as requests for stack allocation on the x86-64,
979 Alpha-32, PPC and SPARC platforms.
980 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
981 the wrong file descriptor; implementation of SOCKET-OPEN-P.
982 (thanks to Tony Martinez)
984 ** gcing a dead thread can no longer lead to lockups
985 ** threads block signals until they are set up properly
986 ** errno is no longer shared by threads
987 ** interrupt-thread restores the eflags register on x86
988 ** fixed some lockups due to gc/thread interaction
989 * fixed some bugs revealed by Paul Dietz' test suite:
990 ** invalid dotted lists no longer raise a read error when
992 ** signal an error if a symbol that names a declaration is used
993 as the name of a type, or vice versa
994 ** allow using the (declare (typespec var*)) abbreviation for
995 (declare (type typespec var*)) with all type specifiers
996 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
997 ** always return NIL from PPRINT-POP when OBJECT is NIL
998 ** don't signal errors when pretty-printing malformed LABELS,
999 FLET or MACROLET forms
1000 ** declarations in a DOLIST body are also in scope for the
1002 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1004 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1007 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1008 * fixed cross-compiler leakages that prevented building a 32-bit
1009 target with a 64-bit host compiler.
1010 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1011 opened with :IF-EXISTS :OVERWRITE.
1012 * fixed bug 281: error for an invalid qualifier in a short-form method
1013 combination method is not signalled until the faulty method is called.
1014 * bug fix: iteration variable type inferrer failed to deal with open
1015 intervals. (reported by Alan Shields)
1016 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1017 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1018 or more runtime options were provided to the sbcl binary.
1019 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1020 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1021 * optimization: there's now a fast-path for fixnum arguments in the
1022 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1023 * optimization: the code generated on x86-64 is more compact thanks
1024 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1025 * bug fix: floating-point exception handling now partly works on
1026 x86-64. (Thanks to James Knight)
1027 * improvement to the MIPS backend from Thiemo Seufer: C-style
1028 64-bit long long arguments and return values to alien functions
1030 * fixed some bugs revealed by Paul Dietz' test suite:
1031 ** the type-error signalled from WARN has a filled-in DATUM slot.
1032 ** the type-error required when a stream is not associated with
1033 a file has the stream as its datum.
1034 ** type-errors on single-floats on x86-64 no longer have
1035 :INVALID-OBJECT as the datum
1036 ** the type-errors signalled for invalid function names now have
1037 a correct expected type
1038 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1039 for structures defined with a :TYPE. Documentation strings for
1040 typed structures are no longer immediately discarded
1041 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1042 broadcast streams and synonym streams. FILE-LENGTH now also works
1043 on broadcast streams.
1045 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1046 * incompatible change: the --noprogrammer option, deprecated since
1047 version 0.7.5, has been removed. Please use the equivalent
1048 --disable-debugger option instead.
1049 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1051 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1052 inoperational for a while now) has been completely removed.
1053 * null lexical environments are now printed as #<NULL-LEXENV>,
1054 significantly reducing the amount of clutter in typical backtraces.
1055 * documentation on weak pointers, finalization, and after GC hooks
1056 has been added to the manual.
1057 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1058 as other specialized array types, is much faster. SUBSEQ and
1059 COPY-SEQ on such arrays have also been sped up.
1060 * optimization: EQL is now more efficient when at least other argument
1061 is known to be of type (OR FIXNUM (NOT NUMBER)).
1062 * fixed bug: the runtime is now less vulnerable to changes in the
1063 size of the SBCL object on OS X, and virtual memory is reserved for
1064 all spaces that need to be at a fixed address.
1065 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1066 * fixed bug: finalizers and after GC hooks that cause consing are now
1068 * fixed bug: compiler error messages and summaries are now printed to
1069 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1070 * fixed inference of the upper bound of an iteration variable.
1071 (reported by Rajat Datta).
1072 * fixed bug 211e: calling local functions with duplicated constant
1073 keyword argument no longer causes a bogus style warning about an
1075 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1076 ftype declaration to be disregarded. (reported by Dave Roberts)
1077 * fixed bug 373: caused by erronous compilation of references to alien
1078 variables in the runtime on ppc/darwin.
1079 * fixed bug 376: CONJUGATE type deriver.
1080 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1081 (reported by Baughn on #lisp)
1082 * a cross-compiler bug on non-x86 platforms has been identified and
1083 fixed. (thanks to Bruno Haible)
1084 * improvements to the MIPS runtime code for increased stability.
1085 (thanks to Thiemo Seufer)
1086 * increased the maximimum compact environment size to allow
1087 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1088 * partial workaround for bug 135: don't name the function we're
1089 calling for hairy cases of EVAL, so as not to accumulate one environment
1090 entry per call to EVAL. (reported by Kevin Reid)
1091 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1092 * optimization: type testing for non-vector arrays should be faster.
1093 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1095 * bug fix: setting 31st element of a bit vector to zero did not work
1097 * bug fix: redefining a class definition which failed due to a
1098 previous accessor / function clash now works (but see BUGS entry
1099 #380 for more problems in this area). (thanks to Zach Beane)
1100 * the long form of DEFINE-METHOD-COMBINATION disables method group
1101 checking when given a single method group with pattern *. (thanks
1103 * on x86 compiler supports stack allocation of results of simple
1104 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1105 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1106 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1107 * fixed some bugs related to Unicode integration:
1108 ** the restarts for recovering from input and output encoding
1109 errors only appear when there is in fact such an error to
1111 * fixed some bugs revealed by Paul Dietz' test suite:
1112 ** MISC.549 and similar: late transformation of unsafe type
1113 assertions into derived types caused unexpected code
1115 ** SCALE-FLOAT type deriver is less wrong.
1116 ** type derivers for EXP, LOG and similar functions compute result
1117 types for complex arguments better.
1118 ** (MISC.563) CONJUGATE type deriver works for very restricted
1120 ** out-of-line type testers for character strings are available.
1121 ** EQUAL compiler transform understands specialness of objects
1123 ** accessing double-floats stored on the stack now works on x86-64.
1124 ** debugger internals could sometimes create invalid lispobjs,
1125 resulting in GC crashes.
1126 ** MISC.548: type check weakening can convert required type into
1128 ** initialization forms for bindings are not in scope of free special
1131 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1132 * incompatible change: thread support for non-NPTL systems has
1133 been removed - locking is buggy and unreliable. A threaded
1134 SBCL build will now warn at startup and refuse to create
1135 new threads, unless futex support is detected in the kernel
1136 * incompatible change: the top level REPL now has only an ABORT
1137 restart associated with it, not TOPLEVEL and ABORT as it used to.
1138 TOP and TOPLEVEL are now available as debugger commands for
1139 returning to the top level.
1140 * incompatible change: forms evaluated in the REPL now use the
1141 global optimization policy.
1142 * incompatible change: user- and system-initialization files are
1143 no longer processed with LOAD, but by READ and EVAL; hence the
1144 global optimization policy, startup package, readtable, etc,
1146 * The .fasl file format number has been incremented because of
1147 various incompatible changes.
1148 * internal entry point details and argument counts no longer appear
1149 in backtraces unless explicitly requested by setting
1150 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1151 * built-in and standard functions no longer have names like "top
1152 level local call to FOO".
1153 * fixed bug 32: functions defined in non-null lexical environments
1154 now have more legible printed representation
1155 * fixed bug 33: functions defined in non-null lexical environemnts
1156 are now more amenable to inspection by INSPECT.
1157 * workaround for bug 354: XEPs no longer appear in backtraces unless
1158 explicitly requested.
1159 * fixed bug: receiving the signal which results from attempting to
1160 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1161 SIGBUS on other *BSDs) on architectures where the C stack is also
1162 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1163 to debugger confusion or wild instability. (reported by Cheuksan
1165 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1166 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1167 (reported by Lutz Euler)
1168 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1169 compiled code. (reported by Paul Dietz)
1170 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1171 specialized parameter to a method to something that is not TYPEP
1172 the specializer is now possible.
1173 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1174 face of package deletion.
1175 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1176 pathname. (reported by David Tolpin on c.l.l)
1177 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1178 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1179 STRING-STREAM and FILE-STREAM, respectively.
1180 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1181 than previously; in addition, its test suite is now run on build.
1182 (thanks to Luke Gorrie)
1183 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1184 * optimization: added a immediate representation for single-floats
1186 * fixed some bugs revealed by Paul Dietz' test suite:
1187 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1188 ** attempting to create a package with a colliding nickname causes
1189 correctable errors to be signalled.
1190 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1191 ** several x86-64 backend bugs related to sign-extension of immediate
1194 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1195 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1196 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1198 * fixed loading of multiply forward-referenced layouts.
1199 (thanks to Cheuksan Wang)
1200 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1201 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1202 (see COMPILE-FILE documentation for details of :PRINT option)
1203 * fixed bugs 19 and 317: fixed-format floating point printing is
1204 more accurate. This also fixes a bug reported by Adam Warner
1205 related to the ~@F format directive.
1206 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1208 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1209 dispatch table if the from-char is a dispatch macro character.
1210 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1211 by the wordsize no longer ignore the last word. (reported by Lutz
1213 * fixed bug in type checking of dynamic-extent variables. (reported
1215 * optimization: sequence traversal functions use their freedom to
1216 coerce function designators to functions.
1217 * optimization: code with many calls to CLOS methods specialized on
1218 CLOS classes has had redundant type checks removed.
1219 * optimization: added declarations to speed up operations that access
1220 the internal character database (for example STRING-UPCASE)
1221 * optimization: comparison operations between floats and sufficiently small
1222 fixnums no longer create extra rationals
1223 * fixed some bugs related to Unicode integration:
1224 ** portions of multibyte characters at the end of buffers for
1225 character-based file input are correctly transferred to the
1226 start of the buffer at the next read.
1227 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1228 passing it through to OPEN.
1229 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1230 argument given to internal calls to OPEN.
1231 ** the built-in comment readers (introduced by character sequences
1232 ";" and "#|") are more forgiving to encoding errors; they will
1233 STYLE-WARN and then attempt to resync the stream at a character
1234 boundary. (thanks to Teemu Kalvas)
1235 * fixed some bugs in the x86-64 port:
1236 ** Negative short int return values from c-calls are sign-extended
1238 ** The stack is aligned to 16-bytes for c-calls, as required by
1239 the ABI. (thanks to Cheuksan Wang)
1240 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1241 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1242 for enums. (thanks to Vincent Arkesteijn)
1243 ** Multiple small optimizations and bugfixes for floating point
1245 * fixed some bugs revealed by Paul Dietz' test suite:
1246 ** Space, Tab, Linefeed, Return and Page have the invalid
1247 secondary constituent character trait.
1248 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1250 ** WITH-INPUT-FROM-STRING should only update the index place on
1252 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1253 works more reliably.
1254 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1255 with one-dimensional arrays with fill pointers.
1256 ** TYPE-OF failed on a complex with an integer realpart and a
1258 ** compiler failure during type inference for the code of form
1259 (IF (EQL X (THE ...)) ...) (MISC.535).
1261 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1262 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1263 platform. The port supports 61-bit fixnums, large memory spaces
1264 and reloading shared object files.
1265 * enhancement: saving cores with foreign code loaded is now
1266 supported on ppc/Darwin in addition to the previously supported
1268 * enhancement: the statistical profiler now walks deeper into the
1269 call stack for more meaningful call-graphs and accrued time
1270 reports (x86/x86-64 only). It also now reports time spent in
1272 * enhancement: it is now possible to trace most individual methods
1273 of a generic function in addition to tracing the generic function
1275 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1277 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1278 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1279 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1280 produces an error. (thanks to Vincent Arkesteijn)
1281 * bug fix: NAMESTRING on pathnames with :WILD components in their
1282 directories works correctly. (thanks to Artem V. Andreev)
1283 * fixed bug 125: compiler preserves identity of closures. (reported
1285 * bug fixed: functions with &REST arguments sometimes failed with
1286 "Undefined function" when compiled with (DEBUG 3). (reported by
1288 * bug fix: overflow during compiling of setting element of a bit
1289 vector with constant index and value. (reported by Timmy Douglas)
1290 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1291 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1292 * on x86 compiler partially supports stack allocation of dynamic-extent
1294 * GO and RETURN-FROM do not check the extent of their exit points
1295 when compiled with SAFETY 0.
1296 * fixed some bugs related to Unicode integration:
1297 ** encoding and decoding errors are now much more robustly
1298 handled; it should now be possible to recover even from invalid
1299 input or output to the terminal. (thanks to Teemu Kalvas)
1300 ** provided a first cut at implementing STRING-TO-OCTETS and
1301 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1302 ** altered the SB-MD5 contributed module slightly, changing the
1303 interface just enough for it to be supportable for builds where
1304 lisp characters are not eight bits.
1305 * fixed some bugs revealed by Paul Dietz' test suite:
1306 ** the FORMATTER-generated functions for ~V[ conditionals require
1307 the correct number of arguments.
1308 ** READ-FROM-STRING returns the mandated second value when applied
1309 to displaced strings.
1310 ** the #\Rubout and #\Backspace characters are treated as invalid
1311 constituent characters by the tokenizer.
1313 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1314 * new feature: reloading changed shared object files with
1315 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1316 * new feature: references to foreign variables and functions
1317 can now be compiled and loaded before the corresponding shared
1318 object file is loaded, as long as the foreign definitions are
1319 available at runtime.
1320 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1321 supported. (thanks to Dan Debertin)
1322 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1323 just keywords. (thanks to Vincent Arkesteijn)
1324 * fixed bug #331: structure-class instances corresponding to
1325 DEFSTRUCT forms are now created eagerly.
1326 * fixed bug #345: backtraces from calls to undefined functions work
1327 on x86 as well. Related bug #61 is now also partially fixed on x86
1328 (backtraces from throws to unknown catch tags.)
1329 * bug fix: lambda-list parsing is now stricter vrt. order and number
1330 of lambda-list keywords.
1331 * bug fix: as specified by AMOP, an error is signalled if a
1332 class-option appears multiple times in a DEFCLASS form. (reported
1334 * bug fix: RANDOM can be compiled when the compiler derives the type
1335 of its numeric argument as a disjoint set of small integers.
1336 (reported by Paul Dietz)
1337 * bug fix: starting a core saved with shared objects loaded when
1338 those objects are not available no longer causes threaded SBCL to
1339 hang. (reported by Sean Ross)
1340 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1342 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1343 argument, which must always be NIL. (reported by Kalle Niemitalo)
1344 * bug fix: printing 1.0d+23 no longer results in an error.
1345 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1346 * bug fix: structure slot setters preserve evaluation order. (thanks
1348 * fixed some bugs related to Unicode integration:
1349 ** RUN-PROGRAM can allow its child to take input from a Lisp
1350 stream. (reported by Stefan Scholl)
1351 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1352 locales. (reported by Ken Causey)
1353 * fixed some bugs revealed by Paul Dietz' test suite:
1354 ** INCF, DECF and REMF evaluate their place form as specified in
1356 ** FORMATTER expands ~{ ~} iteration directives with V or #
1357 parameters correctly.
1358 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1359 consequent uses no arguments correctly.
1360 ** the system has a partial understanding of the (COMPLEX RATIO)
1362 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1365 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1366 * new feature: a build-time option (controlled by the :SB-UNICODE
1367 keyword feature, enabled by default) for building the system with
1368 support for the entire 21-bit character space defined by the
1370 * new feature: the system now has rudimentary external-format
1371 support; the primary user-visible change at this time is that
1372 characters with the high bit set (such as lower-case-e-acute) will
1373 print correctly to a terminal in a UTF-8 environment.
1374 * minor incompatible change: BASE-CHAR no longer names a class;
1375 however, CHARACTER continues to do so, as required by ANSI.
1376 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1377 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1378 has been moved to the SB-EXT package (temporarily re-exported from
1380 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1381 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1383 * fixed bug #351: better error-handlind and reporting for malformed
1385 * fixed bug #350: bignum-printing is now more memory-efficient,
1386 allowing printing of very large bignums, eg. (expt 2 10000000).
1387 (reported by Bruno Haible)
1388 * fixed bug #302: better primitive-type selection for intersection
1390 * fixed bug #308: non-graphic characters now all have names, as
1391 required. (reported by Bruno Haible)
1392 * bug fix: redefining a class with different superclasses now correctly
1393 removes it from the direct-subclasses of its previous superclasses.
1394 (reported by David Morse)
1395 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1396 new value now works. (reported by Bruno Haible)
1397 * bug fix: correct canonicalization of multiple non-standard slot
1398 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1399 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1400 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1401 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1402 now exists, an signals an error.
1403 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1404 during macroexpansion for non-iteration variables as well. (reported
1405 by Bruno Haible for CMUCL)
1406 * bug fix: Cyclic structures and unprintable objects in compiler
1407 messages no longer cause errors. (reported by Bruno Haible)
1408 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1409 and READ-FROM-STRING all now return a primary value of NIL if
1410 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1411 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1412 (reported by Bruno Haible for CMUCL)
1413 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1414 do not designate a condition. (reported by Bruno Haible for
1416 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1417 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1418 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1419 permitting resolution in favour of any of the conflicting symbols.
1420 (reported by Bruno Haible for CMUCL)
1421 * bug fix: EQUAL compiler optimizations is less aggressive on
1422 strings which can potentially compare true despite having distinct
1423 specialized array element types.
1424 * bug fix: unit enumerations can be defined without dividing by
1425 zero. (thanks to Vincent Arkesteijn)
1426 * FORMAT compile-time argument count checking has been enhanced.
1427 (report from Bruno Haible for CMUCL)
1428 * a partial workaround for the bug 262: the compiler does not try to
1429 inline-expand a local function doing RETURN-FROM from a deleted
1430 BLOCK. (thanks to Peter Denno for the bug report and to David
1431 Wragg for the simple test case)
1432 * fixed some bugs revealed by Paul Dietz' test suite:
1433 ** RENAME-PACKAGE allows all package designators as new package
1435 ** constraint propagation and lambda variable substitution are
1436 more cautious in dealing with partially deleted code.
1437 ** compiler sometimes lost reoptimization passes.
1438 ** CERROR, given a condition as condition designator, passes its
1439 remaining arguments to the continue format control without
1441 ** Case-altering FORMAT directives work correctly on non-ASCII
1443 ** The REAL type specifier handles bounds outside the floating
1444 point ranges without signalling FLOATING-POINT-OVERFLOW.
1445 ** Functions with IR1-transformations can create intercomponent
1446 references to global functions.
1447 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1449 ** FORMAT ~R treats a nil value for its first parameter correctly.
1451 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1452 * enhancement: saving cores with foreign code loaded is now
1453 supported on x86/NetBSD and sparc/Linux in addition to the previously
1454 supported platforms.
1455 * bug fix: on some platforms repeated installations caused multiple
1456 copies of HTML documentation to be installed -- should not happen
1457 any more. (reported by Stefan Scholl)
1458 * bug fix: parsing self-recursive alien record types multiple times
1459 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1460 original patch by Helmut Eller for CMUCL)
1461 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1462 (thanks to Richard Kreuter)
1463 * bug fix: defining classes whose accessors are methods on existing
1464 generic functions in other (locked) packages no longer signals
1465 bogus package lock violations. (reported by François-René Rideau)
1466 * bug fix: special variables as DEFMETHOD parameters no longer have
1467 associated bogus type declarations. (reported by David Wragg and
1469 * bug fix: read-write consistency on streams of element-type
1470 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1471 * bug fix: redefiniton of the only method of a generic function with
1472 no DEFGENERIC no longer emits a full WARNING. In addition,
1473 redefinition of generic functions with no DEFGENERIC to an
1474 incompatible lambda list now signals an error. (thanks to Zach
1476 * bug fix: DEFGENERIC now works even when there's a function of the
1477 same name in an enclosing lexical environment. (thanks to Zach
1479 * fixed compiler failure, caused by instrumenting code during
1480 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1481 * optimization: added loop analysis and improved register allocation
1482 * fixed some bugs revealed by Paul Dietz' test suite:
1483 ** POSITION on displaced vectors with non-zero displacement
1484 returns the right answer.
1485 ** (SIMPLE-STRING) is a valid type specifier for sequence
1487 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1489 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1490 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1492 ** The justification version of the FORMAT ~< directive treats
1493 non-zero minpad parameter correctly.
1495 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1496 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1497 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1498 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1499 the supported interface.
1500 * new feature: Single-stepping of code compiled with DEBUG 2 or
1501 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1502 * new feature: saving cores with foreign code loaded is now
1503 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1504 Timothy Moore's work for CMUCL)
1505 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1506 parameters to * instead of NIL if no initform is supplied.
1507 (reported by Johan Bockgaard)
1508 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1509 correctly when FUNCALL appears as the car of the form. Note:
1510 despite this FUNCALL forms are not currently subject to
1511 compiler-macro expansion. (port of Raymond Toy's fix for the
1512 same from CMUCL, reported by Johan Bockgaard)
1513 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1514 (thanks for Teemu Kalvas)
1515 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1516 case instead of terminating the iteration (thanks for Julian
1517 Squires, Sean Champ and Raymond Toy)
1518 * bug fix: incorrect expansion of defgeneric that caused a style
1519 warning. (thanks for Zach Beane)
1520 * bug fix: slot accessor effective method computation works properly
1521 for classes with multiple non-standard applicable methods on
1522 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1523 * on x86 compiler supports stack allocation of results of LIST and
1524 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1525 CMUCL implementation by Gerd Moellmann)
1526 * fixed some bugs revealed by Paul Dietz' test suite:
1527 ** FORMAT strings with both the ~<~:;~> form of the justification
1528 directive and pretty-printing directives cause an error.
1530 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1531 * incompatible change: the internal functions
1532 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1533 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1534 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1535 instead of the old functions.
1536 * new feature: on platforms where "dladdr" is available foreign
1537 function names now appear in backtraces. (based on Helmut Eller's
1539 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1540 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1542 * documentation: ASDF manual is now distributed with SBCL.
1543 * bug fix: SBCL can now load its contributed modules with REQUIRE
1544 even if the system-provided entries have been removed from
1545 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1547 * bug fix: backtraces involving undefined functions or assembly
1548 routines are more informative. (thanks to Brian Downing)
1549 * bug fix: mutually referent alien structures now work correctly.
1550 (reported by Rick Taube)
1551 * bug fix: structures defined by WITH-ALIEN can be referred to
1552 within other definitions in the same WITH-ALIEN.
1553 * bug fix: division operators (MOD, TRUNCATE and the like) with
1554 constant zero divisors and integer dividends no longer generate
1556 * bug fix: provide default methods for INPUT-STREAM-P and
1557 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1558 * bug fix: improve the detection and resolution of MOP metacycles
1559 (where effective method computation depends on the generic function
1560 having its effective method computed). (reported by Bruno Haible)
1561 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1562 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1563 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1564 * bug fix: pathnames with relative directory components can be
1565 represented relative to default pathnames.
1566 * optimization: in taking the GCD of bignums, reduce the two bignums
1567 to approximately the same size (using Euclid's algorithm) before
1568 applying the more sophisticated binary GCD. (thanks to Juho
1570 * optimization: COUNT on bitvectors now operates word-at-a-time.
1571 * optimization: ASH with a positive, but not necessarily constant,
1572 (leftwards) shift, when in a modular context, is compiled to a
1574 * fixed some bugs revealed by Paul Dietz' test suite:
1575 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1576 if the corresponding argument is NIL.
1577 ** FORMAT directives accepting a minpad parameter treat negative
1579 ** literal commas as character parameters to FORMAT directives are
1581 ** literal spaces directly after ~<Newline> directives within a
1582 format-logical-block (~:< ~@:>) do not induce :FILL-style
1583 conditional newlines.
1584 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1585 as specified: it no longer includes conditional newlines).
1586 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1588 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1589 :PER-LINE-PREFIX argument does not evaluate to a string.
1591 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1592 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1593 the manual for details; for now, package locks can be disabled by
1594 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1595 if no major problems are found then it is likely that they will be
1596 compiled in unconditionally.
1597 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1598 unsupported operators on all platforms. To load a shared library
1599 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1600 object file, link it into a shared library outside of SBCL and
1601 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1603 * bug workaround: ROOM T and the GC were not working together
1604 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1605 reporting facility aren't preserved. That reporting has been
1606 disabled, so now until and unless someone figures out how to make it
1607 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1608 * minor incompatible change: as threatened around sbcl-0.8.0, the
1609 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1610 The symbols are also exported from SB-PCL for backwards
1611 compatibility, but more so than before SB-PCL should be treated as
1612 an implementation-internal package.
1613 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1615 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1616 It is known as of this release not to work on the Alpha, however.
1617 * fixed bug #167: errors signalled due to illegal syntax in method
1618 bodies are now more legible.
1619 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1620 designators and can hence be used with TYPEP.
1621 * fixed bug #333: CHECK-TYPE now ensures that the type error
1622 signalled, if any, has the right object to be accessed by
1623 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1624 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1625 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1627 * fixed bug #334: programmatic addition of slots using specialized
1628 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1629 and :ALLOCATION :CLASS slots.
1630 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1631 float exponents. (rereported by Peter Seibel)
1632 * fixed a bug: #\Space (and other whitespace characters) are no
1633 longer considered to be macro characters in standard syntax by
1634 GET-MACRO-CHARACTER.
1635 * fixed bug: initialization of condition class metaobjects no longer
1636 causes an instance of the condition to be created. (reported by
1638 * fixed bug: it is now possible to have slots such that
1639 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1640 description is neither :INSTANCE nor :CLASS.
1641 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1642 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1643 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1644 non-standard SLOT-DEFINITION-ALLOCATION values.
1645 * fixed bugs: various race conditions handling exiting threads.
1646 CL-PPCRE's thread torture test now passes.
1647 * fixed arguably-a-bug: GC time proportional to number of threads in
1648 system even when most of them are idle
1649 * optimization: improved performance of BIT and SBIT on bit-vectors.
1650 * .fasl file incompatibility: The fasl file version number has
1651 been incremented because of changes associated with package locks.
1653 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1654 * minor incompatible change: the system no longer provides
1655 optimization hints (conditions which are TYPEP
1656 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1657 mode; these hints are still emitted when higher SPEED optimization
1659 * new contrib module: a sampling profiler (profiling by statistical
1660 sampling, rather than by instrumenting functions) is available as
1661 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1662 * the behaviour of the standard function ED is now customizeable by
1663 third parties through a hook variable: see ED's documentation
1664 string for information on the protocol.
1665 * the compiler no longer emits efficiency notes for (FUNCALL X)
1666 when the type of X is uncertain under default optimization
1668 * fixed bug 276: mutating a binding of a specialized parameter to a
1669 method to something that is not TYPEP the specializer is now
1671 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1672 exhibit the required behaviour on the x86 platform. (thanks to
1673 Peter van Eynde, Eric Marsden and Bruno Haible)
1674 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1675 even for difficult arguments. (reported by Peter Graves)
1676 * fixed bug 141a: the backquote printer now descends quoted
1678 * fixed another bug in backquote printing: no more destructive
1679 modification of the form's list structure. (reported by Brian
1681 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1682 move between its address being taken and the call to
1683 interrupt_thread, fixing a crashing race condition.
1684 * the SB-POSIX contrib implementation has been adjusted so that it
1685 no longer exhibits ridiculously poor performance when constructing
1686 instances corresponding to C structs.
1688 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1689 * minor incompatible change: the sb-grovel contrib now treats C
1690 structures as alien (in the sense of SB-ALIEN) objects rather than
1691 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1692 has implications for memory management of client code
1693 (sb-grovel-returned objects must now be manually managed) and for
1694 type safety (alien objects now have full types).
1695 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1696 used to control emission of compiler diagnostics, rather than the
1697 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1698 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1699 quality should be considered deprecated.
1700 * install.sh now installs the user manual as well
1701 * (not quite a new documentable feature, but worth considering in
1702 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1703 beginnings of a semantically meaningful condition hierarchy is
1704 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1706 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1707 designator as the defaults argument.
1708 * fixed bug: Displaced arrays whose displaced-to array has become
1709 too small now cause ARRAY-DIMENSION to signal an error, providing
1710 for safer bounds-checking. (reported by Bruno Haible)
1711 * fixed bug: DEFCLASS slot definitions with identical :READER and
1712 :WRITER names now signal a reasonable error. (reported by Thomas
1714 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1716 * fixed bug: Invalid slot specification errors now print correctly.
1717 (thanks to Zach Beane)
1718 * fixed bug 320: Shared to local slot value transfers in class
1719 redefinitions now happen corrently from superclasses as
1720 well. (reported by Bruno Haible)
1721 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1723 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1724 improper lists correctly. (reported by Bruno Haible)
1725 * fixed bug 313: source-transform for <fun-name> was erroneously
1726 applied to a call of a value of a variable with name <fun-name>.
1727 (reported by Antonio Menezes Leitao)
1728 * fixed bug 307: The obsolete instance protocol ensures that
1729 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1730 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1731 remove dead unknown-values globs from the middle of the stack.
1732 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1733 easier to resume long computations after using *BREAK-ON-SIGNALS*
1734 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1735 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1736 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1737 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1738 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1739 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1741 * fixed a (fairly theoretical) bug in string printing: if
1742 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1743 does not have array-element-type equal to the most general string
1745 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1746 function name. (thanks to Nikodemus Siivola)
1747 * fixed bug: compile-time format string checker failed on
1748 non-closed ~{. (reported by Thomas F Burdick)
1749 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1750 don't warn on legal loop code involving more than one aggregate
1751 boolean. (thanks to Nikodemus Siivola)
1752 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1753 in keyword position to MAKE-PACKAGE was not being reported as
1755 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1756 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1758 * fixed bugs 280 and 312: the checking for multiple definitions in a
1759 file is less likely to become confused by uses of inline
1761 * fixed bug: the #S reader macro performs the keyword coercion
1762 specified for slot names. (reported by Kalle Niemitalo)
1763 * fixed bug: lambda lists may contain symbols whose names start with
1764 & but are not lambda-list-keywords; their occurrence triggers a
1766 * fixed bug 321: define-method-combination argument lambda lists do
1767 not cause invalid code to be generated when &optional or &aux
1768 variables have default values. (reported by Bruno Haible)
1769 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1770 have CLOS classes; this appears also to have fixed reported
1771 instability in stack exhaustion detection.
1772 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1773 MUFFLE-WARNING when no associated restart is present is now
1775 * optimization: rearranged the expansion of various defining macros
1776 so that each expands into only one top-level form in a
1777 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1779 * optimization: used a previously little-used slot in symbols to
1780 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1782 * fixed some bugs revealed by Paul Dietz' test suite:
1783 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1785 ** Fixed an optimization bug related to inheritance of initforms
1786 from local to shared slots.
1787 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1788 ** CAST optimizer forgot to flush argument derived type.
1789 ** print/read consistency on floats is now orders of magnitude
1790 more likely. (thanks also to Bruno Haible for a similar report
1792 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1793 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1794 for objects of type REAL. Make it so.
1795 ** SXHASH is specified (infelicitously) to respect similarity,
1796 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1797 it so. (thanks to Markus Ziegler)
1798 ** on the Alpha, the compiler succeeds in compiling functions
1799 returning a known number of arguments greater than 63.
1800 ** fixed handling of invalid NIL arguments in keyword position
1802 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1804 ** provided a readably-printable representation for RANDOM-STATE
1806 ** ensured that pathnames, where they have a namestring, always
1807 print using #P"..." syntax.
1809 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1810 * Support for the forthcoming 2.0 version of the NetBSD kernel
1811 running on x86 hardware has been added. (thanks to Perry
1812 E. Metzger most immediately, and others for their past work)
1813 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1814 current release version; SBCL's OpenBSD support had been broken
1815 since about the time of OpenBSD's switch to ELF binary format.)
1816 * [placeholder for DX summary]
1817 ** user code with &REST lists declared dynamic-extent, under high
1818 speed or space and low safety and debug optimization policy.
1819 * The manual has been converted to Texinfo format and the debugger
1820 chapter from the cmucl manual has been added.
1821 * A facility has been added to extract documentation strings from
1822 sbcl and store them as Texinfo-formatted snippets for inclusion in
1823 the manual (via Texinfo's @include directive)
1824 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1825 the test case to Dave Roberts)
1826 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1827 pointers. (reported by Sean Ross)
1828 * bug fix: PROFILE output is printed nicely even for large numerical
1829 values. (thanks to Zach Beane)
1830 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1831 greater than 32 handle EOF correctly.
1832 * bug fix: on X86 an immediate argument of the IMUL instruction is
1833 correctly printed in disassembly. (thanks to Lutz Euler)
1834 * bug fix: class slots in redefined classes preserve their old
1835 values. (thanks to Bruno Haible and Nikodemus Siivola)
1836 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1837 succeeds. (reported by Marco Baringer)
1838 * fixed some bugs revealed by Paul Dietz' test suite:
1839 ** READ-SEQUENCE now works on ECHO-STREAMs.
1840 ** RATIONALIZE works more according to its specification. (thanks
1843 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1844 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1845 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1846 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1847 should matter to you only if you rebind the printer control
1848 variables and then find you want different bindings in the
1849 debugger than in the ordinary execution of your program.)
1850 * The runtime build system has been tweaked to support building
1851 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1852 assembler and linker. (thanks to Nikodemus Siivola)
1853 * Unbound, undefined, undeclared variables now trigger full
1854 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1855 more often programmer error than deliberate exploitation of undefined
1857 * optimization: the hash algorithm for strings has changed to one
1858 that is less vulnerable to spurious collisions. (thanks to Juho
1860 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1861 needless bounds checking. (thanks to Juho Snellman)
1862 * optimization: implemented multiplication as a modular
1863 (UNSIGNED-BYTE 32) operation on the PPC backend.
1864 * fixed some bugs revealed by Paul Dietz' test suite:
1865 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1867 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1868 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1869 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1871 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1873 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1875 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1876 their output stream on EOF from read.
1877 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1878 have been read to end-of-file.
1879 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1881 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1882 description of determination of which consecutive characters
1884 ** Printing the "Space" character with escaping on now yields "#\\ ",
1885 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1886 ** Reading floating-point numbers with *READ-BASE* set to a number
1887 less than 10 works correctly.
1888 ** Reading floating-point numbers with *READ-BASE* set to a number
1889 more than 10 works correctly.
1890 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1891 the readtable currently in effect.
1893 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1894 * minor incompatible change: parsing of namestrings on a physical
1895 (Unix) host has changed; numbers after the final #\. in a
1896 namestring are no longer interpreted as a version field. This is
1897 intented to be largely invisible to the user, except that the
1898 meaning of the namestring "*.*.*" has changed: it now refers to a
1899 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1900 should usually be replaced by
1901 (make-pathname :name :wild :type :wild :version :wild)
1902 with the added benefit that this is more likely to be portable.
1903 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1904 option to OPEN now signals an error if the file being opened
1905 exists; this may have an impact on existing code.
1906 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1907 signals received, on the PowerPC platforms (both Linux and
1908 Darwin). (thanks to Pierre Mai for pointing out the location of
1910 * several fixes on OS X: The system now builds and runs cleanly on
1911 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1912 (thanks to Brian Mastenbrook)
1913 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1914 second-resolution: integer multiples of 1/3600 between -24 and 24.
1915 (thanks to Vincent Arkesteijn)
1916 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1917 type when called with 1 argument; PEEK-CHAR checked type of
1918 PEEK-TYPE only after having read first character from a
1919 stream. (reported by Peter Graves)
1920 * bug fix: the garbage collector now has much better locality
1921 behaviour, and in particular no longer treats all memory as being
1922 exhausted when many small objects point to each other in a deeply
1924 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1925 recognized as being TYPEP their class.
1926 * bug fix: the PUSHNEW documentation string has been corrected.
1927 (thanks to Vincent Arkesteijn)
1928 * bug fix: defaulting of the value for the last of an atypically
1929 large number of multiple values being bound was not being
1930 performed correctly on the Alpha or PPC platforms
1931 * optimization: implemented multiplication as a modular
1932 (UNSIGNED-BYTE 32) operation on the x86 backend.
1933 * optimization: SEARCH on simple-base-strings can now be open-coded.
1934 (see also contrib/compiler-extras.lisp for inspiration for
1935 teaching the compiler about the Boyer-Moore algorithm).
1936 * value, returned by MAX (and MIN) called with several EQUALP, but
1937 not EQL, arguments now does not depend on compiler settings.
1938 * fixed some bugs revealed by Paul Dietz' test suite:
1939 ** in stack analysis liveness information is propagated from
1940 non-local entry points.
1941 ** pathwise CAST removing failed when the CAST node did not start
1943 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1944 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1946 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1947 host is already defined.
1948 ** RENAME-FILE works on streams instead of signalling an internal
1950 ** PEEK-CHAR uses the current readtable when determining whether
1951 or not a character is whitespace.
1952 ** MERGE-PATHNAMES handles the case when the pathname does not
1953 specify a name while the default-pathname specifies a version.
1954 ** Pathnames now stand a better chance of respecting print/read
1956 ** Attempting to use standardized file system operators with a
1957 pathname with invalid :DIRECTORY components signals a
1959 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1961 ** DIRECTORY on logical pathnames is more correct.
1962 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1963 signal a TYPE-ERROR if their argument is not a stream.
1964 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1965 designator argument does not designate a stream.
1966 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1967 examining the synonym.
1968 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1970 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1971 element-type larger than ([UN]SIGNED-BYTE 32).
1973 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1974 * When built with the :SB-FUTEX feature, threaded builds now take
1975 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1976 for faster/more reliable mutex and condition variable support.
1977 * Incompatible change (but one you probably shouldn't have been using
1978 anyway): the interface and code for arbitrating between multiple
1979 threads in the same user session has been redesigned.
1980 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1981 error when called without an explicit environment argument.
1982 (thanks to Nikodemus Siivola)
1983 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1984 confused about their position. (thanks to Adam Warner and Gerd
1986 * bug fix: Pretty printing backquoted forms with unquotations in the
1987 argument list position of various code constructs such as LAMBDA
1988 now works correctly. (reported by Paul Dietz)
1989 * bug fix: Pretty printing unquotations no longer loses all
1990 stream position information.
1991 * optimization: performance of string output streams is now less
1992 poor for multiple small sequence writes.
1993 * optimization: performance of CSUBTYPEP in the presence of complex
1994 expressions involving CONS and NOT many times has been improved.
1995 (reported by Paul Dietz)
1996 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1998 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1999 streams for the REPL, for the debugger, and so on. (thanks to
2001 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2002 * fixed some bugs revealed by Paul Dietz' test suite:
2003 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2004 is ignored if it is too small, rather than propagating through
2005 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2006 ** extremely complex negations of CONS types were not being
2007 sufficiently canonicalized, leading to inconsistencies in
2009 ** VALUES tranformer lost derived type.
2011 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2012 * fixed a bootstrapping bug: the build process no longer assumes
2013 that the various BOOLE-related constants have the same value in
2014 host and target lisps. (noted by Paul Dietz' test suite on an
2015 SBCL binary built from CLISP)
2016 * The system can now be dynamically linked on the MIPS platform,
2017 which enables dynamic loading of foreign code from Lisp. (thanks
2018 to Ralf Baechle for discussions on the MIPS ABI)
2019 * The system now records debugging information for its own source
2020 files in a filesystem-position-independent manner, relative to
2021 the "SYS" logical host.
2022 * fixed a compiler bug: MV-LET convertion did not check references
2023 to the "max args" entry point. (reported by Brian Downing)
2024 * tweaked disassembly notes to be less confident about proclaiming
2025 some instruction as an LRA. (thanks to Brian Downing)
2026 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2027 now each have their own history, command character, and other
2028 characteristics. (thanks to David Lichteblau)
2029 * fixed some bugs revealed by Paul Dietz' test suite:
2030 ** compiler failure in compiling LOGAND expressions including a
2032 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2033 shift greater than 32.
2034 ** FLUSH-DEST did not mark blocks for type check regeneration.
2035 ** HANDLER-CASE failed to accept declarations in handler clauses
2036 in some circumstances.
2038 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2039 * New code in contrib/sb-introspect (still probably not entirely
2040 stable yet) provides some support for smart Lisp development
2041 environments like SLIME.
2042 * The conditions signalled for errors occurring when loading .fasl
2043 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2044 in a way which should help ASDF recover gracefully.
2045 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2046 stuff has been cleaned up. If you code contrib/ stuff, this might
2047 affect you, and you can look at contrib/README, contrib/STANDARDS,
2048 and/or the 0.8.4.27 diff to check.
2049 * In full calls the compiler now does not generate checks for declared
2050 argument types for all arguments.
2051 * various threading fixes
2052 ** and some experimental patches which didn't make it into
2053 the main tree for this release, but which are shipped in
2054 contrib/experimental-thread.patch as a possible fix for some
2055 failures (deadlock, spinning...) in GC-intensive multithreaded
2057 * fixed PPC build problem (source code incompatibility of different
2058 library versions): added offsetof() hackery which attempts to divine
2059 where glibc maintainers put uc_mcontext today
2060 * fixed bug 282: compiler does not trust type assertions while passing
2061 arguments to a full call.
2062 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2063 &OPTIONAL VALUES type parameter.
2064 * fix bug 214: algorithm for noting rejected templates is now more
2065 similar to that of template seletion. (also reported by rydis on
2067 * fixed bug 141b: printing backquoted information readably and prettily
2068 inserts a space where necessary.
2069 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2070 (CHARACTER 10) are now reported as errors, rather than propagated
2071 as unknown types. (reported by piso on #lisp)
2072 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2073 with values NIL and :ERROR. (thanks to Milan Zamazal)
2074 * fixed bug 191c: CLOS now does proper keyword argument checking as
2075 described in CLHS 7.6.5 and 7.6.5.1.
2076 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2077 longer raise an error; further, using a list as a for-as-arithmetic
2078 counter now raises a meaningful error.
2079 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2080 now understood by sequence creation functions such as MAKE-SEQUENCE
2082 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2083 a STRING-INPUT-STREAM.
2084 * compiler enhancement: SIGNUM is now better able to derive the type
2086 * type declarations inside WITH-SLOTS are checked. (reported by
2088 * fixed some bugs revealed by Paul Dietz' test suite:
2089 ** incorrect optimization of TRUNCATE for a positive first
2090 argument and negative second.
2091 ** compiler failure in let-convertion during flushing dead code.
2092 ** compiler failure while deriving type of TRUNCATE on an
2093 interval, containing 0.
2094 ** ASH of a negative bignum by a negative bignum count now returns
2096 ** intersection of CONS types now canonicalizes properly, fixing
2097 inconsistencies in SUBTYPEP.
2099 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2100 * incompatible change: The --disable-debugger command line
2101 option now clobbers the debugger at a more fundamental
2102 level, by redefining #'INVOKE-DEBUGGER instead of by
2103 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2104 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2105 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2106 enter the debugger REPL and then suffer recursive errors
2107 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2108 while under the new system, BREAK in --disable-debugger mode
2109 terminates the system just as an unhandled error would.
2110 * fixed compiler performance when processing loops with a step >1;
2111 * bug fix: DOCUMENTATION now retrieves generic function
2112 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2113 support has been systematized, and now supports the methods
2114 specified by ANSI, along with a default method and a method for
2115 slot documentation. (reported by Nathan Froyd)
2116 * bug fix: effective methods associated with a generic function are
2117 no longer cached over a change of that generic function's method
2118 combination. (reported by Andreas Fuchs)
2119 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2120 names a non-existent directory. (thanks to Andreas Fuchs)
2121 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2122 the PPC platform, lead to stack corruption; this has been fixed.
2123 (reported by Rainer Joswig)
2124 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2125 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2126 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2127 arithmetic in general, is now much more reliable on the PPC
2129 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2130 platform now returns the right answer.
2131 * optimization: restored some effective method precomputation in
2132 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2133 precomputation is now tunable.
2134 * optimization: compiler-internal data structure use has been
2135 reviewed, and changes have been made that should improve the
2136 performance of the compiler by about 20%.
2137 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2138 simple-base-strings and simple-bit-vectors is improved.
2139 * optimization: the optimization of 32-bit logical and arithmetic
2140 functions introduced in version 0.8.3 on the x86 has been
2141 implemented on the mips, ppc and sparc platforms; an
2142 implementation of the same facility, but for 64-bit arithmetic,
2143 has been added for the alpha.
2144 * microoptimization: the compiler is better able to make use of the
2145 x86 LEA instruction for multiplication by constants.
2146 * bug fix: in some situations compiler did not report usage of
2147 generic arithmetic in (SPEED 3) policy.
2148 * bug 145b fix: compiler used wrong type specifier while converting
2149 MEMBER-types to numeric.
2150 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2152 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2153 index argument now works properly on the Alpha platform.
2154 * bug fix: floating point exception treatment on the Alpha platform
2156 * bug fix: FILE-POSITION works much better on string input and
2157 output streams. (thanks to Nikodemus Siivola)
2158 * bug fix: many threading/garbage collection symptoms sorted.
2159 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2160 might be pseudo-atomic.
2161 * internal change: Stopping for GC is now done with signals not
2162 ptrace. GC is now done in whichever thread wanted it, instead of
2164 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2166 * bug fix: result form in DO is not contained in the implicit
2168 * incompatible change: ICR structure is changed; the value part of
2169 CONTINUATION is now called LVAR; corresponding functions are
2170 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2171 * added type deriver for ISQRT (thanks to Robert E. Brown).
2172 * bug fix: better support for loading from the command line when an
2173 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2175 * fixed some bugs revealed by Paul Dietz' test suite:
2176 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2177 ** ROUND and FROUND now give the right answer when given very
2178 small float arguments.
2179 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2181 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2182 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2183 ** LCM should return a non-negative integer.
2184 ** PARSE-INTEGER returned the index of a terminator instead of the
2185 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2186 ** PARSE-INTEGER returned an incorrect index being applied to a
2188 ** LCM with two arguments of 0 returns 0 rather than signalling
2190 ** unsigned addition of a 32-bit constant with the high bit set no
2191 longer causes an internal compiler error on the x86.
2192 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2193 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2194 with negative last argument.
2195 ** byte specifiers with zero size and position no longer cause
2196 an error during type derivation.
2197 ** bignum multiplication on the Alpha platform now returns the
2199 * porting: The system now builds on SuSE AMD64, although it still
2200 generates a 32-bit binary.
2201 * .fasl file incompatibility: The fasl file version number has
2202 been incremented (because of the changes to internal compiler
2203 data structures referred to above).
2205 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2206 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2207 more accurately, on the Darwin kernel running on PowerPC hardware.
2208 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2209 * Compiler code deletion notes now signal a condition of type
2210 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2211 an associated MUFFLE-WARNING restart.
2212 * The compiler now performs limited argument count validation of
2213 constant format strings in FORMAT, and where appropriate in ERROR,
2214 CERROR and WARN. (thanks to Gerd Moellmann)
2215 * New ASDF-INSTALL contrib can be used for automatic download and
2216 installation of third-party Lisp code from CCLAN or other sites
2218 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2219 which forces another thread to execute a function supplied by the
2221 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2222 now accept and act upon their :ELEMENT-TYPE keyword argument.
2223 (reported by Martin Atzmueller, Edi Weitz)
2224 * bug fix: FILE-POSITION now accepts position designators up to
2225 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2226 is the greater. (thanks to Patrik Nordebo)
2227 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2228 Kalle Olavi Niemitalo)
2229 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2230 printable. (reported by Eric Marsden)
2231 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2232 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2233 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2234 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2235 (thanks to Stig Sandoe)
2236 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2237 UNREAD-CHAR now correctly includes the unread character in the
2238 target sequence. (thanks to Gerd Moellmann)
2239 * bug fix in threaded builds: the system can now be suspended and
2240 resumed by shell job control with minimal disruption.
2241 * bug fixes in times and timezones >2038AD
2242 * better handling of "where is GNU make?" problem in build scripts
2243 (thanks to Nikodemus Siivola)
2244 * new optimization: inside a named function any reference to a
2245 function with the same name is considered to be a self-reference;
2246 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2247 optimization quality.
2248 * new optimization on x86: logical functions and + now have
2249 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2250 used when the result is truncated to 32 bits.
2251 * VALUES declaration is partially enabled.
2252 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2253 * fixed some bugs revealed by Paul Dietz' test suite:
2254 ** The system now obeys the constraint imposed by
2255 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2256 types form a lattice under type intersection.
2257 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2258 ** ASSOC now ignores NIL elements in an alist.
2259 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2260 and (1+ MOST-POSITIVE-FIXNUM) answers.
2261 ** The addition of a method with invalid qualifiers to a generic
2262 function does not cause an error to be signalled immediately;
2263 a warning is signalled, and the error is generated only on
2264 calling the generic function.
2265 * changed .fasl file version number, in part to add type codes for
2266 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2267 obscure ANSI requirements
2269 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2270 * fixed bug 148: failure to inline-expand a local function left
2271 garbage, confusing the compiler.
2272 * fixed bugs 3cd: structure slot readers perform type check if the
2273 slot can have an invalid value (i.e. it is either not initialized
2274 or can be written with a less specific slot writer).
2275 * bug fix: the compiler now traps array references to elements off
2276 the end of an array; previously, the bounds checking in some
2277 circumstances could go off-by-one.
2278 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2280 * type declarations for array element types now obey the description
2281 on the CLHS page "Declaration TYPE", as per discussions on
2282 sbcl-help around 2003-05-08. This means that a declaration
2283 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2284 declaration, all references to BAR will be asserted or assumed
2285 (with THE, so dependent on compiler policy) to involve objects of
2286 type FOO. Note that no such declaration is implied in
2287 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2288 * declared types of functions from the "Conditions"
2289 chapter. (reported by Paul Dietz)
2290 * bug fix: CERROR accepts a function as its first argument.
2291 * bug fix: NTH an NTHCDR accept a bignum as index
2292 arguments. (reported by Adam Warner)
2293 * optimization: character compare routines now optimize comparing
2294 against a constant character. (reported by Gilbert Baumann)
2295 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2296 argument now works properly on the MIPS platform.
2297 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2298 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2299 declared to have a functional type, does not cause an error
2301 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2302 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2303 * fixed bug: if last continuation of a deleted block has a
2304 destination, this destination should be deleted too. (reported by
2306 * fixed a bug in the bootstrap process: the host compiler's values
2307 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2308 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2309 test case from Patrik Nordebo)
2310 * improved the ability of the disassembler on the PPC platform to
2311 provide helpful disassembly notes.
2312 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2313 the class in more cases than previously.
2314 * bug fix: FILE-POSITION now understands :START and :END for
2315 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2316 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2317 8) data. (thanks to David Lichteblau)
2318 * bug fix: it is possible to add a method to a generic function
2319 without lambda list.
2320 * bug fix: reader failed to signal END-OF-FILE inside an
2321 object representation. (reported by Nikodemus Siivola)
2322 * fixed some bugs revealed by Paul Dietz' test suite:
2323 ** LAST and [N]BUTLAST should accept a bignum.
2324 ** condition slot accessors are methods.
2325 ** (VECTOR NIL) is a subtype of STRING.
2327 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2328 * minor incompatible change: some nonsensical specialized lambda
2329 lists (used in DEFMETHOD) which were previously ignored now signal
2331 * minor incompatible change: the system is now aware of the types of
2332 variables in the COMMON-LISP package, and will signal errors for
2333 most violations of these type constraints (where previously they
2334 were silently accepted).
2335 * minor incompatible change: COMPILE-FILE now uses the freedom
2336 afforded (ANSI 3.2.2.3) to use derived function types for
2337 functions defined in the same file. This also permits the system
2338 to warn on static type mismatches and function
2339 redefinition. (Currently it does not work with high DEBUG level.)
2340 * minor incompatible change: VALUES declaration is disabled.
2341 * When issuing notes, the compiler now signals a condition of type
2342 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2343 restart for use in user handlers. It is expected that the
2344 COMPILER-NOTE condition will eventually become a condition
2345 supertype to a hierarchy of note types, which will then be
2346 handleable in a similar fashion. However, at the moment, no such
2347 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2348 but it's an implementation detail, not a classification for the
2350 * Changes in type checking closed the following bugs:
2351 ** type checking of unused values (192b, 194d, 203);
2352 ** template selection based on unsafe type assertions (192c, 236);
2353 ** type checking in branches (194bc).
2354 * A short form of VALUES type specifier has ANSI meaning (it has
2355 increased the number of situations when SBCL cannot perform type
2357 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2358 keywords or constants is permissible.
2359 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2360 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2361 classes. (thanks to Antonio Martinez)
2362 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2363 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2364 argument to a pretty-print stream. Instead, it leaves any such
2365 operations to DESCRIBE-OBJECT methods.
2366 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2368 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2369 lambda lists. (thanks to David Lichteblau)
2370 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2371 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2372 (thanks to Gerd Moellmann)
2373 * fixed bug 246: increased compilation speed of long
2374 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2376 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2378 * DEFCONSTANT now throws a condition of type
2379 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2380 constant to a non-EQL value; CONTINUE and ABORT restarts
2381 respectively change and preserve the value.
2382 * fixed bug 63: The code walker, part of the implementation of CLOS,
2383 is now better at handling symbol macros.
2384 * bug fix: There is no longer an internal implementation type named
2385 CL:LENGTH. (reported by Raymond Toy)
2386 * bug fix: In macro-like defining macros/special operators the
2387 implicit block does not enclose the lambda list.
2388 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2389 enclosed in parentheses) are not suitable as type specifiers, and
2390 their use properly signals an error now.
2391 * bug fix: An argument count mismatch for a type specifier in code
2392 being compiled no longer causes an unhandled error at compile
2393 time, but signals a compile-time warning.
2394 * fixed simple vector readable printing
2395 * bug fix: DESCRIBE takes more care over whether the class
2396 precedence list slot of a class is bound before accessing it.
2397 (reported by Markus Krummenacker)
2398 * bug fix: FORMATTER can successfully compile pretty-printer format
2399 strings which use variants of the ~* directive inside.
2400 * bug fix: SEARCH now applies its TEST predicate to the elements of
2401 the arguments in the correct order. (thanks to Wolfhard Buss)
2402 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2403 in type inference. (thanks to Robert E. Brown)
2404 * bug fix: Reading in symbols with an explicit package name of ""
2405 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2406 * fixed some bugs revealed by Paul Dietz' test suite:
2407 ** NIL is now allowed as a structure slot name.
2408 ** Arbitrary numbers, not just REALs, are allowed in certain
2409 circumstances in LOOP for-as-arithmetic clauses.
2410 ** Multiple class redefinitions before slot access no longer
2411 causes a type error.
2412 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2413 association between the name and a class.
2414 ** Generic functions with non-standard method-combination and over
2415 five methods all of which return constants no longer return NIL
2416 after the first few invocations. (thanks to Gerd Moellmann)
2417 ** CALL-NEXT-METHOD with no arguments now passes the original
2418 values of the arguments, even in the presence of assignment.
2419 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2421 ** DISASSEMBLE works with closures and funcallable instances.
2422 ** ADD-METHOD now returns the generic function, not the new method.
2423 ** FIND-METHOD signals an error if the lengths of the specializers
2424 is incompatible with the generic function, even if the ERRORP
2426 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2427 which its argument is a member.
2428 ** DEFCLASS only redefines the class named by its class-name
2429 argument if that name is the proper name of the class;
2430 otherwise, it creates a new class.
2431 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2432 of the UNBOUND-SLOT condition to the name of the slot.
2433 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2434 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2435 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2436 SLOT-MAKUNBOUND in the specified fashion.
2438 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2439 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2440 cross-compilation host. As a consequence, we can now bootstrap our
2441 way up to SBCL starting with a bare gcc toolchain and human-readable
2442 source code (first the source to CLISP, then the source to SBCL).
2443 * A contributed module containing a partial implementation of the
2444 simple-streams interface has been included. (thanks to Rudi
2446 * A contributed module implementing the RFC1321 Message Digest
2447 Algorithm, known as MD5, has been included.
2448 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2449 no longer has any effect, as the code controlled by this feature
2450 has been deleted. (As far as we know, no-one has ever built using
2451 this feature, and its semantics were confused in any case).
2452 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2453 work on conditions (as required by the ANSI specification),
2454 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2455 expected behaviour on conditions. Users should note, however,
2456 that such behaviour is not required by the ANSI specification,
2457 and so use of this behaviour may render their code unportable.
2458 * fixed some bugs revealed by Paul Dietz' test suite:
2459 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2461 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2462 ** SXHASH on condition objects no longer returns NIL.
2463 ** :ALLOCATION :CLASS slots are better treated; their values are
2464 updated on class redefinition, and initforms inherited from
2465 superclasses are applied.
2466 ** REMOVE-METHOD returns its generic function argument even when
2467 no method was removed.
2468 ** SHARED-INITIALIZE now initializes the values of the requested
2469 slots, including those with :ALLOCATION :CLASS.
2470 ** ALLOCATE-INSTANCE now works on structure classes defined via
2471 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2473 ** SLOT-EXISTS-P now works on conditions, as well as structures
2475 ** MAKE-LOAD-FORM now has the required methods on
2476 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2477 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2478 keyword, and now implements the SLOT-NAMES argument.
2479 ** methods with &OPTIONAL arguments no longer allow too many
2480 arguments to be passed in the call without error.
2481 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2482 option is consistent with the required arguments of the generic
2483 function lambda list.
2484 * bug fix: REQUIRE accepts a string designator. (Thanks to
2486 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2487 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2488 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2489 * bug fix: The compiler now checks for duplicated variables in macro
2491 * bug fix: SETQ on globals returns the correct value.
2492 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2493 not a COMPILER-ERROR (followed by some other strange error on
2494 choosing the CONTINUE restart).
2495 * bug fix: make.sh and friends are now more consistent in the way that
2496 they look for GNU "make".
2498 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2499 * experimental native threads support (on x86 Linux >=2.4 only).
2500 This is not compiled in by default: you need to add :SB-THREAD to
2501 the target features. See the "Beyond ANSI" chapter of the manual
2503 * fix for longstanding nonANSIism: The old distinction between
2504 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2505 The return value from CL:FIND-CLASS is now a CLOS class, and
2506 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2507 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2508 * An interface to the MetaObject Protocol, as described in Kiczales,
2509 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2510 MIT Press, 1991, is available from the SB-MOP package.
2511 * incompatible change: the SB-PCL package should now be considered
2512 a private implementation detail, and no longer a semi-private MOP
2514 * minor incompatible change: due to rearrangement for threads, the
2515 control stack and binding stack are now allocated at arbitrary
2516 addresses instead of being hardcoded per-port. Users affected by
2517 this probably have to be doing advanced things with shared
2518 libraries, and will know who they are.
2519 * minor incompatible change: Previously, all --eval forms used were
2520 processed with READ before any of them were processed with EVAL.
2521 Now each --eval form is processed with both READ and EVAL before
2522 the next --eval form is processed. (Thus package operations like
2523 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2524 work as the user might reasonably expect.)
2525 * minor incompatible change: *STANDARD-INPUT* is now only an
2526 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2528 * minor incompatible change: Y-OR-N-P is now character-oriented,
2529 not line oriented. Also, YES-OR-NO-P now works without errors.
2530 (thanks to Antonio Martinez)
2531 * sb-aclrepl module improvements: an integrated inspector, added
2532 repl features, and a bug fix to :trace command.
2533 * Known functions, which cannot be open coded by the backend, are
2534 considered to be able to check types of their arguments. (fixing
2535 a bug report by Nathan J. Froyd)
2536 * fixed a bug in computing method discriminating functions: It is
2537 now possible to define methods specialized on classes which have
2538 forward-referenced superclasses. (thanks to Gerd Moellmann)
2539 * fixed evaluation order in optional entries (reported by Gilbert
2541 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2542 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2543 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2544 required-by-AMOP NAME argument, as well as CLASS and
2545 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2546 * fixed bug 20: DEFMETHOD can define methods using names that are
2547 not the proper names of classes to designate class specializers.
2548 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2549 Unix file descriptors, instead of blowing up. (thanks to Antonio
2551 * Garbage collection refactoring: user-visible change is that a
2552 call to the GC function during WITHOUT-GCING will not do garbage
2553 collection until the end of the WITHOUT-GCING. If you were doing
2554 this you were probably losing anyway.
2555 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2556 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2557 TYPEP the latter but not the former.
2558 * The compiler issues a full WARNING for calls to undefined functions
2559 with names from the CL package.
2560 * MAP-INTO for a vector destination is open coded. (reported by
2561 Brian Downing on c.l.l)
2562 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2563 documentation string.
2564 * fixed some bugs revealed by Paul Dietz' test suite:
2565 ** COPY-ALIST now signals an error if its argument is a dotted
2567 ** Condition slots are now accessed more correctly in the presence
2568 of multiple initargs for a given slot.
2569 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2570 exclude restarts of the same name associated with a different
2572 ** DEFCLASS of forward-referenced classes with another
2573 forward-referenced class in the superclasses list no longer
2575 ** Condition slots are now initialized once each, not multiple
2576 times. (thanks to Gerd Moellmann)
2577 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2579 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2580 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2581 arguments contain duplicated elements.
2582 ** RESTART-CASE understands local macros.
2583 ** RESTART-CASE associates exactly its own restarts with a condition.
2584 ** ENDP in safe mode checks its argument to be of type LIST.
2585 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2586 in question is unbound.
2587 ** Optimized MAKE-INSTANCE functions no longer cause internal
2588 assertion failures in the presence of duplicate initargs.
2589 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2590 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2591 propagated even in the ctor optimized implementation of
2593 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2595 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2596 * a better implementation of SXHASH on (simple) bit vectors,
2597 measured both in execution speed and in distribution of results
2598 over the positive fixnums, has been installed. Likewise, a better
2599 implementation of EQUAL for simple bit vectors is now available.
2600 * fixed CEILING optimization for a divisor of form 2^k.
2601 * fixed bug 240 (emitting extra style warnings "using the lexical
2602 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2603 by Antonio Martinez)
2604 * fixed SXHASH, giving different results for NIL depending on type
2605 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2606 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2607 variable. (found by Rolf Wester)
2608 * SBCL does not ignore type declarations for special
2609 variables. (reported by rif on c.l.l 2003-03-05)
2610 * some bug fixes in contrib/sb-aclrepl/
2611 * fixed some bugs revealed by Paul Dietz' test suite:
2612 ** a bug in the CONS type specifier, whereby the CAR and CDR
2613 types got intertwined, has been fixed;
2614 ** the type system is now able to reason about the interaction
2615 between INTEGER and RATIO types more completely;
2616 ** APPEND, [N]REVERSE and NRECONC check that those their
2617 arguments, which must be proper lists, are really so;
2618 ** An array specialized to be unable to hold elements has been
2619 implemented, as required -- yes, really -- by ANSI;
2620 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2621 on malformed property lists;
2623 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2624 * incompatible packaging change: in line with Unix convention,
2625 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2626 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2627 * REQUIRE and PROVIDE are now optionally capable of doing something
2628 useful. See the documentation string for REQUIRE.
2629 * infrastructure for a managed SBCL contrib system: contributed
2630 modules in this release include:
2631 ** the ASDF system definition facility;
2632 ** an interface to the BSD Sockets API;
2633 ** an ACL-like convenience interface to the repl;
2634 (thanks to Kevin Rosenberg)
2635 ** an implementation of ROTATE-BYTE, with efficient implementation
2637 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2638 gives a closer approximation to the right answer than previously.
2639 (thanks to Raymond Toy)
2640 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2641 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2642 optional environment arguments, as required by ANSI.
2643 * fixed bugs in other functions taking environment objects, allowing
2644 calls with an explicit NIL environment argument to be compiled
2646 * fixed bug 228: primary return values from
2647 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2648 COMPILE or FUNCTION.
2649 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2650 :TYPE will no longer signal errors on innocuous objects.
2651 * fixed bug 231b: SETQ is better at respecting type declarations in
2652 the lexical environment.
2653 * fixed a bug in DEFCLASS: classes named by symbols with no or
2654 unprintable packages can now be defined.
2655 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2656 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2657 * fixed compiler failure related to checking types of functions.
2658 (reported by Robert E. Brown)
2659 * the compiler is now much more consistent in its error-checking
2660 treatment of bounding index arguments to sequence functions: in
2661 (SAFETY 3) code, errors will be signalled in almost all cases if
2662 invalid sequence bounding indices are passed to functions defined
2663 by ANSI to operate on sequences.
2664 * fixed a bug in the build procedure: documentation of SBCL-specific
2665 packages is now preserved and available in the final Lisp image.
2666 * lifted FDEFINITION lookup out of loops in the implementation of
2667 many list operations. (thanks to Robert E. Brown)
2668 * fixed a bug in the reader: the #n# reader macro now works for
2669 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2670 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2671 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2672 E. Brown and Tony Martinez respectively)
2673 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2674 SLOT-BOUNDP: in particular, it is now able to optimize them much
2675 better, and is now not vulnerable to having packages renamed.
2676 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2677 always called when a slot is not present in an instance. (thanks
2679 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2680 longer signal an error on unreading a character following EOF on
2681 the previous constituent stream. (thanks to Tony Martinez)
2682 * fixed some bugs revealed by Paul Dietz' test suite:
2683 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2684 not just nonnegative fixnums;
2685 ** the logical bit-array operators such as BIT-AND now accept an
2686 explicit NIL for their "opt-arg" argument (to indicate a
2687 freshly-consed result bit-array);
2688 ** ELT now signals an error on an invalid sequence index in safe
2690 ** the type system is now cleverer about negations of numeric
2691 types, and consequently understands the BIGNUM and RATIO types
2693 ** the type system is now cleverer about the interaction between
2694 INTEGER and RATIO types: while bugs still remain, many more
2695 cases are accurately computed;
2696 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2697 if it is in the last clause;
2698 ** CONSTANTLY now correctly returns a side-effect-free function in
2700 ** DECLARE is no longer treated as a special-operator; in
2701 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2702 * incremented fasl file version number due to the change in the
2703 DEFSTRUCT-SLOT-DESCRIPTION structure.
2705 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2706 * minor incompatible change: code processed by the "interpreter" or
2707 EVAL now has a compilation optimization policy of (DEBUG 2)
2708 (changed from (DEBUG 1)) to improve debuggability of interactive
2709 development, and to allow the use of the debug RETURN command in
2711 * an experimental implementation of the RETURN command for the
2712 debugger has been included. (thanks to Frederik Kuivinen)
2713 * fixed bug 62: constraints were not propagated into a loop.
2714 * fixed bug in embedded calls of SORT (reported and investigated by
2716 * fixed some bugs revealed by Paul Dietz' test suite:
2717 ** printing and reading of arrays with some dimensions having
2718 length 0 (thanks to Gerd Moellmann);
2719 ** BOA constructor with &AUX argument without a default value does
2720 not cause a type error;
2721 ** CONSTANTP now returns true for all self-evaluating objects.
2723 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2724 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2725 accessors that are related by inheritance, as specified in the
2726 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2727 Valtteri Vuorikoski)
2728 * The compiler is now able to inline functions that were defined in
2729 a complex lexical environment (e.g. inside a MACROLET).
2730 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2732 * Support for the upcoming FreeBSD-5.0 release has been included.
2733 (thanks to Dag-Erling Smorgrav)
2734 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2735 effect when it is not in a toplevel context.
2736 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2737 stand a better chance of being correct. (thanks to Gerd
2739 * fixed bug in COERCE, which now signals an error on coercing a
2740 rational to a bounded real type which excludes the expected
2742 * The compiler is now able to derive types more accurately from the
2743 COERCE and COMPILE functions.
2744 * fixed bug 223: functional binding is considered to be constant
2745 only for symbols in the CL package.
2746 * fixed bug 231: SETQ did not check the type of a variable being set
2747 (reported by Robert E. Brown)
2748 * A new optimization for MAKE-INSTANCE has been included, fixing
2749 various bugs (including relating to :ALLOCATION :CLASS slots and
2750 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2752 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2753 ** As required by ANSI, LOOP now disallows anonymous collection
2754 clauses such as COLLECT I in conjunction with aggregate boolean
2755 clauses such as THEREIS (= I 1);
2756 ** LOOP now signals an error when any variable is reused in the
2757 same loop (including the potentially useful construct analogous
2758 to WITH A = 1 WITH A = (1+ A);
2759 ** IT is only a special loop symbol within the first clause of a
2760 conditional loop clause;
2761 ** LOOP with a typed iteration variable over a hashtable now
2762 signals a type error iff it should.
2763 * fixed some other bugs revealed by Paul Dietz' test suite:
2764 ** FILE-STREAM now names the class previously known as FD-STREAM;
2765 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2766 argument) no longer signals an error;
2767 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2768 of the slot symbol, rather than using the current package
2769 ((:CONC-NAME "") continues to intern the slot's name in the
2771 * incremented fasl file version number, because of the incompatible
2772 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2773 of the new implementation of DEFINE-COMPILER-MACRO.
2775 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2776 * Support for building SBCL for MIPS platforms running in
2777 little-endian mode has now been checked in, and basic
2778 functionality on said platforms verified.
2779 * minor incompatible change: PCL now records the pathname of a file
2780 in which methods and the like are defined, rather than its
2782 * minor incompatible change: TRUENAME now considers the truename of
2783 a file naming a directory to be the pathname with :DIRECTORY
2784 component indicating that directory.
2785 * minor incompatible change: a NAMED clause in the extended form of
2786 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2787 reason for the previous behaviour is unclear.
2788 * more systematization and improvement of CLOS and MOP conformance
2789 in PCL (thanks to Gerd Moellman and Pierre Mai):
2790 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2792 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2793 primary methods with no specializers;
2794 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2796 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2797 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2798 FORWARD-REFERENCED-CLASSes; error reporting on
2799 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2801 ** SXHASH on CLOS instances now uses a slot internal to the
2802 instance to return different numbers on distinct instances,
2803 while preserving the same return value through invocations of
2805 ** DEFMETHOD signals errors when methods with longer incongruent
2806 lambda lists are added to generic functions;
2807 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2808 CLASS, as specified in AMOP;
2809 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2810 on the order returned by the primary method for classes of
2811 class STANDARD-CLASS;
2812 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2813 * fixed some bugs shown by Paul Dietz' test suite:
2814 ** DOLIST puts its body in TAGBODY;
2815 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2817 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2818 value producing form;
2819 ** if more variables are given to PROGV than values, extra
2820 variables are bound and made to have no value;
2821 ** NSUBSTITUTE on list arguments gets the right answer with
2823 ** ELT signals an error of type TYPE-ERROR when the index argument
2824 is not a valid sequence index;
2825 ** LOOP signals (at macroexpansion time) an error of type
2826 PROGRAM-ERROR when duplicate variable names are found;
2827 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2828 ** FUNCALL of special-operators now cause an error of type
2830 ** PSETQ now works as required in the presence of side-effecting
2831 symbol-macro places;
2832 ** NCONC accepts any object as its last argument;
2833 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2835 ** loop-for-as-package does not require a package to be explicitely
2837 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2839 * fixed bug 166: compiler preserves "there is a way to go"
2840 invariant when deleting code.
2841 * fixed bug 172: macro lambda lists with required arguments after
2842 &REST arguments now cause an error to be signalled. (thanks to
2844 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2845 bug. (thanks to Matthew Danish)
2846 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2848 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2849 arguments are lexically rebound. (thanks to Gerd Moellmann and
2851 * fixed bug 194: error messages are now more informative when there
2852 is no primary method applicable in a call to a generic
2853 function. (thanks to Gerd Moellmann)
2854 * fixed bug in command line argument checking (thanks to Julian
2856 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2857 (thanks to Matthew Danish)
2858 * incremented fasl file version number, because of the
2859 SXHASH-related changes in the layout of CLOS data structures
2861 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2862 * minor incompatible change: The runtime (the Unix executable named
2863 "sbcl") is now much pickier about the .core files it will load.
2864 Essentially it now requires .core files to descend from the same
2865 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2866 as the runtime does. (The intent is to prevent the crashes which
2867 can occur, and which can even be reported as mysterious failures,
2868 when people patch the sources or change the build parameters
2869 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2870 sbcl and .core files.)
2871 * fixed bug: VALUES-LIST is no longer optimized away.
2872 * fixed bug 142: The FFI conversion of C string values to Lisp
2873 string values no longer conses excessively. (thanks to Nathan
2874 Froyd porting Raymond Toy's fix to CMU CL)
2875 * began to systematize and improve MOP conformance in PCL (thanks to
2876 Nathan Froyd, Gerd Moellman and Pierre Mai):
2877 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2879 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2880 ** FINALIZE-INHERITANCE is now called on class finalization;
2881 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2882 argument precedence order.
2883 * fixed bug 202: The compiler no longer fails on functions whose
2884 derived types contradict their declared type.
2885 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2886 so it can be non-toplevel.
2887 * The fasl file version number has changed (because of the new
2888 implementation of DEFMACRO).
2889 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2890 safe code, that any length requirement by their type-specifier
2891 argument is valid. The exceptions to this are described in bug
2893 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2894 functions now check the types of their inputs as required by ANSI.
2895 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2896 introduced symbol is DECLAREd to be SPECIAL.
2897 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2898 * fixed bug: :COUNT argument to sequence functions may be negative
2899 * fixed bug: body of DO-SYMBOLS may contain declarations
2900 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2901 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2902 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2903 symbol macro only once
2904 * fixed printing of call frame when argument list is unavailable
2905 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2906 * compiler no longer signals WARNING on unknown keyword
2909 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2910 * A beta-quality port to the mips architecture running Linux,
2911 based on the old CMUCL backend, has been made. It has been tested
2912 on a big-endian kernel, and works sufficiently well to be able to
2913 rebuild itself; it has not been tested in little-endian mode.
2914 * fixed an inconsistency between gencgc.c and purify.c which made
2915 dumping/loading .core files unreliable
2916 * fixed bug 120a: The compiler now deals correctly with IFs where
2917 the consequent is the same as the alternative, instead of
2918 misderiving the return type. (thanks to Alexey Dejneka)
2919 * fixed bug 113: Logical pathnames are now dumpable (the logical
2920 host is resolved at load-time, throwing an error if it is not
2922 * fixed bug 174: FORMAT's error message is slightly clearer when a
2923 non-printing character is used in a format directive.
2924 * fixed several bugs in compiler checking of type declarations, i.e.
2925 violations of the Python "declarations are assertions" principle
2926 (thanks to Alexey Dejneka)
2927 * fixed several bugs in PCL's error checking (thanks to Gerd
2929 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2931 * fixed bug in compilation of functions as first class values
2932 (thanks to Antonio Martinez-Shotton)
2933 * The compiler's handling TYPE-ERRORs which it can prove will
2934 inevitably happen at runtime has been cleaned up and corrected
2935 in several ways. (thanks to Alexey Dejneka)
2936 * improved argument type checking for various basic arithmetic
2937 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2938 lavished on them in the past that they can be compiled in many
2939 ways in different special cases
2940 * fixed bug 181: compiler checks validity of user supplied type
2942 * cleaned up code flushing in optimization: Function calls which
2943 should signal errors for safety purposes (e.g. which ANSI says
2944 should signal errors when their arguments are of incorrect type)
2945 are no longer optimized away.
2946 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2947 * incremented fasl file version number, because changes in the
2948 implementation of sequence functions like COERCE caused
2949 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2950 in old inline expansions) to become undefined. (Actually these
2951 changes were later undone, so we might very well be binary
2952 compatible with 0.7.7 after all, but leaving the version number
2953 incremented seemed like the simplest and most conservative
2956 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2957 * An alpha-quality port to the parisc architecture running Linux,
2958 based on the old CMUCL backend, has been made. This, even more so
2959 than the other backends, should be considered still a work in
2960 progress; known problems include that the Linux kernel in 64-bit
2961 mode does not propagate the correct sigcontext structure to
2962 userspace, and consequently SBCL on a parisc64 kernel will not
2964 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2965 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2966 suppressed.) Also now that the compiler is looking at declarations
2967 in the environment, it checks optimization declarations as well,
2968 and suppresses inlining when (> DEBUG SPEED).
2969 * More fixes have been made to treatment of floating point exception
2970 treatment and other Unix signals. In particular, floating point
2971 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2972 * The detection and handling of control stack exhaustion (infinite
2973 or very deeply nested recursion) has changed. Stack exhaustion
2974 detection is now done by write-protecting pages at the OS level
2975 and applies at all optimization settings; when found, a
2976 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2977 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2978 be caught using IGNORE-ERRORS.
2979 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2980 correctly when transforming recursive LABELS functions to LETs.
2981 (thanks to Alexey Dejneka)
2982 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2983 that are names of constants or global variables.
2984 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2985 alien routines with docstrings.
2986 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2987 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2989 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2990 (thanks to Eric Marsden)
2991 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2992 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2993 * Bug fix: The SPARC backend can now compile functions involving
2994 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2995 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2996 to a functions expecting a PATHNAME-DESIGNATOR.
2997 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2998 lambda lists. (thanks to Alexey Dejneka)
2999 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3000 functions so that EXPORTing the name of the class doesn't cause
3001 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3002 to Antonio Martinez for reporting this)
3003 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3004 constant-folded without failing an assertion. (thanks to Einar
3005 Floystad Dorum for reporting this)
3006 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3007 types are now dealt with more correctly by the compiler.
3008 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3009 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3010 behaviour analogous to RENAME-FILE. This puts its behaviour more
3011 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3013 * The fasl file version number has changed again. (because of the
3014 bug fix involving the names of PCL MAKE-INSTANCE functions)
3016 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3017 * bug fix: Floating point exceptions are treated much more
3018 consistently on the x86/Linux and PPC/Linux platforms.
3019 * Array initialization with :INITIAL-ELEMENT is now much faster for
3020 cases when the compiler cannot open code the array creation, but
3021 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3022 array accesses have also seen a speed increase.
3023 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3024 specified. (This was at the root of some bad interactions between
3025 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3027 * bug fix: Internal error arguments for undefined functions are now
3028 computed correctly on the PPC/Linux platform.
3029 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3030 Raymond Toy's patch for CMU CL)
3031 * Support for the Solaris 9 operating environment has been included
3032 (thanks to Daniel Merritt)
3033 * A very ugly but hopefully complete draft of the missing FFI chapter
3034 of the manual has been created by reformatting the corresponding
3035 CMU CL manual chapter into (currently very ugly and incoherent)
3036 DocBook and bringing it up to date for SBCL behavior. Thus, the
3037 manual is now essentially complete, at least by my extreme
3038 once-and-only-once standards, whereby it's acceptable to refer to
3039 the doc strings of SB-EXT functions as the primary documentation.
3040 * The fasl file version number has changed again, due to cleanup of
3041 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3042 is no longer a static symbol.)
3044 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3045 * SBCL now builds with OpenMCL (version 0.12) as the
3046 cross-compilation host; also, more progress has been made toward
3047 bootstrapping under CLISP.
3048 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3050 * bug 158 fixed: The compiler can now deal with integer loop
3051 increments different from 1; fixing this turned out also to fix
3053 * bug 169 fixed: no more bogus warnings about using lexical bindings
3054 despite the presence of perfectly good SPECIAL declarations (thanks
3055 to David Lichteblau)
3056 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3057 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3058 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3060 * bug 180 fixed: Method combination specifications no longer ignore
3061 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3062 * bug fix: Structure type predicate functions now check their argument
3063 count as they should.
3064 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3065 correctly. (thanks to Pierre Mai)
3066 * minor incompatible change: The --noprogrammer option is deprecated
3067 in favor of the new --disable-debugger option, which is very similar.
3068 (The major difference is that it takes effect at a slightly different
3069 time at startup, causing handling of errors in --sysinit and
3070 --userinit files will be affected differently.) The
3071 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3072 been added to allow this functionality to be controlled from ordinary
3073 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3074 maintainers, who might want to run non-interactive scripts to
3075 build SBCL cores which will later be used interactively.)
3076 * minor incompatible change: The LOAD function no longer, when given
3077 a wild pathname to load, loads all files matching that pathname.
3078 Instead, an error of type FILE-ERROR is signalled.
3080 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3081 * bug 147 fixed: The compiler preserves its block link/count
3082 invariants more correctly now so that it doesn't crash. (thanks
3084 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3086 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3087 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3088 no-value-for-this-character as NIL (as specified by ANSI).
3089 * HOST-NAMESTRING on physical pathnames now returns a string that is
3090 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3091 (thanks to Christophe Rhodes)
3092 * The Alpha port handles icache flushing more correctly. (thanks to
3094 * More progress has been made toward bootstrapping under CLISP. (thanks
3095 to Christophe Rhodes)
3096 * The fasl file format has changed again, because dynamic loading
3097 on OpenBSD (which has non-ELF object files) motivated some cleanups
3098 in the way that foreign symbols are transformed and passed around.
3099 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3100 is no longer treated as whitespace by the reader, but instead as
3101 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3102 |A
\7fB|, instead of A as it used to.
3104 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3105 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3106 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3107 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3108 this as of 0.7.1.45, but was left out of the previous news section
3109 (thanks to Dan Barlow)
3110 * SBCL now runs on the Solaris operating system on SPARC architectures
3111 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3112 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3113 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3114 * SPARC backend cleanups, allowing builds of cores optimized for V8
3115 and V9 SPARCS, and also emission of code targeted to a particular
3116 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3118 * SBCL is closer to bootstrapping under CLISP, thanks to various
3119 fixes by Christophe Rhodes.
3120 * The fasl file format has changed again, to allow the compiler's
3121 INFO database to support symbol macros.
3122 * The user manual (in doc/) is formatted into HTML more nicely.
3123 (thanks to coreythomas)
3124 * The system is smarter about SUBTYPEP relationships, especially
3125 those involving NOT types (including types such as ATOM which are
3126 represented internally using NOT types). Thus SUBTYPEP is less
3127 likely to return (VALUES NIL NIL) in general, and in particular
3128 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3130 * The fasl file format has changed again, because the internal
3131 representation of types now includes a new slot to support the new
3132 SUBTYPEP-of-NOT-types logic.
3133 * (not a change in the main branch of SBCL, but a related prototype
3134 which can hopefully be merged into the main branch of SBCL in the
3135 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3136 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3137 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3138 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3139 dispatch macro character combinations. (thanks to Alexey Dejneka)
3140 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3141 that superficially look like logical namestrings correctly.
3142 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3144 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3145 default supplied-p) for &optional and &key arguments. (thanks to
3148 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3149 * incompatible change: The compiler is now less aggressive about
3150 tail call optimization, doing it only when (> SPACE DEBUG) or
3151 (> SPEED DEBUG). (This is an incompatible change because there are
3152 programs which relied on the old CMU-CL-style behavior to optimize
3153 away their unbounded recursion which will now die of stack overflow.)
3154 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3155 for non-GENCGC systems has been increased to 20M (since that
3156 seems much closer to the likely performance optimum for modern
3157 systems than the old 4M value was)
3158 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3159 and *DEBUG-PRINT-LEVEL*
3160 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3161 of CMU CL's support for SPARC, and various endianness and other
3162 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3163 * new syntactic sugar for the Unix command line: --load foo.bar is now
3164 an alternate notation for --eval '(load "foo.bar")'.
3166 ** The system now detects stack overflow and handles it gracefully,
3167 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3168 optimization settings. (This is a good thing in general, and
3169 its introduction in this version should be particularly timely
3170 for anyone whose code fails because of suppression of tail
3172 ** The system now hunts for the C variable "environ" in a more
3173 devious way, to avoid segfaults when the C library version differs
3174 between compile time and run time. (thanks to Christophe Rhodes)
3175 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3176 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3177 ** The compiler no longer issues bogus style warnings for undefined
3178 classes in the same source file as the DEFCLASSes which defined
3179 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3181 ** fixes in CONDITION class precedence list for undefined function
3182 errors (thanks to Alexei Dejneka)
3183 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3184 correctly. (thanks to Dan Barlow)
3185 ** portability fixes aiming at bootstrapping under CLISP (thanks
3186 to Dave McDonald and Christophe Rhodes)
3187 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3188 ** fixes in type translation and and type inference (thanks to
3190 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3191 ** various fixes in the new ports (thanks to Dan Barlow)
3192 * several changes related to debugging:
3193 ** suppression of tail recursion, as noted above
3194 ** stack overflow detection, as noted above
3195 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3196 is now the default. (For some time encapsulation has been more
3197 reliable than the breakpoint-based :ENCAPSULATE NIL
3198 implementation, at least on X86 systems; and I just noticed that
3199 encapsulation also seems closer to the spirit of the ANSI
3202 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3204 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3205 up properly again. (There was a packaging bug in 0.7.0 which
3206 left their definitions in SB-SYS::LOAD-FOREIGN and
3207 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3208 vital for most things which interface to C-level interfaces,
3209 like extensions working with sockets or databases or
3210 Perl-compatible regexes or whatever, and the need to fix
3211 this bug is the main reason that 0.7.1 was released so
3213 ** DEFGENERIC is now choosier about the methods it redefines, so that
3214 reLOADing a previously-LOADed file containing DEFGENERICs does
3215 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3216 cycle now works as it should. (thanks to Alexey Dejneka)
3217 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3218 fixed by Christophe Rhodes. (He actually submitted this patch
3219 months ago, and I delayed until after 0.7.0.)
3220 ** Bug 111 (internal compiler confusion about runtime checks on
3221 FUNCTION types) was fixed by Alexey Dejneka.
3222 * Some internal cleanups (getting rid of variables which aren't
3223 needed now that the byte interpreter is gone) caused the fasl
3224 file format number to change again.
3226 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3227 * major incompatible change: The default fasl file extension, i.e. the
3228 default extension for files produced by COMPILE-FILE, has changed
3229 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3231 ** There are many changes in the implementation of the compiler.
3232 SBCL is now essentially a compiler-only implementation of ANSI
3233 Common Lisp. EVAL still "interprets" a few special cases, but
3234 almost all the interesting cases are handled by creating
3235 a LAMBDA expression, calling COMPILE on it, then calling
3236 FUNCALL on the result.
3237 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3238 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3239 Since the code is newer, there might still be some new bugs
3240 (though not as many as before Martin Atzmueller's fixes:-). But
3241 the new code is substantially simpler and clearer, and hopefully
3242 any remaining bugs will be simpler, less fundamental, and more
3243 fixable then the bugs in the old code.
3244 ** The revised compiler is still a little unsteady on its feet.
3246 *** The debugging information it produces (particularly the names
3247 of FUNCTION objects) is sometimes much less useful than what
3248 the old compiler produced.
3249 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3250 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3251 within a MACROLET) has been temporarily weakened.
3252 ** There are new compiler optimizations for various functions:
3253 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3254 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3255 *** the math functions TRUNCATE, FLOOR, and CEILING
3256 *** the function-of-all-trades COERCE
3257 Mostly these should be transparent, but there's one
3258 potentially-annoying problem (bug 117): when the compiler
3259 inline-expands a function and does type analysis on the result,
3260 it can create control paths which have type mismatches, and
3261 when it can't prove that those control paths aren't taken,
3262 it will issue WARNINGs about the type mismatches. This is
3263 a particular problem in practice for the new sequence functions.
3264 It's not clear how this should be fixed, and for now, a
3265 workaround is given in the entry for 117 in the BUGS file.
3266 ** (Because of the interaction between the two previous items --
3267 occasional inlining problems and new inline expansions -- some
3268 of the new sequence function optimizations won't really kick in
3269 completely until debugging information, and then inlining, are
3270 straightened out in some future version.)
3271 * minor incompatible changes:
3272 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3273 DIRECTORY no longer implicitly promotes NIL slots of its
3274 pathname argument to :WILD. In particular, when you ask for the
3275 contents of a directory (which you used to be able to do without
3276 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3277 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3278 ** changes in behavior that ANSI explicitly defines to be
3279 implementation dependent:
3280 *** The new compiler-only implementation still conforms with ANSI,
3281 but acts a little different than before. Besides the obvious
3282 changes in performance tradeoffs (that the cost per form passed
3283 to EVAL has gone up, and the cost per form executed by EVAL
3284 has gone down), the behavior of the system changes a little
3285 because there are no longer any interpreted function objects.
3286 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3287 e.g. doing COMPILE on the output of interactive DEFUN is
3289 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3291 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3292 has changed from (:CL) to NIL.
3293 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3294 e.g. #\Newline and #\Tab, don't have names specified in the
3295 ANSI Common Lisp standard, is now based on their ASCII symbolic
3296 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3297 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3298 are no longer used for output.
3299 ** changes in internal implementation constants:
3300 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3301 4 million. (If your application spends a lot of time GCing and
3302 you have a lot of RAM, you might want to experiment with
3303 increasing it even more.)
3304 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3305 However, almost all old code should still continue to work without
3306 immediate update, as SB-C-CALL is now a (deprecated) nickname
3308 ** Old operator names in the style DEF-FOO are now deprecated in
3309 favor of new corresponding names DEFINE-FOO, for consistency with
3310 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3311 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3312 internal symbols, but a few supported extensions like
3313 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3314 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3315 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3316 etc. as you get deeper into recursive calls to the debugger
3317 command loop, instead of the old "5]", "5]]", "5]]]"
3318 sequence. (I was motivated to do this when squabbles between
3319 ILISP and SBCL left me very deeply nested in the debugger. In the
3320 short term, this change will probably provoke more ILISP/SBCL
3321 squabbles, but hopefully it will be an improvement in the long run.)
3322 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3323 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3324 function, since declaiming return types involving aliens is
3325 (1) annoyingly messy to do by hand and (2) vital to efficient
3326 compilation of code which calls such functions.
3327 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3328 longer reexported by the SB-EXT package. They're solely useful
3329 for alien code, so it seems more logical that you should get
3330 them from the SB-ALIEN package, not in SB-EXT.
3331 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3332 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3333 features. Instead, the code that they used to control is always
3334 built into the system.
3335 * many other bug fixes
3336 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3337 advantage of the new EVAL-WHEN stuff and to clean them up in
3338 general, and they are now more ANSI-compliant in a number of
3339 ways. Martin Atzmueller is responsible for a lot of this.
3340 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3342 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3343 *** correct ERROR type for various file operations
3344 *** some fixes for Lisp streams
3345 *** DEFMETHOD syntax checking
3346 *** changing old weird representation of debug information as
3347 strings (which, among their other deficiencies, don't transform
3348 correctly when you rename packages, and don't change their
3349 print representation when you change things like *PACKAGE*
3350 and *PRINT-LENGTH*) to symbols and lists of symbols
3351 He also made several improvements and fixed several bugs in DESCRIBE.
3352 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3354 *** misbehavior of WRITE-STRING/WRITE-LINE
3355 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3356 and several other LOOP problems as well
3357 *** DIRECTORY when similar filenames are present
3358 *** DEFGENERIC with :METHOD options
3359 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3360 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3361 *** argument ordering in FIND with :TEST option
3362 *** mishandled package designator argument in APROPOS-LIST
3363 *** various problems in the backquote readmacro
3364 *** a bug in APROPOS
3365 *** probably some others that I'm not describing very well here,
3366 since the CVS log documents them by reference to sbcl-devel
3367 messages, and the SourceForge archives aren't working well.:-(
3368 ** Dan Barlow improved the Alpha port (and is making progress on the
3369 PPC port, for those of you who think different).
3370 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3371 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3372 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3373 continues to work on the SPARC port (for those of you in a position
3374 to look down upon our little PC-compatible boxes from a great height).
3375 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3376 and uses that copy, rather than the current dynamic value, when
3377 it's trying to decide whether to truncate output. Thus e.g.
3378 (let ((*print-lines* 50))
3379 (pprint-logical-block (stream nil)
3381 (let ((*print-lines* 8))
3382 (print (aref possiblybigthings i) stream)))))
3383 should now truncate the logical block only at 50 lines, instead of
3384 often truncating it at 8 lines, as it did before.
3385 * The doc/cmucl/ directory, containing old CMU CL documentation
3386 from the time of the fork, is no longer part of the base system.
3387 SourceForge has shut down its anonymous FTP service, and with it
3388 my original plan for distributing the old CMU CL documentation
3389 there. For now, if you need these files you can download an old
3390 SBCL source release and extract them from it.
3391 * The fasl file version number changed again, for dozens of reasons,
3392 some of which are apparent above.
3394 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3395 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3396 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3397 patch, so that DISASSEMBLE output is much nicer.
3398 * The code in the SB-PROFILE package now seems reasonably stable.
3399 I still haven't decided what the final interface should look like
3400 (I'd like PROFILE to interact cleanly with TRACE, since both
3401 facilities use function encapsulation) but if you have a need
3402 for profiling now, you can probably use it successfully with
3403 the current CMU-CL-style interface.
3404 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3405 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3406 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3407 link now returns the dangling link itself, and for similar
3408 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3409 link itself. (In these cases the old code signalled an error and
3410 looped endlessly, respectively.) Thus, DIRECTORY now works even
3411 in the presence of dangling and cyclic symbolic links.
3412 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3413 is now a supported extension again, since the consensus on
3414 sbcl-devel was that it can be useful for ordinary development
3415 work, not just for debugging SBCL itself.
3416 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3417 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3418 that currently-defined functions might be redefined later with
3419 different return types.
3420 * Hash tables can be printed readably, as inspired by CMU CL code
3421 of Eric Marsden and SBCL code of Martin Atzmueller.
3422 * better error handling in CLOS method combination, thanks to
3423 Martin Atzmueller porting Pierre Mai's CMU CL patches
3424 * more overflow fixes for >16Mbyte I/O buffers
3425 * A bug in READ has been fixed, so that now a single Ctrl-D
3426 character suffices to cause end-of-file on character streams.
3427 In particular, now you only need one Ctrl-D at the command
3428 line (not two) to exit SBCL.
3429 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3431 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3432 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3433 does the right thing.
3434 * The implementation of some type tests, especially for CONDITION
3435 types, is now tidier and maybe faster, due to CMU CL code
3436 originally by Douglas Crosher, ported by Martin Atzmueller.
3437 * Some math functions have been fixed, and there are new
3438 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3439 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3440 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3441 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3442 likely to be built into sbcl-0.7.0.)
3443 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3444 CL, ported to SBCL by Martin Atzmueller)
3445 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3446 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3447 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3448 * There's a new slam.sh hack to shorten the edit/compile/debug
3449 cycle for low-level changes to SBCL itself, and a new
3450 :SB-AFTER-XC-CORE target feature to control the generation of
3451 the after-xc.core file needed by slam.sh.
3452 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3453 COMPILE-FILE is no longer supported, so that now every function
3454 gets an entry point, so that block compilation looks a little
3455 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3456 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3457 returns the number of bytes consed since the system started,
3458 rather than the number consed since the first time the function
3459 was called. (The new definition parallels ANSI functions like
3460 CL:GET-INTERNAL-RUN-TIME.)
3461 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3462 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3463 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3464 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3465 * Fasl file version numbers are now independent of the target CPU,
3466 since historically most system changes which required version
3467 number changes have affected all CPUs equally. Similarly,
3468 the byte fasl file version is now equal to the ordinary
3471 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3472 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3473 is no longer recognized. I apologize for this, because it was
3474 listed in SB-EXT as a supported extension, but I found that
3475 its existing behavior was poorly specified, as well as incorrectly
3476 specified, and it looked like too much of a mess to straighten it
3477 out. I have enough on my hands trying to get ANSI stuff to work..
3478 * many patches ported from CMU CL by Martin Atzmueller, with
3479 half a dozen bug fixes in pretty-printing and the debugger, and
3480 half a dozen others elsewhere
3481 * fixed bug 13: Floating point infinities are now supported again.
3482 They might still be a little bit flaky, but thanks to bug reports
3483 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3484 as flaky as they were.
3485 * The --noprogrammer command line option is now supported. (Its
3486 behavior is slightly different in detail from what the old man
3487 page claimed it would do, but it's still appropriate under the
3488 same circumstances that the man page talks about.)
3489 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3490 are now supported, and enabled by default. Thus, the compiler can
3491 handle many floating point and complex operations much less
3492 inefficiently. (Thus e.g. you can implement a complex FFT
3494 * The compiler now detects type mismatches between DECLAIM FTYPE
3495 and DEFUN better, and implements CHECK-TYPE more correctly, and
3496 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3497 thanks to patches from Martin Atzmueller.
3498 * various fixes to make the cross-compiler more portable to
3499 ANSI-conforming-but-different cross-compilation hosts (notably
3500 Lispworks for Windows, following bug reports from Arthur Lemmens)
3501 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3502 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3503 fixed thanks to Pierre Mai's CMU CL patches.
3504 * fixes to keep the system from overflowing internal counters when
3505 it tries to use i/o buffers larger than 16M bytes
3506 * fixed bug 45a: Various internal functions required to support
3507 complex special functions have been merged from CMU CL sources.
3508 (When I was first setting up SBCL, I misunderstood a compile-time
3509 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3510 * improved support for type intersection and union, fixing bug 12
3511 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3512 more obscure bugs as well
3513 * some steps toward byte-compiling non-performance-critical
3514 parts of the system, courtesy of patches from Martin Atzmueller
3515 * Christophe Rhodes has made some debian packages of sbcl at
3516 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3517 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3518 stable, but are nonetheless usable. When he's ready, I'd be happy
3519 to add them to the SourceForge "File Releases" section. (And if
3520 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3521 * new fasl file format version number (because of changes in
3522 internal representation of (OR ..) types to accommodate the new
3523 support for (AND ..) types, among other things)
3525 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3526 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3528 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3530 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3531 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3532 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3533 * DESCRIBE now works on structure objects again.
3534 * Most function call argument type mismatches are now handled as
3535 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3536 know whether the function will be redefined before the call is
3537 executed. (The compiler could flag local calls with full WARNINGs,
3538 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3539 it doesn't keep track of enough information to know whether calls
3540 are local in this sense.)
3541 * Compiler output is now more verbose, with messages truncated
3542 later than before. (There should be some supported way for users
3543 to override the default verbosity, but I haven't decided how to
3544 provide it yet, so this behavior is still controlled by the internal
3545 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3546 src/compiler/ir1util.lisp.)
3547 * Fasl file format version numbers have increased again, because
3548 support for the Gray streams extension changes the layout of the
3549 system's STREAM objects.
3550 * The Gray subclassable streams extension now works, thanks to a
3551 patch from Martin Atzmueller.
3552 * The full LOAD-FOREIGN extension (not just the primitive
3553 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3554 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3555 but like most other programs, it defaults to copying the Unix
3556 environment from the original process instead of starting the
3557 new process in an empty environment.
3558 * Extensions which manipulate the Unix environment now support
3559 an :ENVIRONMENT keyword option which doesn't smash case or
3560 do other bad things. The CMU-CL-style :ENV option is retained
3561 for porting convenience.
3562 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3563 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3565 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3567 * A patch from Martin Atzmueller seems to have solved the SIGINT
3568 problem, and as far as we know, signal-handling now works cleanly.
3569 (If you find any new bugs, please report them!)
3570 * The system no longer defaults Lisp source file names to types
3571 ".l", ".cl", or ".lsp", but only to ".lisp".
3572 * The compiler no longer uses special default file extensions for
3573 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3574 seems to expect a single default extension for all compiled code,
3575 and there's no compelling reason to try to stretch the standard
3576 to allow two different extensions.) Instead, byte-compiled files
3577 default to the same extension as native-compiled files.
3578 * Fasl file format version numbers have increased again, because
3579 a rearrangement of internal implementation packages made some
3580 dumped symbols in old fasl files unreadable in new cores.
3581 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3582 many fewer weird special cases.
3583 * Bug #17 (differing COMPILE-FILE behavior between logical and
3584 physical pathnames) has been fixed, and some related misbehavior too,
3585 thanks to a patch from Martin Atzmueller.
3586 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3587 by Tim Moore, ported to SBCL by Martin Atzmueller.
3588 * Martin Atzmueller fixed several filesystem-related problems,
3589 including bug #36, in part by porting CMU CL patches, which were
3590 written in part by Paul Werkowski.
3591 * More compiler warnings in src/runtime/ are gone, thanks to
3592 more patches from Martin Atzmueller.
3593 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3596 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3598 * DESCRIBE now works on CONDITION objects.
3599 * The debugger now handles errors which arise when trying to print
3600 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3602 * The build system now uses an additional file, customize-target-features.lisp,
3603 to allow local modifications to the target *FEATURES* list. (The point of
3604 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3605 debugging features enabled, without having to worry about propagating it
3606 into everyone's system when I do a "cvs update".) When no
3607 customize-target-features.lisp file exists, the target *FEATURES* list
3608 should be constructed the same way as before.
3609 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3610 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3611 warn about behavior which is undefined under the ANSI spec. Note: This
3612 is specified by ANSI, but it's not very popular with programmers.
3613 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3614 macro in the SBCL sources for an example of a workaround which you
3615 might use to make portable ANSI-standard code which does what you want.
3616 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3617 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3618 when it's used as a non-toplevel form. (This is required in order
3619 to implement the DEFCONSTANT-EQX macro.)
3620 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3621 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3622 funny things, due to bugs in the implementation of EVAL-WHEN
3623 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3624 non-toplevel DEFCONSTANTs are uncommon.)
3625 * The core file version number and fasl file version number have been
3626 incremented, because the old noncompliant DEFCONSTANT behavior involved
3627 calling functions which no longer exist, and because I also took the
3628 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3629 * fixed bug 1 (error handling before read-eval-print loop starts), and
3630 redid debugger restarts and related debugger commands somewhat while
3632 ** The QUIT debugger command is gone, since it did something
3633 rather different than the SB-EXT:QUIT command, and since it never
3634 worked properly outside the main toplevel read/eval/print loop.
3635 Invoking the new TOPLEVEL restart provides the same functionality.
3636 ** The GO debugger command is also gone, since you can just invoke
3637 the CONTINUE restart directly instead.
3638 ** The TOP debugger command is also gone, since it's redundant with the
3639 FRAME 0 command, and since it interfered with abbreviations for the
3641 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3643 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3644 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3645 some porting work by Martin Atzmueller)
3646 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3647 with a boatload of patches which he ported to SBCL. Now that those
3649 ** The system tries to make sure that its low-priority messages
3650 are prefixed by semicolons, to help people who like to use
3651 syntax highlighting in their ILISP buffer. (This patch
3652 was originally due to Raymond Toy.)
3653 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3654 patches originally written by Raymond Toy.
3655 ** The compiler understands coercion between single-value and
3656 multiple-VALUES type expressions better, getting rid of some very
3657 weird behavior, thanks to patches originally by Robert MacLachlan
3658 and Douglas Crosher.
3659 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3660 lambda lists, thanks to a patch originally by Pierre Mai.
3661 ** The system no longer bogusly warns about "abbreviated type
3663 ** The compiler gets less confused by inlining and RETURN-FROM,
3664 thanks to some patches originally by Tim Moore.
3665 ** The system no longer hangs when dumping circular lists to fasl
3666 files, thanks to a patch originally from Douglas Crosher.
3667 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3668 undefined function error.
3669 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3670 in structure slot initforms) for now, documented workaround instead:-|
3671 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3672 * fixed bug 5: added stubs for various Gray stream functions called
3673 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3674 installed, at least appropriate type errors are generated
3675 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3676 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3677 consistently in DEFMETHOD forms.
3678 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3679 it doesn't seem to affect SBCL after all
3680 * The C runtime system now builds with better optimization and many
3681 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3683 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3685 * The system is now under CVS at SourceForge (instead of the
3686 CVS repository on my home machine).
3687 * The new signal handling code has been tweaked to treat register
3688 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3689 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3690 downstream, e.g. in debug-int.lisp, has implicit dependencies
3691 on the unsignedness of integer representation of machine words,
3692 and that caused the system to bomb out with infinite regress
3693 when trying to recover from type errors involving signed values,
3694 e.g. (BUTLAST '(1 2 3) -1).)
3695 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3696 (This was one of the bugs Peter Van Eynde reported back in July.)
3697 * The system now uses code inspired by Colin Walters' O(N)
3698 implementation of MAP (from the cmucl-imp@cons.org mailing
3699 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3700 inline the MAP operation, and there is more than one
3701 sequence argument to the MAP call (so that it can't just
3702 do ETYPECASE once and for all based on the type of the
3703 single sequence argument). (The old non-inline implementation
3704 of the general M-argument sequence-of-length-N case required
3705 O(M*N*N) time when any of the sequence arguments were LISTs.)
3706 * The QUIT :UNIX-CODE keyword argument has been renamed to
3707 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3708 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3710 * Raymond Wiker's patches to port dynamic loading from Linux to
3711 FreeBSD have been added.
3712 * The BUGS file is now more nearly up to date, thanks in large part
3713 to Martin Atzmueller's review of it.
3714 * The debugger now flushes standard output streams before it begins
3715 its output ("debugger invoked" and so forth).
3716 * The core version number and fasl file version number have both
3717 been incremented, because of incompatible changes in the layout
3719 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3720 used to not be called for a saved Lisp image.)
3721 * Martin Atzmueller's version of a patch to fix a compiler crash,
3722 as posted on sbcl-devel 13 September 2000, has been installed.
3723 * Instead of installing Martin Atzmueller's patch for the
3724 compiler transform for SUBSEQ, I deleted the compiler transform,
3725 and transforms for some similar consing operations.
3726 * A bug in signal handling which kept TRACE from working on OpenBSD
3728 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3729 away by constant folding
3730 * The system now defines its address space constants in one place
3731 (in the Lisp sources), and propagates them automatically elsewhere
3732 (through GENESIS and the sbcl.h file). Therefore, patching the
3733 address map is less unnecessarily tedious and error-prone. The
3734 Lisp names of address space constants have also been systematized.
3735 * CVS tags like dollar-Header-dollar have been removed from
3736 the sources, because they have never saved me trouble and
3737 they've been source of trouble working with patches and other
3738 diff-related operations.
3739 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3740 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3742 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3744 * The system has been ported to OpenBSD.
3745 * The system now compiles with a simple "sh make.sh" on the systems
3746 that it's supported on. I.e., now you no longer need to tweak
3747 text in the target-features.lisp-expr and symlinks in src/runtime/
3748 by hand, the make.sh takes care of it for you.
3749 * The system is no longer so grossly inefficient when compiling code
3750 involving vectors implemented as general (not simple) vectors (VECTOR T),
3751 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3752 which dares to use the various sequence functions on non-simple
3753 vectors, takes less of a performance hit.
3754 * There is now a primitive type predicate VECTOR-T-P
3755 to test for the (VECTOR T) type, so that e.g.
3756 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3757 can now be compiled with some semblance of efficiency. (The old code
3758 turned the type declaration into a full call to %TYPEP at runtime!)
3759 * AREF on (VECTOR T) is still not fast, since it's still compiled
3760 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3761 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3762 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3763 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3764 without making full calls to SUBTYPEP at runtime.
3765 (Some analogous efficiency issues for non-simple vectors specialized to
3766 element types other than T, or for non-simple multidimensional arrays,
3767 have not been addressed. They could almost certainly be handled the
3768 same way if anyone is motivated to do so.)
3769 * The changes in array handling break binary compatibility, so
3770 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3771 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3772 instead of (VALUES T T).
3773 * By following the instructions that Dan Barlow posted to sbcl-devel
3774 on 2 July 2000, I was able to enable primitive dynamic object
3775 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3776 functionality is not implemented (since it calls ld to resolve
3777 library references automatically, requiring RUN-PROGRAM for its
3778 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3779 try to resolve library references) is now supported.
3780 * The system now flushes the standard output streams when it terminates,
3781 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3782 them at several other probably-convenient times, e.g. in each pass of
3783 the toplevel read-eval-print loop, and after evaluating a form given
3784 as an "--eval" command-line option. (These changes were motivated by a
3785 discussion of stream flushing issues on cmucl-imp in August 2000.)
3786 * The source transform for TYPEP of array types no longer assumes
3787 that an array whose element type is a not-yet-defined type
3788 is implemented as an array of T, but instead punts, so that the
3789 type will be interpreted at runtime.
3790 * There is now some support for cross-compiling in make.sh: each of
3791 the phases of make.sh has its own script. (This should be transparent
3792 to people doing ordinary, non-cross-compile builds.)
3793 * Since my laptop doesn't have hundreds of megabytes of memory like
3794 my desktop machine, I became more motivated to do some items on
3795 my to-do list in order to reduce the size of the system a little:
3796 ** Arrange for various needed-only-at-cold-init things to be
3797 uninterned after cold init. To support this, those things have
3798 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3799 symbols with such names are now uninterned after cold init).
3800 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3801 fasl files for cold load.
3802 ** Remove the old compiler structure pooling code (which used to
3803 be conditional on the target feature :SB-ALLOC) completely.
3804 ** Redo the representation of some data in cold init to be more compact.
3805 (I also looked into supporting byte compiled code at bootstrap time,
3806 which would probably reduce the size of the system a lot, but that
3807 looked too complicated, so I punted for now.)
3808 * The maximum signal nesting depth in the src/runtime/ support code has
3809 been reduced from 4096 to 256. (I don't know any reason for the very
3810 large old value. If the new smaller value turns out to break something,
3811 I'll probably just bump it back up.)
3812 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3814 * Many, many bugs reported by Peter Van Eynde have been added to
3815 the BUGS list; some have even been fixed.
3816 * While enabling dynamic object file loading, I tried to make the
3817 code easier to understand, renaming various functions and variables
3818 with less ambiguous names, and changing some function calling
3819 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3820 * While trying to figure out how to do the OpenBSD port, I tried to
3821 clean up some of the code in src/runtime/. In particular, I dropped
3822 support for non-POSIX signal handling, added various comments,
3823 tweaked the code to reduce the number of compilation warnings, and
3824 renamed some files to increase consistency.
3825 * To support the new automatic configuration functionality in make.sh,
3826 the source file target-features.lisp-expr has been replaced with the
3827 source file base-target-features.lisp-expr and the machine-generated
3828 file local-target-features.lisp-expr.
3829 * fixed a stupid quoting error in make.sh so that using CMU CL
3830 "lisp -batch" as cross-compilation host works again
3832 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3834 * DESCRIBE no longer tries to call itself recursively to describe
3835 bound/fbound values, so that it no longer fails on symbols which are
3836 bound to themselves (like keywords, T, and NIL).
3837 * DESCRIBE now works on generic functions.
3838 * The printer now prints less-screwed-up representations of closures
3839 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3840 * A private symbol is used instead of the :EMPTY keyword previously
3841 used to mark empty slots in hash tables. Thus
3842 (DEFVAR *HT* (MAKE-HASH-TABLE))
3843 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3844 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3845 now does what ANSI says that it should. (You can still get
3846 similar noncompliant behavior if bang on the hash table
3847 implementation with all the symbols you get back from
3848 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3849 This breaks binary compatibility, since tests for equality to
3850 :EMPTY are wired into things like the macroexpansion of
3851 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3853 * There's now a minimal placeholder implementation for CL:STEP,
3854 as required by ANSI.
3855 * An obscure bug in the interaction of the normal compiler, the byte
3856 compiler, inlining, and structure predicates has been patched
3857 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3858 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3860 * Missing ordinary arguments in a macro call are now detected even
3861 when the macro lambda list contains &KEY or &REST.
3862 * The debugger no longer complains about encountering the top of the
3863 stack when you type "FRAME 0" to explicitly instruct it to go to
3864 the top of the stack. And it now prints the frame you request even
3865 if it's the current frame (instead of saying "You are here.").
3866 * As specified by ANSI, the system now always prints keywords
3867 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3869 * The default initial SIZE of HASH-TABLEs is now smaller.
3870 * Type information from CLOS class dispatch is now propagated
3871 into DEFMETHOD bodies, so that e.g.
3872 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3874 is now basically equivalent to
3875 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3876 (DECLARE (TYPE SINGLE-FLOAT X))
3878 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3879 operation, without having to do run-time type dispatch.
3880 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3881 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3882 * Since I don't seem to be making big file reorganizations very often
3883 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3884 is overflowing my ability to conveniently back them up), I've finally
3885 checked the system into CVS. (The CVS repository is on my home system,
3886 not at SourceForge -- putting it on SourceForge might come later.)
3887 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3888 high-level GC-NOTIFY-FOO functions send their output. (There's
3889 still very little control of where low-level verbose GC functions
3890 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3891 less than it used to -- the GC-NOTIFY-FOO functions are now under
3892 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3893 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3894 in only one place in the source code, and propagates it automatically
3895 everywhere that it's needed. Thus e.g. when I bump the version from
3896 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3897 * The C source files now include boilerplate legalese and documentation
3898 at the head of each file (just as the Lisp source files already did).
3899 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3900 to his page will be replaced with a link to his new CLiki service.
3902 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3904 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3905 * The build process now looks for GNU make under the default name "gmake",
3906 instead of "make" as it used to. If GNU make is not available as "gmake"
3907 on your system, you can change this default behavior by setting the
3908 GNUMAKE environment variable.
3909 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3910 can build without error under CMU CL.
3912 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3914 * There is now a partial SBCL user manual (with some new text and some
3915 text cribbed from the CMU CL manual).
3916 * The beginnings of a profiler have been added (starting with the
3917 CMU CL profiler and simplifying and cleaning up). Eventually the
3918 main interface should be through the TRACE macro, but for now,
3919 it's still accessed through vaguely CMU-CL-style functions and macros
3920 exported from the package SB-PROFILE.
3921 * Some problems left over from porting CMU CL to the new
3922 cross-compilation bootstrap process have been cleaned up:
3923 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3924 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3925 part when getting rid of the compiler *BACKEND* stuff.)
3926 ** The value of *NULL-TYPE* was screwed up, because it was
3927 being initialized before the type system knew the final
3928 definition of the 'NULL type. This screwed up several key
3929 optimizations in the compiler, causing inefficiency in all sorts
3930 of places. (I found it because I wanted to understand why
3931 GET-INTERNAL-RUN-TIME was consing.)
3932 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3933 PROCLAIM FTYPE information. Unfortunately this broke binary
3934 compatibility again, since now the forms output by DEFGENERIC
3935 to refer to functions which didn't exist in 0.6.3.
3936 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3937 can use the new (as of 0.6.3) transform for SOME into MAP into
3939 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3940 values to (INTEGER 0 1000000), so that the time code will no longer
3941 occasionally get blown up by Linux returning 1000000 microseconds
3942 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3943 its output conform to the ANSI spec. (Alas, this makes its output
3944 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3945 * A full call to MAP NIL with a single sequence argument no longer conses.
3946 * fixes to problems pointed out by Martin Atzmueller:
3947 * The manual page no longer talks about multiprocessing as though
3948 it were currently supported.
3949 * The ILISP support patches have been removed from the distribution,
3950 because as of version 5.10.1, ILISP now supports SBCL without us
3951 having to maintain patches.
3952 * added a modified version of Raymond Toy's recent CMU CL patch for
3953 EQUALP comparison of HASH-TABLE
3955 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3957 * The system still can't cross-compile itself with
3958 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3959 entails), but at least it can compile more of itself that way
3960 than it used to be able to, and various buglets which were uncovered
3961 by trying to cross-compile itself that way have now been fixed.
3962 * This release breaks binary compatibility again. This time
3963 at least I've incremented the FASL file format version to 2, so that the
3964 problem can be detected reliably instead of just causing weird errors.
3965 * various new style warnings:
3966 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3967 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3968 ** using the lexical binding of a variable named in the *FOO* style
3969 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3970 as specified by ANSI.
3971 * *RANDOM-STATE* is no longer automatically initialized from
3972 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3973 default behavior of the system is to repeat its behavior every time
3974 it's run. If you'd like to change this behavior, you can always
3975 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3976 old convention there was no comparably easy way to get the system to
3977 repeat its behavior every time it was run.
3978 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3979 FTYPE declarations has been removed, in favor of their ANSI
3980 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3981 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3982 the types of their sequence arguments can be determined at compile time.
3983 This is done through a new open code expansion for MAP which eliminates
3984 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3985 argument types can be determined at compile time.
3986 * The optimizer now transforms COERCE into an identity operation when it
3987 can prove that the coerced object is already of the correct type. (This
3988 can be a win for machine generated code, including the output of other
3989 optimization transforms, such as the MAP transform above.)
3990 * Credit information has been moved from source file headers into CREDITS.
3991 * Source file headers have been made more standard.
3992 * The CASE macro now compiles without complaining even when it has
3995 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3997 * (Note that the way that the PCL macroexpansions were rewritten
3998 to accommodate the change in DEFGENERIC below breaks binary
3999 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4000 not run under sbcl-0.6.2. Once we get out of alpha releases,
4001 i.e. hit release 1.0.0, we'll probably try to maintain binary
4002 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4003 and sbcl-1.4.4. Until then, however, it might be fairly common
4004 for maintenance releases to break binary compatibility.)
4005 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4007 * The reporting of unhandled signals has been changed to print some
4008 explanatory text as well as the report form. (Previously only
4009 the report form was printed.)
4010 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4011 it defines, so that the compiler no longer issues undefined function
4012 warnings for compiled-but-not-yet-loaded generic functions.
4013 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4014 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4015 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4016 section of the ANSI spec are supported.
4017 * The "" nickname for the "KEYWORD" package has been removed.
4018 The reader still handles symbol tokens which begin with a package marker
4019 as keywords, but it doesn't expose its mechanism for doing so in the
4020 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4021 * The system now issues STYLE-WARNINGs for contradictory TYPE
4022 proclamations. (Warnings for contradictory FTYPE proclamations would
4023 be nice too, but those can't be done usefully unless the type system
4024 is made smarter about FUNCTION types.)
4025 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4026 systematized, so that "*target-*.lisp is supposed to exist only on the
4027 target and imply that there's a related file which exists on the
4028 host, and *host-*.lisp is supposed to exist only on the host and imply
4029 that there's a related file which exists on the target. This involves a
4030 lot of renaming. Hopefully the acute confusion caused by the renaming
4031 will be justified by the reduction in chronic confusion..
4032 ** runtime-type.lisp -> early-target-type.lisp
4033 ** target-type.lisp -> late-target-type.lisp
4034 ** early-host-format.lisp -> early-format.lisp
4035 ** late-host-format.lisp -> late-format.lisp
4036 ** host-error.lisp -> misc-error.lisp
4037 ** early-error.lisp -> early-target-error.lisp
4038 ** late-error.lisp -> late-target-error.lisp
4039 ** host-defboot.lisp -> early-defboot.lisp
4040 ** code/misc.lisp -> code/target-misc.lisp
4041 ** code/host-misc.lisp -> code/misc.lisp
4042 ** code/numbers.lisp -> code/target-numbers.lisp
4043 ** code/early-numbers.lisp -> numbers.lisp
4044 ** early-host-type.lisp -> early-type.lisp
4045 ** late-host-type.lisp -> late-type.lisp
4046 ** host-typep.lisp -> typep.lisp
4047 ** load.lisp -> target-load.lisp
4048 ** host-load.lisp -> load.lisp
4049 ** host-disassem.lisp -> disassem.lisp
4050 ** host-insts.lisp -> insts.lisp
4051 ** byte-comp.lisp -> target-byte-comp.lisp
4052 ** host-byte-comp.lisp -> byte-comp.lisp
4053 ** host-signal.lisp -> signal.lisp
4054 ** host-defstruct.lisp -> defstruct.lisp
4055 ** late-target-type.lisp -> deftypes-for-target.lisp
4056 Furthermore, several other previously target-only files foo.lisp (e.g.
4057 hash-table.lisp and random.lisp) have been split into a target-and-host
4058 foo.lisp file and a target-only target-foo.lisp file, with their key type
4059 definitions in the target-and-host part, so that the cross-compiler will
4060 know more about target types.
4061 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4062 gone away. In their place are various *BACKEND-FOO* variables
4063 corresponding to the slots of the old structure.
4064 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4065 into the target SBCL has been fixed.
4066 * The chill.lisp system for loading cold code into a running SBCL
4068 * Support for the CMU CL "scavenger hook" extension has been removed.
4069 (It was undocumented and unused in the CMU CL sources that SBCL was
4070 derived from, and stale in sbcl-0.6.1.)
4071 * Various errors in the cross-compiler type system were detected
4072 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4073 (enabling various consistency checks). Many of them were fixed,
4074 but some hard problems remain, so the compiler is back to
4075 running without *TYPE-SYSTEM-INITIALIZED* for now.
4076 * As part of the cross-compiler type system cleanup, I implemented
4077 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4078 * I have started adding UNCROSS calls throughout the type system
4079 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4080 on cross-compiler input files will be able to go away, and various
4082 * CONSTANTP now returns true for quoted forms (as explicitly required
4085 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4087 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4088 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4089 (MAKE-LIST -1). In the longer term, it ought to become true
4090 that declarations are assertions even at SAFETY 1. For now, it's not
4091 quite true even at SAFETY 3, but it's at least more nearly true..
4092 (Note that this change seems to increases the size of the system by
4093 O(5%) and to decrease the speed of the compiler by 20% or more.)
4094 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4095 for the problem of printing dozens of lines of distracting information
4096 about low-level system machinery as part of the top stack frame
4097 on entry to the debugger when an undefined function was called.
4098 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4099 is not reset to COMMON-LISP-USER.
4100 * Compilation of stuff related to dyncount.lisp has been made conditional
4101 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4102 smaller. The various dyncount-related symbols have been moved into
4103 a new "SB-DYNCOUNT" package.
4104 * tty-inspect.lisp has been renamed to inspect.lisp.
4105 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4106 feature has gone away. (When we eventually port to other flavors of
4107 libc and/or Unix, we'll try to make the differences between flavors
4108 invisible at the user level.)
4109 * Various other *FEATURES* tags, and/or their associated conditionals,
4110 have been removed if obsolescent, or given better documentation, or
4111 sometimes given more-mnemonic names.
4113 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4115 * tidied up "make.sh" script
4116 * tidied up system directory structure
4117 * better "clean.sh" behavior
4118 * added doc/FOR-CMUCL-DEVELOPERS
4119 * many many small tweaks to output format, e.g. removing possibly-confusing
4120 trailing #\. character in DESCRIBE-INSTANCE
4121 * (EQUALP #\A 'A) no longer signals an error.
4122 * new hashing code, including EQUALP hashing
4123 * tidied up Lisp initialization and toplevel
4124 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4125 * command line argument processing
4126 * added POSIX-GETENV function to deal with Unix-ish environment variables
4127 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4128 terminating SBCL on EOF
4129 * non-verbose GC by default
4130 * There is no more "sbcl" shell script; the sbcl file is now the C
4131 runtime executable (just like CMU CL).
4132 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4134 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4135 symbol and package information as Lisp native symbol and package objects
4136 instead of strings naming symbols and strings naming packages. This way,
4137 whenever packages are renamed (as in warm init), debug information is
4138 transformed along with everything else.
4139 * tweaked the optimization policy declarations which control the building
4140 of SBCL itself. Now, among other things, the system no longer saves
4141 source location debugging information. (This helps two problems at once
4142 by reducing SBCL size and by keeping SBCL from trying to look for its
4143 sources -- which may not exist -- when reporting errors.)
4144 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4145 debugging and testing purposes
4146 * cleaned up printing, making the printer call PRINT-OBJECT for
4147 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4148 methods, giving nearly-ANSI behavior
4149 * converted almost all special variables to use *FOO* naming convention
4150 * deleted PARSE-TIME functionality, since it can be done portably
4151 * moved some files out of cold init into warm init
4152 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4154 * regularized formatting of source files
4155 * added an install.sh script
4156 * fixed ridiculous memory usage of cross-compiler by making
4157 compiler/alloc.lisp not try to do pooling unless it can hook
4158 itself into the GC of the cross-compilation host. Now the system
4159 builds nicely on my old laptop.
4160 * added :SB-ALLOC in target-features.lisp-expr
4161 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4162 was not implemented)
4163 * re-did condition handling and note reporting in the compiler. Notes
4164 are no longer handled by signalling conditions. Style warnings
4165 and warnings are handled more correctly and reported in such a way
4166 that it's easy to find one or the other in your output (so that you
4167 can e.g. figure out which of many problems caused COMPILE-FILE to
4169 * changed the severity of several compiler warnings from full WARNING
4170 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4171 compiler note reporting so that it doesn't use the condition system
4172 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4173 COMPILE-FILE command)
4174 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4175 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4176 no longer supported, since their implementation was deeply intertwingled
4177 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4178 * removed bogus "support" for compiler macros named (SETF FOO), and
4179 removed the compiler macro for SETF INFO (but only after making a fool
4180 of myself on the cmucl-imp mailing list by posting a bogus patch for
4181 DEFINE-COMPILER-MACRO..)
4182 * Compiled files containing forms which have side effects on the Lisp
4183 reader (such as DEFPACKAGE forms) are now handled more correctly.
4184 (Compiler queuing of top level lambdas has been suppressed by setting
4185 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4186 * deleted various currently-unused source files, e.g. gengc.lisp. They
4187 may be added back at some point e.g. when porting to other architectures,
4188 but until they are it's distracting to distribute them and to try to
4190 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4191 there were so many of them they're just distractions, and UNCROSS is
4192 known to be able to handle the current sources
4193 * moved PROFILE functionality into TRACE, so that it will be clear
4194 how the wrapping and unwrapping of functions when you profile them
4195 interacts with the wrapping and unwrapping of functions when you
4196 trace them. (Actually, the functionality isn't there yet, but at least
4197 the interface specification is there. Hopefully, the functionality will
4198 arrive with some future maintenance release.)
4199 * removed host-oops.lisp
4200 * changed signature of QUIT function to allow UNIX-CODE argument
4201 * fixed READ-SEQUENCE bug
4202 * tweaked verbose GC output so that it looks more like the progress
4203 output that ANSI specifies for functions like LOAD
4204 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4205 * added <http://sbcl.sourceforge.com> to the banner information printed by