1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
3 * minor incompatible change: changed experimental JOIN-THREAD interface
4 * documentation: unwinding from asyncronous events has been
6 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
7 in multithreaded application code.
8 * bug fix: GC deadlocks from asynchronous interrupts has been fixed
9 by disabling interrupts for the duration of any
10 SB-SYS:WITHOUT-GCING section.
12 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
13 * new platform: experimental support for x86-64/darwin (MacOS).
14 * incompatible change: the thread-safe (on most platforms) getaddrinfo
15 and getnameinfo sockets functions are used instead of gethostbyaddr
16 and gethostbyname, on platforms where the newer functions are available.
17 As a result, the ALIASES field of HOST-ENT will always be NIL on these
19 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
20 don't print the startup banner, but behave as if --noinform was passed
21 as a command line argument. (thanks to Kevin Reid)
22 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
23 * optimization: code using alien values with undeclared types is much faster.
24 * optimization: the compiler is now able to open code SEARCH in more cases.
25 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
26 * bug fix: using standardized COMMON-LISP special variables as loop
27 variables no longer signals bogus package lock violations. (reported
29 * bug fix: declaring local loop variables to be of a range-limited type
30 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
31 (reported by Andras Simon)
32 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
33 bugs remain on x86-64.)
34 * bug fix: the #= and ## reader macros now interact reasonably with
35 funcallable instances.
36 * bug fix: type-checks for function arguments were compiled using the
37 compiler policy settings of the wrong lexical environment (in 1.0.2
39 * bug fix: SHADOW accepts characters as string designators, as required
40 by the spec (thanks to Eric Marsden)
41 * bug fix: fixed GC safety issues when foreign functions are called with
42 non-base strings as arguments
43 * bug fix: more consistent error output for fatal-compiler-warnings, like
45 * bug fix: functions with proper names were showing up as NIL in
46 backtraces (reported by Edi Weitz, regression in 1.0.2)
48 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
49 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
50 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
51 produce more efficient code on the x86 if the compiler can determine
52 that (+ INDEX <constant>) does not require a bounds check and FOO
53 has an element type at least 8 bits wide.
54 * bug fix: references to undefined variables are handled the same way
55 in toplevel forms as in the normal compiler
56 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
58 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
59 (thanks to Jon Buller)
60 * improvement: faster compilation times for complex functions
61 * improvement: added readlink support to SB-POSIX (thanks to Richard
64 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
65 * improvement: experimental support for mach exception handling on
66 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
68 * improvement: support for GBK external format.
69 (thanks to Chun Tian (binghe))
70 * improvement: the debugger now displays variables that have been closed
71 over, in code compiled with (DEBUG 2) or higher
72 * improvement: support for executable cores on NetBSD (thanks to
74 * new feature: added a RESTART-FRAME debugger command
75 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
76 be used to provide an implementation for FILE-POSITION on Gray streams
77 (thanks to Eric Marsden)
78 * new feature: add syslog, openlog and closelog support to SB-POSIX
79 (thanks to Richard Kreuter)
80 * optimization: the function call overhead in code compiled with
81 a high DEBUG optimization setting is significantly
82 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
83 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
84 (thanks to Tony Martinez)
85 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
86 evaluated once (reported by Kevin Reid)
87 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
88 works again (thanks to Kilian Sprotte)
89 * bug fix: an error is signaled for tagbodies with duplicate tags
90 (thanks to Stephen Wilson)
91 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
92 * bug fix: Win32 port can now handle foreign code unwinding Lisp
93 stack frames from alien callbacks.
94 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
95 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
96 65k pages (thanks to David Woodhouse)
97 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
99 changes in sbcl-1.0.1 relative to sbcl-1.0:
100 * new platform: FreeBSD/x86-64, including support for threading.
101 * new feature: the compiler stores cross-referencing information
102 abount function calls (who-calls), macroexpansion (who-macroexpands)
103 and special variables (who-binds, who-sets, who-references) for code
104 compiled with (< SPACE 3). This information is available through the
105 sb-introspect contrib.
106 * new feature: users may subclass SEQUENCE, and have instances of
107 these classes interoperate with standard Common Lisp functions if
108 a number of methods are defined. (This feature is experimental
109 and the interface subject to change based on feedback from SBCL
110 users and the general community)
111 * improvement: sb-sprof traces call stacks to an arbitrary depth on
112 x86/x86-64, rather than the previous fixed depth of 8
113 * improvement: another pthread back-end of mutex "pthread-futex".
114 * bug fix: non-ascii command-line arguments are processed correctly
115 (thanks to Yaroslav Kavenchuk)
116 * bug fix: non-required arguments were not passed correctly when a method
117 defined using DEFMETHOD was called from a mop-generated method using
118 CALL-NEXT-METHOD (reported by Pascal Costanza)
119 * bug fix: recursion is now permitted in accessors through
120 SLOT-UNBOUND. (reported by Pascal Costanza)
121 * bug fix: an error was signaled at startup if the HOME environment
122 variable was defined, but had an empty value (reported by Peter Van Eynde)
123 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
124 signal a PROGRAM-ERROR, not a BUG.
125 * bug fix: SB-POSIX:READDIR works when built with large file support.
126 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
128 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
129 for better automatic indentation support. (thanks to Matt Pillsbury)
130 * optimization: loading generic functions no longer takes O(n^2) time,
131 proportional to the amount of methods in the generic function
132 (reported by Todd Sabin and Jeremy Brown)
133 * optimization: the FIND and POSITION family of sequence functions
134 are significantly faster on arrays whose element types have been
136 * improvements to the Windows port:
137 ** Intermittent heap corruption problems have been fixed. (thanks
138 to Alastair Bridgewater)
139 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
141 ** Lisp is able to unwind foreign exception frames from alien
142 callbacks. (thanks to Alastair Bridgewater)
144 changes in sbcl-1.0 relative to sbcl-0.9.18:
145 * improvement: experimental support for threading on FreeBSD/x86.
146 (thanks to NIIMI Satoshi)
147 * improvement: runtime option --dynamic-space-size can be used
148 to set the size of the dynamic space reserved on startup.
149 * improvement: floating point modes in effect are now saved in
150 core, and restored on startup.
151 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
152 startup, not time since first call to GET-INTERNAL-REAL-TIME.
153 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
154 threads are not running after *SAVE-HOOKS* have run.
155 * improvement: writes to CLOS instance slots are type-checked in code
156 compiled with (SAFETY 3)
157 * improvement: floating-point exception handling on FreeBSD (thanks to
159 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
160 (thanks to Zach Beane)
161 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
163 * improvement: added support for the Shift-JIS external format.
164 (contributed by NIIMI Satoshi)
165 * improvement: callbacks are supported on Linux/PPC. (thanks to
167 * bug fix: compiler bug triggered by a (non-standard) VALUES
168 declaration in a LET* was fixed. (reported by Kaersten Poeck)
169 * bug fix: file compiler no longer confuses validated and already
170 dumped structurres. (reported by Kaersten Poeck)
171 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
172 fill-pointer signals a type-error as required. (thanks to
174 * bug fix: disassemly of funcallable instances works.
175 * bug fix: single stepping on PPC.
176 * bug fix: fix thread-safety problems in the type system (generally
177 manifesting as nonsensical errors like "STRING is a bad type specifier
178 for sequences" or "The value 1 is not of type FIXNUM")
179 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
180 (reported by Josip Gracin)
181 * bug fix: an error is signaled for attempts to displace arrays with
182 incompatible element types (thanks to Mario Mommer)
183 * bug fix: more correct handling of wide characters in debug info
184 (bug reported by Attila Lendvai and fixed by Juho Snellman)
185 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
186 and don't cause extra consing
187 * optimization: MAP and MAP-INTO are significantly faster on vectors
188 whose elements types have been declared.
189 * Improvements to SB-SPROF:
190 ** Support for allocation profiling
191 ** Reduced profiling overhead, especially for long profiling runs
192 * Improvements to the Windows port:
193 ** floating point exceptions are now reported correctly.
194 ** stack exhaustion detection works partially.
195 ** more accurate GET-INTERNAL-REAL-TIME.
196 ** better RUN-PROGRAM behaviour in case of errors in the child
198 ** PROBE-FILE now simplifies pathnames correctly.
199 ** DIRECTORY now works correctly with :WILD-INFERIORS.
200 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
201 1st 1970 (although time timezone data will be unreliable).
202 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
204 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
205 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
206 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
208 * bug fix: two potential GC deadlocks affecting threaded builds.
209 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
210 returns T (reported by Anton Kazennikov)
211 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
212 with non-variable places
213 * bug fix: remove a race condition in the setting of
214 funcallable-instance functions, this should make threaded CLOS
215 code more stable against memory faults.
216 * bug fix: corruption of specials when unbinding is interrupted by an
217 asynchronous unwind (reported by Hannu Koivisto)
218 * improvement: the debugger will now also display local variables that
219 are only used once, for code compiled with a DEBUG optimization quality
222 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
223 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
224 * incompatible change: External-format support for FFI calls. The
225 SB-ALIEN:C-STRING no longer implies an ASCII
226 external-format. Instead the string is subject to external-format
227 conversion. Additionally return values of type SB-ALIEN:C-STRING
228 are by default of type (SIMPLE-ARRAY CHARACTER), not
229 SIMPLE-BASE-STRING. For an alien type that behaves like the old
230 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
231 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
232 * incompatible change: SB-EXT package no longer contains the
233 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
234 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
236 * incompatible change: the single-stepper is no longer available
237 on Alpha, Mips and Sparc platforms.
238 * minor incompatible change: the direct superclasses of
239 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
240 not (STANDARD-OBJECT FUNCTION). This makes the
241 class-precedence-lists of GENERIC-FUNCTION and
242 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
244 * minor incompatible change: the default stream external format on
245 non-unicode SBCL is no longer detected from the locale, but is always
247 * new feature: Add a version of evaluator that uses an interpreter instead
248 of the compiler. EVAL still uses the compiler by default, to switch it
249 to use the interpreter, set the value of the variable
250 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
251 * minor incompatible change: the single-stepper REPL has been merged
252 with the normal debugger (see the "Stepping" heading of the debugger help
253 for more details). The debugger command STEP will no longer switch
254 to the single-stepper REPL.
255 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
256 for a type now works.
257 * bug fix: loading of default sysinit file works. (thanks to Leonid
259 * bug fix: better detection of circularities in the file-compiler.
260 (reported by Marco Monteiro)
261 * bug fix: the CL pathname functions now work with files that have
262 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
263 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
264 systems with Unix98 pty semantics.
265 * bug fix: ASDF-INSTALL will now work with bsd tar.
266 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
268 * bug fix: timers expiring in dead threads no longer cause a
269 type-error (reported by Paul "Nonny Mouse").
270 * bug fix: thanks to more lightweight single-stepper instrumentation,
271 code compiled with (DEBUG 3) will compile and execute significantly faster,
272 and will have more accurate type-inferencing than before
273 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
274 whose bindings are modified
275 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
276 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
277 ** replace dummy implementations of CL:MACHINE-INSTANCE and
278 CL:SOFTWARE-VERSION with proper version
280 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
281 * feature: implemented the READER-METHOD-CLASS and
282 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
283 as specified by AMOP.
284 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
286 * optimization: faster LOGCOUNT implementation on x86 and x86-64
287 (thanks to Lutz Euler)
288 * optimization: hashing of general arrays and vectors has been
289 improved. (reported by Any Fingerhut)
290 * enhancement: SB-INTROSPECT is now able to find definitions of
291 profiled functions. (thanks to Troels Henriksen)
292 * enhancement: compiler-macro expansion applies now to FUNCALL forms
293 as well, allowing compiler-macros for SETF-functions to expand.
294 * enhancement: step-instrumentation no longer wraps known
295 single-value functions with multiple-value context, allowing
296 better type inference.
297 * fixed bug #337: use of MAKE-METHOD in method combination now works
298 even in the presence of user-defined method classes. (reported by
299 Bruno Haible and Pascal Costanza)
300 * fixed bug #339(c): if there are applicable methods not part of any
301 long-form method-combination group, call INVALID-METHOD-ERROR.
302 (reported by Bruno Haible)
303 * fixed bug #361: the :FUNCTION initarg in the protocol for
304 initialization of methods can now be used to override
305 internally-produced optimized functions. (reported by Bruno
307 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
308 system-provided lambda expression no longer cause warnings about
309 unbound #:|pv-table| symbols.
310 * bug fix: improved the handling of type declarations and the
311 detection of violations for keyword arguments with non-constant
313 * bug fix: potentially erronous calls to PATHNAME and
314 MERGE-PATHNAMES were being flushed in some cases.
315 (reported by Richard Kreuter)
316 * bug fix: compiled calls to TYPEP were mishandling obsolete
317 instances. (reported by James Bielman and Attila Lendvai)
318 * bug fix: format strings with ~> without matching ~< no longer
319 trigger an AVER, but signal an understandable error instead.
320 (reported by Antonio Martinez)
321 * bug fix: specifying an output-file without a pathname-type for
322 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
323 of input-file instead of "fasl". (reported by Robert Dodier)
324 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
325 (reported by James Y Knight).
326 * bug fix: compiler-macros-function did not consider the environment
327 argument for shadowing by local functions.
328 * bug fix: compiler-macros expansion was inhibited by local INLINE
330 * bug fix: inline expansions of known functions were subject to
331 step-instrumentation in high debug policies, leading to problems
333 * bug fix: compiler failed to differentiate between different CONS
335 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
336 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
337 * bug fix: #368: incorrect use of expressed vs. upgraded array
339 * bug fix: #306a: more precise unions of array types.
340 * thread-safety improvements:
341 ** CONDITION-WAIT could return early on Linux, if the thread was
342 interrupted and subsequently continued with SIGCONT.
343 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
345 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
346 * added support for the ucs-2 external format. (contributed by Ivan
348 * minor incompatible change: pretty printing of objects of type
349 (cons symbol) is, in the default pprint-dispatch-table, now
350 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
352 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
353 called later in a class's lifetime, possibly as late as when the
354 first instance of the class is created. Previously,
355 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
356 class became finalizeable.
357 * fixed bug: FILE-POSITION sometimes returned inconsistent results
358 for multibyte external-format streams. (thanks to "vbzoli")
359 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
360 with :ALLOCATION :CLASS inherited from superclasses of the
362 * fixed bug: anonymous classes can now be created using the :NAME
363 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
364 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
365 * fixed bug: core-files saved with :EXECUTABLE T can again be
366 executed when SBCL_HOME isn't set. (reported by James Knight)
367 * fixed bug: toplevel LOCALLY forms with declarations could
368 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
369 * fixed bug: printing from several different threads using different
370 values of *print-case* could cause invalid output, due to
371 some internal special variables of the printer not being bound
372 thread-locally (reported by Max Mikhanosha)
373 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
374 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
375 (reported by Pascal Costanza)
376 * fixed bug: FINALIZE-INHERITANCE is called from
377 REINITIALIZE-INSTANCE on classes when the class has previously
378 been finalized, as required by AMOP.
379 * minor code generation optimizations:
380 ** better register allocation in CLOS dispatching functions
381 ** overflow detection when coercing signed bytes to fixnums on x86-64
382 ** is now implemented with one IMUL instruction instead of three shifts
383 ** more efficient bit-vector access on x86 and x86-64
384 ** more efficient access to raw structure slots on x86 and x86-64
385 * fixed some bugs revealed by Paul Dietz' test suite:
386 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
389 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
390 * feature: thread support on Solaris/x86, and experimental thread support
392 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
394 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
395 default method for SLOT-UNBOUND.
396 * minor incompatible change: prevent the user from specializing the
397 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
398 somewhat counter to the intent of the protocol, I (CSR) think, and
399 additionally it just doesn't work in SBCL as currently
400 implemented, thanks to optimizations (that are always valid for
401 the other three SLOT-VALUEish functions, but not for the setter).
402 * minor incompatible change: the :SB-LDB feature is now enabled by
403 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
404 the low-level debugger.
405 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
406 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
407 down to an SB-SYS:FD-STREAM, allowing direct communication in
408 more cases. Temporary files and pipes are still used as a fallback
410 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
411 on the non-reentrant C functions localtime(3) and gmtime(3).
412 * bug fix: global optimization policy was not visible in LOCALLY and
414 * bug fix: class objects can be used as specializers in methods.
415 (reported by Pascal Costanza)
416 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
417 with a type error. (reported by blitz_ on #lisp)
418 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
419 failed with a type error. (reported by Pascal Bourguignon)
420 * bug fix: merging pathnames against defaults with :DIRECTORY
421 starting with '(:RELATIVE :BACK) should preserve the :BACK.
422 (reported by James Y Knight)
423 * bug fix: saving large (>2GB) cores on x86-64 now works
424 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
425 constant index and a (simple-array (signed-byte 32)) array
426 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
427 error (patch by Robert J. Macomber)
428 * feature: TIME now displays the amount of run-time spent in GC
429 * bug fix: The debugger now does a better job of respecting
430 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
431 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
432 workaround for bug 403.)
433 * bug fix: floating point exception handling now works on FreeBSD
434 (thanks to NIIMI Satoshi)
435 * fixed some bugs revealed by Paul Dietz' test suite:
436 ** MISC.641: LET-conversion were not supposed to work in late
438 * improvements to the Win32/x86 port:
439 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
440 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
442 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
443 * new feature: source path information is generated for macro-expansion
444 errors for use in IDE's like Slime (thanks to Helmut Eller)
445 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
446 * bug fix: compiler does not loop forever on an invalid type in
448 * improvement: compilation of most CLOS applications is significantly
450 * optimization: added a limited bytecode compiler for simple toplevel
451 forms, speeding up compilation and FASL loading
452 * bug fix: the statistical profiler now properly distinguishes anonymous
455 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
456 * minor incompatible change: in sbcl-0.9.11 (but not earlier
457 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
458 into subclasses of the finalized class. Now user calls to
459 FINALIZE-INHERITANCE finalize just the one class, and calls by the
460 system return before any subclasses are finalized.
461 * minor incompatible change: The reader no longer ignores errors
462 regarding non-existent packages in #+ and #- feature tests.
463 * new feature: command line options --no-sysinit, --no-userinit to
464 inhibit loading the corresponding init files
465 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
466 for sending data through UDP sockets (thanks to François-René Rideau)
467 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
468 error is signalled instead (thanks to François-René Rideau)
469 * minor incompatible change: Error signalling behaviour of lexical
470 operations violating package locks has changed slightly. Refer to
471 documentation on package locks for details.
472 * enhancement: EVAL can process IF-expressions without resorting to the
474 * bug fix: Some operations involving SIN, COS, or TAN weren't being
475 constant-folded properly due to missing out-of-line functions.
476 (reported by Mika Pihlajamäki)
477 * bug fix: LISTEN sometimes returned T even in cases where no data was
478 immediately available from the stream
479 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
480 were reversed. (reported by Levente Mészáros)
481 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
482 when given a vector displaced to an adjustable array. (reported
484 * bug fix: derivation of float boundaries from numbers outside the
485 appropriate float range (reported by John Wiseman)
486 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
488 * bug fix: precomputing cache entries for generic functions with
489 some subclasses of specializers as yet invalid does not attempt to
490 fill a cache line with a negative offset. (reported by Levente
492 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
493 allow condition class objects as arguments to DOCUMENTATION and
494 (SETF DOCUMENTATION); only find and set documentation for
495 structure names for the STRUCTURE doc-type. (suggested by Gary
497 * improvements to the Win32/x86 port:
498 ** (user-homedir-pathname) and default initialization file
499 locations now know about the user's "Documents and Settings"
500 directory (thanks to Yaroslav Kavenchuk)
501 ** run-program is implemented (thanks to Mike Thomas)
502 ** sockets support (thanks to Timothy Ritchey)
503 ** better backtrace support (thanks to Alastair Bridgewater)
504 ** sb-grovel supported
505 ** asdf-install and sb-posix work somewhat
506 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
507 * improvements to the Solaris/x86 port:
508 ** works on Solaris 11/Solaris Express
509 ** floating-point exception handling support
510 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
511 * fixed some bugs revealed by Paul Dietz' test suite:
512 ** REMOVE-METHOD returns its generic function argument even when
513 the method is not one of the generic functions' methods.
514 ** objects declared as MEMBER types can be admissible arguments to
516 ** printing characters should simply be printed by the FORMAT ~:C
518 ** compiler failure when compiling functions with hairy constant
519 defaults for optional parameters.
520 ** compiler produces wrong code when MAYBE-INLINE-expanding a
521 function, which is already optimized.
523 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
524 * new platform: experimental support for SBCL x86/Darwin, including
525 MacOS X 10.4.5 on Intel.
526 * new feature: Unicode character names are now known to the system
527 (through CHAR-NAME and NAME-CHAR).
528 * new feature: the filesystem location of SBCL's core file is
529 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
530 * minor incompatible change: the contrib modules SB-POSIX and
531 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
532 this change is to make it easier to distribute
533 location-independent binaries.
534 * bug fix: as implied by AMOP, standardized classes no longer have
535 slots named by external symbols of public packages. (reported by
537 * optimization: calling functions via constant symbols -- as in
538 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
539 via the function object as in (FUNCALL #'FOO).
540 * optimization: on x86 and x86-64, the calling convention for the
541 case of dealing with an unknown number of values has been altered
542 to be friendlier to the prediction heuristics implemented,
543 particularly with reference to CALL and RET pairing. (thanks to
544 Alastair Bridgewater)
545 * optimization: on x86, the code for access to thread-local symbol
546 values and binding/unbinding of thread-local symbols is smaller.
547 (thanks to Alastair Bridgewater)
548 * enhancement: CONSTANTP is now able to determine constantness of
549 more complex forms, including calls to constant-foldable standardized
550 functions and some special forms beyond QUOTE.
551 * fixed bug: occasional GC crashes on Solaris/x86
552 * optimization: x86-64 supports stack allocation of results of simple
553 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
554 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
555 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
556 (thanks to James Knight)
557 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
558 by the standard (thanks to Stephen Compall)
560 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
561 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
562 be used for bundling the runtime and the core file into one
563 executable binary. This feature is not currently supported on all SBCL
564 platforms. (thanks to James Bielman and NIIMI Satoshi)
565 * new feature: a generational or ephemeral garbage collector is now
566 the default on the PowerPC platform (both Linux and Darwin). The
567 old Cheney (stop and copy) collector is a build-time option.
568 (thanks to Cyrus Harmon, after Raymond Toy)
569 * minor incompatible change: the method by which SBCL finds its
570 contributed modules has changed; it no longer relies on symbolic
571 links from an $SBCL_HOME/systems directory, but searches directly
572 in the subdirectories of $SBCL_HOME.
573 * enhancement: the dynamic heap size on the Linux/PPC platform is
574 markedly larger, even using the older Cheney garbage collector.
575 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
576 certain complicated string types. (reported by Gary King)
577 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
578 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
579 * fixed bug: DOCUMENTATION on structures no longer signals an
580 error if no structure type of the right name exists. (reported by
582 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
583 applicable non-standard primary or :AROUND method on
584 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
585 no longer cause unbound variable errors. (reported by Kevin Reid)
586 * fixed bug: metacircle resolution in cases where methods have slots
587 added before the slots from STANDARD-METHOD. (reported by Jean
589 * fixed bug: the Power PC assembly code for calling into lisp
590 sometimes computed its return address wrongly (depending on the
591 15th bit of the address, fixed at link time). This used to
592 manifest itself in a segmentation violation while building PCL.
593 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
594 many others over the years)
595 * fixed bug: ROOM no longer reports silly numbers for stack usage if
596 the stack pointer (treated as a signed integer) is negative.
597 (thanks to Peter van Eynde)
599 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
600 * new platform: experimental support for the Windows operating
601 system has been added. (thanks to Alastair Bridgewater)
602 * fixed several bugs in and robustified the PPC FFI (including
603 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
604 * bug fix: allow non-simple string symbol names (reported by Paul
606 * bug fix: interrupt handling on NetBSD (thanks to Richard M
608 * bug fix: saving a core corrupted callbacks on x86/x86-64
609 * bug fix: closed a loophole in metacircularity detection and
610 grounding in the PCL implementation of CLOS.
611 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
612 (reported by tomppa on #lisp)
613 * bug fix: clear the direction flag on Lisp -> C transitions, as
614 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
615 (reported by Andrej Grozin and Hendrik Maryns)
616 * optimization: major improvements to GC efficiency on GENCGC platforms
617 * optimization: faster implementation of EQUAL
618 * optimization: emit more efficient opcodes for some common
619 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
620 * optimization: several other minor code-generation improvements on x86-64
621 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
623 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
624 * minor incompatible change: (SETF CLASS-NAME) and (SETF
625 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
626 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
627 by ANSI as a generic function, and for consistency (SETF
628 GENERIC-FUNCTION-NAME) is treated likewise.
629 * fixed bug #233.b: make constraint propagation notice when a variable
630 value is changed after it is referenced but before it is used
631 * fixed bug #296: no more arbitrary behaviour from filenames with
632 odd characters as --load arguments. (reported by Adam Warner)
633 * fixed bug #390: :CHARACTER-SET pathname components now work as
634 expected. (reported by Tim Daly Jr)
635 * fixed bug #391: complicated :TYPE intersections in slot
636 definitions no longer cause an error in PCL internals.
637 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
638 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
639 returning the number of octets which would be written to the
640 file-stream. (thanks to Robert J. Macomber)
641 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
642 arguments to the internal stream functions. (thanks to David
644 * bug fix: stack exhaustion is now handled on the x86/SunOS
645 platform. (thanks to Daisuke Homma)
646 * bug fix: more accurate ROOM results on GENCGC platforms
647 * optimization: improved type inference for arithmetic-for
648 index variables in LOOP
649 * optimization: faster floating-point SQRT on x86-64
650 * fixed some bugs revealed by Paul Dietz' test suite:
651 ** DOCUMENTATION returns NIL instead of "" for method combinations
652 that don't have a docstring
654 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
655 * minor incompatible change: (SETF CLASS-NAME) and (SETF
656 GENERIC-FUNCTION-NAME) are no longer generic functions, and
657 therefore are not customizeable by user code (as seems to be at
658 least permitted and maybe required by AMOP). As a consolation,
659 however, the SBCL implementation of these functions now calls
660 REINITIALIZE-INSTANCE as specified by AMOP.
661 * bug fix: slot-definition documentation is propagated as per ANSI
662 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
663 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
664 argument, as do eslotd initialization methods. (from Pascal
665 Costanza's "Closer" project)
666 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
667 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
669 * bug fix: it is now possible to have more than one subclass of
670 STANDARD-GENERIC-FUNCTION without causing stack overflow.
671 (reported by Bruno Haible, Pascal Costanza and others)
672 * bug fix: the dependent update protocol now works for generic
673 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
675 * bug fix: condition-class instances corresponding to
676 DEFINE-CONDITION forms are now created eagerly. (reported by
677 Kalle Olavi Niemitalo on comp.lang.lisp)
678 * bug fix: floating point printing is more accurate in some
679 circumstances. (thanks to Simon Alexander)
680 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
681 merged with *DEFAULT-PATHNAME-DEFAULTS*.
682 * bug fix: callbacks on OS X now preserve stack-alignment.
683 (thanks to Cyrus Harmon)
684 * enhancement: the x86-64 disassembler is much better at
685 disassembling SSE instructions. (thanks to Lutz Euler)
686 * enhancement: CHECK-TYPE on a variable now allows type-inference
687 (thanks to Vincent Arkesteijn)
688 * optimization: improved performance of EUC-JP external format.
689 (thanks to NIIMI Satoshi)
690 * optimization: performance improvements to IO on file streams of
691 :ELEMENT-TYPE CHARACTER
692 * optimization: much faster memory allocation on x86-64
693 * optimization: faster garbage collections (latency and throughput)
695 * optimization: faster bignum subtraction and division on x86 and x86-64
696 (thanks to Lutz Euler)
697 * optimization: more accurate type-inference for loops that use a
698 floating point index variable or a negative step.
700 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
701 * bug fix: add a workaround to SBCL looping infinitely at startup on
702 Linux kernels with apparently buggy implementations of personality().
703 (thanks to Svein Ove Aas)
704 * bug fix: Unicode symbols are correctly printed in LDB backtraces
705 (thanks to David Lichteblau)
706 * bug fix: local bindings shadow global symbol macros for
707 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
708 * optimization: non-open coded uses of numeric comparison operators
709 (e.g. >) no longer cons when called with more than one parameter
710 on platforms supporting dynamic-extent allocation.
711 * enhancement: saving cores with foreign code loaded is now supported
712 on MIPS/Linux in addition to the previously supported platforms.
713 * bug fix: FIND-RESTART now tests for activity, not applicability when given
714 a restart object as identifier. (reported by Helmut Eller for CMUCL)
715 * bug fix: division by zero in sb-sprof when no samples were collected
716 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
717 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
718 (thanks to Alexander Kjeldaas)
719 * new feature: ignoring the return values of destructive functions
720 like NREVERSE signals a compile-time style-warning.
721 (thanks to Kevin Reid)
722 * new feature: the alignment of alien structure fields can be
723 explicitly specified. (thanks to Cyrus Harmon)
724 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
725 and dump core on SIGQUIT
727 ** incompatible change: threads do not inherit values of specials
728 from their parents (see manual)
729 ** bug fix: threads stacks belonging to dead threads are freed by the
730 next exiting thread, no need to gc to collect thread stacks anymore
731 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
732 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
733 ** bug fix: dynamic variable and thread start related gc lossage
734 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
736 * fixed some bugs revealed by Paul Dietz' test suite:
737 ** SUBTYPEP is slightly more accurate on heinously complicated
738 CONS types where some of the members have uncertain (in the
739 NIL, NIL sense) type relationships to each other.
741 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
742 no longer purified unless :PURIFY T is explicitly specified.
743 ** Non-purified cores are significantly smaller than before
745 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
746 * new feature: timers based on Zach Beane's excellent timer package
747 * added support for the following external formats: koi8-u,
748 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
749 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
750 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
751 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
752 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
753 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
754 * incompatible change: a threaded SBCL will no longer revert to
755 non-threaded mode on non-NPTL systems, but refuse to start entirely.
756 * bug fix: interrupts are disabled until startup is complete; no
757 more sigsegvs when receiving a signal to soon
758 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
760 * bug fix: add a workaround for the memory randomization features in
761 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
762 workaround will only be in effect on systems with the proc filesystem
764 * bug fix: printing objects of type HASH-TABLE signals a
765 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
767 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
768 have been running for over 50 days. (reported by Gilbert Baumann)
769 * bug fix: the logic for getting names of functions gets less
770 confused when confronded with alternate-metaclass
771 funcallable-instances. (reported by Cyrus Harmon)
772 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
773 consistent, even on internal alternate-metaclass objects.
774 * bug fix: flush closure information collected by physical
775 environment analysis prepass before the main pass. (bug reported
777 * bug fix: compiler pack phase does not modify a hashtable, which is
778 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
779 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
780 may be expanded. (reported by Alexander <malishev> on c.l.l)
781 * bug fix: no more style-warnings when compiling code using LAST
782 under high-SPEED low-SPACE settings. (reported by David Wallin
784 * fixed some incompatibilities between SBCL's MOP and the MOP
786 ** the METAOBJECT class is now implemented;
787 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
788 STANDARD-OBJECT, as required;
789 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
790 now compatible (as required by VALIDATE-SUPERCLASS); there
791 remains a consistency requirement over the presence or absence
792 of the FUNCTION class in the superclass list of finalized
793 classes; see the manual for more details;
794 ** the :AROUND method for COMPUTE-SLOTS on
795 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
796 requested slot ordering.
798 ** bug fix: parent thread now can be gc'ed even with a live
800 ** bug fix: binding a special with PROGV to no value is not
801 the same as that symbol not having been bound (thanks to
803 ** bug fix: binding specials is thread safe (thanks to
805 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
806 and SB-SPROF do not die with 'no handler for signal XX in
807 interrupt_handle_now(..)' anymore
808 ** bug fix: WITH-TIMEOUT works with multiple running threads
809 * fixed some bugs revealed by Paul Dietz' test suite:
810 ** ENSURE-GENERIC-FUNCTION should take a method class object for
811 the :method-class keyword argument.
813 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
814 * new port: the Solaris operating system on x86 processors is now
815 mostly supported, though some rough edges in the environment
816 remain. (thanks to Daisuke Homma)
817 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
818 space for its heap. (thanks to Thiemo Seufer)
819 * optimization: SBCL on MIPS platforms now supports dynamic-extent
820 closures. (thanks to Thiemo Seufer)
821 * minor incompatible change: eof selects abort in the debugger.
822 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
823 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
825 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
826 deprecated, and will go away in a future revision of SBCL.
827 * minor incompatible change: GC-ON and GC-OFF are no longer
828 implemented with a counter, it does not matter how many times gc
829 is switched on or off
830 * bug fix: discriminating functions for generic function classes
831 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
832 make invalid assumptions about method precedence order. (reported
834 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
835 platforms. (thanks to Thiemo Seufer)
836 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
837 * bug fix: correct pathname printing with printer escaping is on.
838 (thanks to Kevin Reid)
839 * bug fix: complex VOP definitions in "user-space" no longer trigger
840 package locks. (reported by Zach Beane)
841 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
842 for funcallable-instance functions, and is no different from
844 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
845 characters in the current readtable. (reported by Nicholas Neuss)
846 * bug fix: SUBTYPEP on various CONS types returns more a more
847 accurate acknowledgment of its certainty.
848 * optimizations: REMOVE-DUPLICATES now runs in linear time on
849 lists in some cases. This partially fixes bug 384.
850 * flush all standard streams before prompting in the REPL and the
852 * bug fix: signal handling and triggering gc do not conflict
853 directly anymore, in particular a high frequency sb-sprof does
854 not prevent gc from running
855 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
856 approximation for timezone and DST information between the
857 universal time epoch and the smallest negative 32-bit time_t.
858 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
859 year 1899 so long as the encoded time is non-negative.
860 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
861 the correct answer for non-constant rotate amounts when performing
862 an inline 32-bit rotation.
864 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
865 there is only one thread in the session
866 ** bug fix: memory leak for streams created in one thread and
867 written to in another
868 ** bug fix: lockup when compiled with gcc4
869 ** bug fix: race that allows the gc to be triggered when gc is
871 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
872 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
874 ** bug fix: finished threads are gc'ed properly
875 * fixed some bugs revealed by Paul Dietz' test suite:
876 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
877 has arguments with a different set of applicable methods from
878 the orignal arguments.
879 ** The type error thrown by MAP now has a correct expected-type
881 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
882 name a compiled function.
883 ** Three MISC tests where a large bignum was improperly coerced to
884 a float that couldn't represent that bignum during type
885 derivation were fixed.
886 ** SUBTYPEP can now handle types involving the negation of a
887 list-form FUNCTION type.
888 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
889 as (COMPLEX (AND RATIO (NOT FIXNUM))).
890 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
892 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
893 * New feature: Experimental support for bivalent streams: streams
894 opened with :element-type :default now allow character and binary
895 (unsigned-byte 8) I/O
896 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
897 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
899 * bug fix: no more highly sporadic "couldn't check whether ~S is
900 readable" when reading a stream and an interrupt hits in the middle
901 of a select system call
902 * compiler better recognizes complex arrays (reported by Eduardo
904 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
905 for variables on SBCL built with linkage-tables. (reported by Luis
907 * various error reporting improvements.
908 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
909 (Thanks to Thiemo Seufer)
910 * bug fix: nested reader invokations work correctly
911 * bug fix: it is possible to have simultaneous references to foreign
912 code and foreign data with the same name.
914 ** added x86-64 support
915 ** incompatible change: the threading api now works with thread
916 objects instead of thread ids
917 ** bug fix: threads are protected from signals and interruption when
918 starting up or going down
919 ** bug fix: a race where an exiting thread could lose its stack to gc
920 ** bug fix: don't halt on infinite error in threads if possible
921 ** fixed numerous gc deadlocks introduced in the pthread merge
922 ** bug fix: fixed thread safety issues in read and print
923 ** bug fix: debugger doesn't hang on session lock if interrupted at
924 an inappropriate moment
925 ** bug fix: run-program is now thread safe(r)
926 ** bug fix: inner with-recursive-lock no longer releases the mutex
927 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
928 * fixed some bugs revealed by Paul Dietz' test suite:
929 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
930 EXPECTED-TYPE slots filled.
931 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
933 ** MAKE-CONDITION accepts classes as type-designators.
934 ** COMPILE may never return NIL.
935 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
936 range before calling Unix time functions
938 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
939 * numerous signal handling fixes to increase stability
940 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
941 * minor incompatible change: we now correctly canonize default
942 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
943 as per the MOP, rather than the historical (INITARG INITFUNCTION
944 INITFORM). (reported by Bruno Haible)
945 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
946 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
947 for more information.
948 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
949 pathname is a directory pathname.
950 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
951 * dynamic space size on PPC has been increased to 768Mb. (thanks to
953 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
954 :METACLASS argument in addition to a class name. (reported by
955 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
956 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
957 (reported by Bruno Haible; patch from Gerd Moellmann)
958 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
960 * bug fix: more cleanups to the floating point exception handling on
961 x86-64 (thanks to James Knight)
962 * bug fix: the compiler does not try to constant fold calls of
963 COERCE and BIT-* functions when they return freshly constructed
964 objects. (reported by Nikodemus Siivola and Paul Dietz)
965 * optimization: improved the MIPS versions of generic (in the
966 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
967 * optimization: direct conversion of (unsigned-byte 32) to floats on
968 the PowerPC platform.
969 * optimization: structure instances with raw slots now use less
970 memory, and probably show better memory locality. (thanks to
972 * optimization: DYNAMIC-EXTENT declarations for lists and closures
973 are treated as requests for stack allocation on the x86-64,
974 Alpha-32, PPC and SPARC platforms.
975 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
976 the wrong file descriptor; implementation of SOCKET-OPEN-P.
977 (thanks to Tony Martinez)
979 ** gcing a dead thread can no longer lead to lockups
980 ** threads block signals until they are set up properly
981 ** errno is no longer shared by threads
982 ** interrupt-thread restores the eflags register on x86
983 ** fixed some lockups due to gc/thread interaction
984 * fixed some bugs revealed by Paul Dietz' test suite:
985 ** invalid dotted lists no longer raise a read error when
987 ** signal an error if a symbol that names a declaration is used
988 as the name of a type, or vice versa
989 ** allow using the (declare (typespec var*)) abbreviation for
990 (declare (type typespec var*)) with all type specifiers
991 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
992 ** always return NIL from PPRINT-POP when OBJECT is NIL
993 ** don't signal errors when pretty-printing malformed LABELS,
994 FLET or MACROLET forms
995 ** declarations in a DOLIST body are also in scope for the
997 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
999 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1002 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1003 * fixed cross-compiler leakages that prevented building a 32-bit
1004 target with a 64-bit host compiler.
1005 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1006 opened with :IF-EXISTS :OVERWRITE.
1007 * fixed bug 281: error for an invalid qualifier in a short-form method
1008 combination method is not signalled until the faulty method is called.
1009 * bug fix: iteration variable type inferrer failed to deal with open
1010 intervals. (reported by Alan Shields)
1011 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1012 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1013 or more runtime options were provided to the sbcl binary.
1014 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1015 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1016 * optimization: there's now a fast-path for fixnum arguments in the
1017 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1018 * optimization: the code generated on x86-64 is more compact thanks
1019 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1020 * bug fix: floating-point exception handling now partly works on
1021 x86-64. (Thanks to James Knight)
1022 * improvement to the MIPS backend from Thiemo Seufer: C-style
1023 64-bit long long arguments and return values to alien functions
1025 * fixed some bugs revealed by Paul Dietz' test suite:
1026 ** the type-error signalled from WARN has a filled-in DATUM slot.
1027 ** the type-error required when a stream is not associated with
1028 a file has the stream as its datum.
1029 ** type-errors on single-floats on x86-64 no longer have
1030 :INVALID-OBJECT as the datum
1031 ** the type-errors signalled for invalid function names now have
1032 a correct expected type
1033 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1034 for structures defined with a :TYPE. Documentation strings for
1035 typed structures are no longer immediately discarded
1036 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1037 broadcast streams and synonym streams. FILE-LENGTH now also works
1038 on broadcast streams.
1040 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1041 * incompatible change: the --noprogrammer option, deprecated since
1042 version 0.7.5, has been removed. Please use the equivalent
1043 --disable-debugger option instead.
1044 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1046 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1047 inoperational for a while now) has been completely removed.
1048 * null lexical environments are now printed as #<NULL-LEXENV>,
1049 significantly reducing the amount of clutter in typical backtraces.
1050 * documentation on weak pointers, finalization, and after GC hooks
1051 has been added to the manual.
1052 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1053 as other specialized array types, is much faster. SUBSEQ and
1054 COPY-SEQ on such arrays have also been sped up.
1055 * optimization: EQL is now more efficient when at least other argument
1056 is known to be of type (OR FIXNUM (NOT NUMBER)).
1057 * fixed bug: the runtime is now less vulnerable to changes in the
1058 size of the SBCL object on OS X, and virtual memory is reserved for
1059 all spaces that need to be at a fixed address.
1060 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1061 * fixed bug: finalizers and after GC hooks that cause consing are now
1063 * fixed bug: compiler error messages and summaries are now printed to
1064 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1065 * fixed inference of the upper bound of an iteration variable.
1066 (reported by Rajat Datta).
1067 * fixed bug 211e: calling local functions with duplicated constant
1068 keyword argument no longer causes a bogus style warning about an
1070 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1071 ftype declaration to be disregarded. (reported by Dave Roberts)
1072 * fixed bug 373: caused by erronous compilation of references to alien
1073 variables in the runtime on ppc/darwin.
1074 * fixed bug 376: CONJUGATE type deriver.
1075 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1076 (reported by Baughn on #lisp)
1077 * a cross-compiler bug on non-x86 platforms has been identified and
1078 fixed. (thanks to Bruno Haible)
1079 * improvements to the MIPS runtime code for increased stability.
1080 (thanks to Thiemo Seufer)
1081 * increased the maximimum compact environment size to allow
1082 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1083 * partial workaround for bug 135: don't name the function we're
1084 calling for hairy cases of EVAL, so as not to accumulate one environment
1085 entry per call to EVAL. (reported by Kevin Reid)
1086 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1087 * optimization: type testing for non-vector arrays should be faster.
1088 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1090 * bug fix: setting 31st element of a bit vector to zero did not work
1092 * bug fix: redefining a class definition which failed due to a
1093 previous accessor / function clash now works (but see BUGS entry
1094 #380 for more problems in this area). (thanks to Zach Beane)
1095 * the long form of DEFINE-METHOD-COMBINATION disables method group
1096 checking when given a single method group with pattern *. (thanks
1098 * on x86 compiler supports stack allocation of results of simple
1099 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1100 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1101 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1102 * fixed some bugs related to Unicode integration:
1103 ** the restarts for recovering from input and output encoding
1104 errors only appear when there is in fact such an error to
1106 * fixed some bugs revealed by Paul Dietz' test suite:
1107 ** MISC.549 and similar: late transformation of unsafe type
1108 assertions into derived types caused unexpected code
1110 ** SCALE-FLOAT type deriver is less wrong.
1111 ** type derivers for EXP, LOG and similar functions compute result
1112 types for complex arguments better.
1113 ** (MISC.563) CONJUGATE type deriver works for very restricted
1115 ** out-of-line type testers for character strings are available.
1116 ** EQUAL compiler transform understands specialness of objects
1118 ** accessing double-floats stored on the stack now works on x86-64.
1119 ** debugger internals could sometimes create invalid lispobjs,
1120 resulting in GC crashes.
1121 ** MISC.548: type check weakening can convert required type into
1123 ** initialization forms for bindings are not in scope of free special
1126 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1127 * incompatible change: thread support for non-NPTL systems has
1128 been removed - locking is buggy and unreliable. A threaded
1129 SBCL build will now warn at startup and refuse to create
1130 new threads, unless futex support is detected in the kernel
1131 * incompatible change: the top level REPL now has only an ABORT
1132 restart associated with it, not TOPLEVEL and ABORT as it used to.
1133 TOP and TOPLEVEL are now available as debugger commands for
1134 returning to the top level.
1135 * incompatible change: forms evaluated in the REPL now use the
1136 global optimization policy.
1137 * incompatible change: user- and system-initialization files are
1138 no longer processed with LOAD, but by READ and EVAL; hence the
1139 global optimization policy, startup package, readtable, etc,
1141 * The .fasl file format number has been incremented because of
1142 various incompatible changes.
1143 * internal entry point details and argument counts no longer appear
1144 in backtraces unless explicitly requested by setting
1145 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1146 * built-in and standard functions no longer have names like "top
1147 level local call to FOO".
1148 * fixed bug 32: functions defined in non-null lexical environments
1149 now have more legible printed representation
1150 * fixed bug 33: functions defined in non-null lexical environemnts
1151 are now more amenable to inspection by INSPECT.
1152 * workaround for bug 354: XEPs no longer appear in backtraces unless
1153 explicitly requested.
1154 * fixed bug: receiving the signal which results from attempting to
1155 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1156 SIGBUS on other *BSDs) on architectures where the C stack is also
1157 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1158 to debugger confusion or wild instability. (reported by Cheuksan
1160 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1161 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1162 (reported by Lutz Euler)
1163 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1164 compiled code. (reported by Paul Dietz)
1165 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1166 specialized parameter to a method to something that is not TYPEP
1167 the specializer is now possible.
1168 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1169 face of package deletion.
1170 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1171 pathname. (reported by David Tolpin on c.l.l)
1172 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1173 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1174 STRING-STREAM and FILE-STREAM, respectively.
1175 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1176 than previously; in addition, its test suite is now run on build.
1177 (thanks to Luke Gorrie)
1178 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1179 * optimization: added a immediate representation for single-floats
1181 * fixed some bugs revealed by Paul Dietz' test suite:
1182 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1183 ** attempting to create a package with a colliding nickname causes
1184 correctable errors to be signalled.
1185 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1186 ** several x86-64 backend bugs related to sign-extension of immediate
1189 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1190 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1191 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1193 * fixed loading of multiply forward-referenced layouts.
1194 (thanks to Cheuksan Wang)
1195 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1196 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1197 (see COMPILE-FILE documentation for details of :PRINT option)
1198 * fixed bugs 19 and 317: fixed-format floating point printing is
1199 more accurate. This also fixes a bug reported by Adam Warner
1200 related to the ~@F format directive.
1201 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1203 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1204 dispatch table if the from-char is a dispatch macro character.
1205 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1206 by the wordsize no longer ignore the last word. (reported by Lutz
1208 * fixed bug in type checking of dynamic-extent variables. (reported
1210 * optimization: sequence traversal functions use their freedom to
1211 coerce function designators to functions.
1212 * optimization: code with many calls to CLOS methods specialized on
1213 CLOS classes has had redundant type checks removed.
1214 * optimization: added declarations to speed up operations that access
1215 the internal character database (for example STRING-UPCASE)
1216 * optimization: comparison operations between floats and sufficiently small
1217 fixnums no longer create extra rationals
1218 * fixed some bugs related to Unicode integration:
1219 ** portions of multibyte characters at the end of buffers for
1220 character-based file input are correctly transferred to the
1221 start of the buffer at the next read.
1222 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1223 passing it through to OPEN.
1224 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1225 argument given to internal calls to OPEN.
1226 ** the built-in comment readers (introduced by character sequences
1227 ";" and "#|") are more forgiving to encoding errors; they will
1228 STYLE-WARN and then attempt to resync the stream at a character
1229 boundary. (thanks to Teemu Kalvas)
1230 * fixed some bugs in the x86-64 port:
1231 ** Negative short int return values from c-calls are sign-extended
1233 ** The stack is aligned to 16-bytes for c-calls, as required by
1234 the ABI. (thanks to Cheuksan Wang)
1235 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1236 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1237 for enums. (thanks to Vincent Arkesteijn)
1238 ** Multiple small optimizations and bugfixes for floating point
1240 * fixed some bugs revealed by Paul Dietz' test suite:
1241 ** Space, Tab, Linefeed, Return and Page have the invalid
1242 secondary constituent character trait.
1243 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1245 ** WITH-INPUT-FROM-STRING should only update the index place on
1247 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1248 works more reliably.
1249 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1250 with one-dimensional arrays with fill pointers.
1251 ** TYPE-OF failed on a complex with an integer realpart and a
1253 ** compiler failure during type inference for the code of form
1254 (IF (EQL X (THE ...)) ...) (MISC.535).
1256 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1257 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1258 platform. The port supports 61-bit fixnums, large memory spaces
1259 and reloading shared object files.
1260 * enhancement: saving cores with foreign code loaded is now
1261 supported on ppc/Darwin in addition to the previously supported
1263 * enhancement: the statistical profiler now walks deeper into the
1264 call stack for more meaningful call-graphs and accrued time
1265 reports (x86/x86-64 only). It also now reports time spent in
1267 * enhancement: it is now possible to trace most individual methods
1268 of a generic function in addition to tracing the generic function
1270 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1272 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1273 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1274 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1275 produces an error. (thanks to Vincent Arkesteijn)
1276 * bug fix: NAMESTRING on pathnames with :WILD components in their
1277 directories works correctly. (thanks to Artem V. Andreev)
1278 * fixed bug 125: compiler preserves identity of closures. (reported
1280 * bug fixed: functions with &REST arguments sometimes failed with
1281 "Undefined function" when compiled with (DEBUG 3). (reported by
1283 * bug fix: overflow during compiling of setting element of a bit
1284 vector with constant index and value. (reported by Timmy Douglas)
1285 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1286 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1287 * on x86 compiler partially supports stack allocation of dynamic-extent
1289 * GO and RETURN-FROM do not check the extent of their exit points
1290 when compiled with SAFETY 0.
1291 * fixed some bugs related to Unicode integration:
1292 ** encoding and decoding errors are now much more robustly
1293 handled; it should now be possible to recover even from invalid
1294 input or output to the terminal. (thanks to Teemu Kalvas)
1295 ** provided a first cut at implementing STRING-TO-OCTETS and
1296 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1297 ** altered the SB-MD5 contributed module slightly, changing the
1298 interface just enough for it to be supportable for builds where
1299 lisp characters are not eight bits.
1300 * fixed some bugs revealed by Paul Dietz' test suite:
1301 ** the FORMATTER-generated functions for ~V[ conditionals require
1302 the correct number of arguments.
1303 ** READ-FROM-STRING returns the mandated second value when applied
1304 to displaced strings.
1305 ** the #\Rubout and #\Backspace characters are treated as invalid
1306 constituent characters by the tokenizer.
1308 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1309 * new feature: reloading changed shared object files with
1310 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1311 * new feature: references to foreign variables and functions
1312 can now be compiled and loaded before the corresponding shared
1313 object file is loaded, as long as the foreign definitions are
1314 available at runtime.
1315 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1316 supported. (thanks to Dan Debertin)
1317 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1318 just keywords. (thanks to Vincent Arkesteijn)
1319 * fixed bug #331: structure-class instances corresponding to
1320 DEFSTRUCT forms are now created eagerly.
1321 * fixed bug #345: backtraces from calls to undefined functions work
1322 on x86 as well. Related bug #61 is now also partially fixed on x86
1323 (backtraces from throws to unknown catch tags.)
1324 * bug fix: lambda-list parsing is now stricter vrt. order and number
1325 of lambda-list keywords.
1326 * bug fix: as specified by AMOP, an error is signalled if a
1327 class-option appears multiple times in a DEFCLASS form. (reported
1329 * bug fix: RANDOM can be compiled when the compiler derives the type
1330 of its numeric argument as a disjoint set of small integers.
1331 (reported by Paul Dietz)
1332 * bug fix: starting a core saved with shared objects loaded when
1333 those objects are not available no longer causes threaded SBCL to
1334 hang. (reported by Sean Ross)
1335 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1337 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1338 argument, which must always be NIL. (reported by Kalle Niemitalo)
1339 * bug fix: printing 1.0d+23 no longer results in an error.
1340 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1341 * bug fix: structure slot setters preserve evaluation order. (thanks
1343 * fixed some bugs related to Unicode integration:
1344 ** RUN-PROGRAM can allow its child to take input from a Lisp
1345 stream. (reported by Stefan Scholl)
1346 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1347 locales. (reported by Ken Causey)
1348 * fixed some bugs revealed by Paul Dietz' test suite:
1349 ** INCF, DECF and REMF evaluate their place form as specified in
1351 ** FORMATTER expands ~{ ~} iteration directives with V or #
1352 parameters correctly.
1353 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1354 consequent uses no arguments correctly.
1355 ** the system has a partial understanding of the (COMPLEX RATIO)
1357 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1360 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1361 * new feature: a build-time option (controlled by the :SB-UNICODE
1362 keyword feature, enabled by default) for building the system with
1363 support for the entire 21-bit character space defined by the
1365 * new feature: the system now has rudimentary external-format
1366 support; the primary user-visible change at this time is that
1367 characters with the high bit set (such as lower-case-e-acute) will
1368 print correctly to a terminal in a UTF-8 environment.
1369 * minor incompatible change: BASE-CHAR no longer names a class;
1370 however, CHARACTER continues to do so, as required by ANSI.
1371 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1372 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1373 has been moved to the SB-EXT package (temporarily re-exported from
1375 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1376 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1378 * fixed bug #351: better error-handlind and reporting for malformed
1380 * fixed bug #350: bignum-printing is now more memory-efficient,
1381 allowing printing of very large bignums, eg. (expt 2 10000000).
1382 (reported by Bruno Haible)
1383 * fixed bug #302: better primitive-type selection for intersection
1385 * fixed bug #308: non-graphic characters now all have names, as
1386 required. (reported by Bruno Haible)
1387 * bug fix: redefining a class with different superclasses now correctly
1388 removes it from the direct-subclasses of its previous superclasses.
1389 (reported by David Morse)
1390 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1391 new value now works. (reported by Bruno Haible)
1392 * bug fix: correct canonicalization of multiple non-standard slot
1393 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1394 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1395 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1396 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1397 now exists, an signals an error.
1398 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1399 during macroexpansion for non-iteration variables as well. (reported
1400 by Bruno Haible for CMUCL)
1401 * bug fix: Cyclic structures and unprintable objects in compiler
1402 messages no longer cause errors. (reported by Bruno Haible)
1403 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1404 and READ-FROM-STRING all now return a primary value of NIL if
1405 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1406 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1407 (reported by Bruno Haible for CMUCL)
1408 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1409 do not designate a condition. (reported by Bruno Haible for
1411 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1412 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1413 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1414 permitting resolution in favour of any of the conflicting symbols.
1415 (reported by Bruno Haible for CMUCL)
1416 * bug fix: EQUAL compiler optimizations is less aggressive on
1417 strings which can potentially compare true despite having distinct
1418 specialized array element types.
1419 * bug fix: unit enumerations can be defined without dividing by
1420 zero. (thanks to Vincent Arkesteijn)
1421 * FORMAT compile-time argument count checking has been enhanced.
1422 (report from Bruno Haible for CMUCL)
1423 * a partial workaround for the bug 262: the compiler does not try to
1424 inline-expand a local function doing RETURN-FROM from a deleted
1425 BLOCK. (thanks to Peter Denno for the bug report and to David
1426 Wragg for the simple test case)
1427 * fixed some bugs revealed by Paul Dietz' test suite:
1428 ** RENAME-PACKAGE allows all package designators as new package
1430 ** constraint propagation and lambda variable substitution are
1431 more cautious in dealing with partially deleted code.
1432 ** compiler sometimes lost reoptimization passes.
1433 ** CERROR, given a condition as condition designator, passes its
1434 remaining arguments to the continue format control without
1436 ** Case-altering FORMAT directives work correctly on non-ASCII
1438 ** The REAL type specifier handles bounds outside the floating
1439 point ranges without signalling FLOATING-POINT-OVERFLOW.
1440 ** Functions with IR1-transformations can create intercomponent
1441 references to global functions.
1442 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1444 ** FORMAT ~R treats a nil value for its first parameter correctly.
1446 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1447 * enhancement: saving cores with foreign code loaded is now
1448 supported on x86/NetBSD and sparc/Linux in addition to the previously
1449 supported platforms.
1450 * bug fix: on some platforms repeated installations caused multiple
1451 copies of HTML documentation to be installed -- should not happen
1452 any more. (reported by Stefan Scholl)
1453 * bug fix: parsing self-recursive alien record types multiple times
1454 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1455 original patch by Helmut Eller for CMUCL)
1456 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1457 (thanks to Richard Kreuter)
1458 * bug fix: defining classes whose accessors are methods on existing
1459 generic functions in other (locked) packages no longer signals
1460 bogus package lock violations. (reported by François-René Rideau)
1461 * bug fix: special variables as DEFMETHOD parameters no longer have
1462 associated bogus type declarations. (reported by David Wragg and
1464 * bug fix: read-write consistency on streams of element-type
1465 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1466 * bug fix: redefiniton of the only method of a generic function with
1467 no DEFGENERIC no longer emits a full WARNING. In addition,
1468 redefinition of generic functions with no DEFGENERIC to an
1469 incompatible lambda list now signals an error. (thanks to Zach
1471 * bug fix: DEFGENERIC now works even when there's a function of the
1472 same name in an enclosing lexical environment. (thanks to Zach
1474 * fixed compiler failure, caused by instrumenting code during
1475 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1476 * optimization: added loop analysis and improved register allocation
1477 * fixed some bugs revealed by Paul Dietz' test suite:
1478 ** POSITION on displaced vectors with non-zero displacement
1479 returns the right answer.
1480 ** (SIMPLE-STRING) is a valid type specifier for sequence
1482 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1484 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1485 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1487 ** The justification version of the FORMAT ~< directive treats
1488 non-zero minpad parameter correctly.
1490 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1491 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1492 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1493 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1494 the supported interface.
1495 * new feature: Single-stepping of code compiled with DEBUG 2 or
1496 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1497 * new feature: saving cores with foreign code loaded is now
1498 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1499 Timothy Moore's work for CMUCL)
1500 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1501 parameters to * instead of NIL if no initform is supplied.
1502 (reported by Johan Bockgaard)
1503 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1504 correctly when FUNCALL appears as the car of the form. Note:
1505 despite this FUNCALL forms are not currently subject to
1506 compiler-macro expansion. (port of Raymond Toy's fix for the
1507 same from CMUCL, reported by Johan Bockgaard)
1508 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1509 (thanks for Teemu Kalvas)
1510 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1511 case instead of terminating the iteration (thanks for Julian
1512 Squires, Sean Champ and Raymond Toy)
1513 * bug fix: incorrect expansion of defgeneric that caused a style
1514 warning. (thanks for Zach Beane)
1515 * bug fix: slot accessor effective method computation works properly
1516 for classes with multiple non-standard applicable methods on
1517 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1518 * on x86 compiler supports stack allocation of results of LIST and
1519 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1520 CMUCL implementation by Gerd Moellmann)
1521 * fixed some bugs revealed by Paul Dietz' test suite:
1522 ** FORMAT strings with both the ~<~:;~> form of the justification
1523 directive and pretty-printing directives cause an error.
1525 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1526 * incompatible change: the internal functions
1527 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1528 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1529 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1530 instead of the old functions.
1531 * new feature: on platforms where "dladdr" is available foreign
1532 function names now appear in backtraces. (based on Helmut Eller's
1534 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1535 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1537 * documentation: ASDF manual is now distributed with SBCL.
1538 * bug fix: SBCL can now load its contributed modules with REQUIRE
1539 even if the system-provided entries have been removed from
1540 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1542 * bug fix: backtraces involving undefined functions or assembly
1543 routines are more informative. (thanks to Brian Downing)
1544 * bug fix: mutually referent alien structures now work correctly.
1545 (reported by Rick Taube)
1546 * bug fix: structures defined by WITH-ALIEN can be referred to
1547 within other definitions in the same WITH-ALIEN.
1548 * bug fix: division operators (MOD, TRUNCATE and the like) with
1549 constant zero divisors and integer dividends no longer generate
1551 * bug fix: provide default methods for INPUT-STREAM-P and
1552 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1553 * bug fix: improve the detection and resolution of MOP metacycles
1554 (where effective method computation depends on the generic function
1555 having its effective method computed). (reported by Bruno Haible)
1556 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1557 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1558 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1559 * bug fix: pathnames with relative directory components can be
1560 represented relative to default pathnames.
1561 * optimization: in taking the GCD of bignums, reduce the two bignums
1562 to approximately the same size (using Euclid's algorithm) before
1563 applying the more sophisticated binary GCD. (thanks to Juho
1565 * optimization: COUNT on bitvectors now operates word-at-a-time.
1566 * optimization: ASH with a positive, but not necessarily constant,
1567 (leftwards) shift, when in a modular context, is compiled to a
1569 * fixed some bugs revealed by Paul Dietz' test suite:
1570 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1571 if the corresponding argument is NIL.
1572 ** FORMAT directives accepting a minpad parameter treat negative
1574 ** literal commas as character parameters to FORMAT directives are
1576 ** literal spaces directly after ~<Newline> directives within a
1577 format-logical-block (~:< ~@:>) do not induce :FILL-style
1578 conditional newlines.
1579 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1580 as specified: it no longer includes conditional newlines).
1581 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1583 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1584 :PER-LINE-PREFIX argument does not evaluate to a string.
1586 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1587 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1588 the manual for details; for now, package locks can be disabled by
1589 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1590 if no major problems are found then it is likely that they will be
1591 compiled in unconditionally.
1592 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1593 unsupported operators on all platforms. To load a shared library
1594 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1595 object file, link it into a shared library outside of SBCL and
1596 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1598 * bug workaround: ROOM T and the GC were not working together
1599 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1600 reporting facility aren't preserved. That reporting has been
1601 disabled, so now until and unless someone figures out how to make it
1602 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1603 * minor incompatible change: as threatened around sbcl-0.8.0, the
1604 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1605 The symbols are also exported from SB-PCL for backwards
1606 compatibility, but more so than before SB-PCL should be treated as
1607 an implementation-internal package.
1608 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1610 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1611 It is known as of this release not to work on the Alpha, however.
1612 * fixed bug #167: errors signalled due to illegal syntax in method
1613 bodies are now more legible.
1614 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1615 designators and can hence be used with TYPEP.
1616 * fixed bug #333: CHECK-TYPE now ensures that the type error
1617 signalled, if any, has the right object to be accessed by
1618 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1619 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1620 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1622 * fixed bug #334: programmatic addition of slots using specialized
1623 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1624 and :ALLOCATION :CLASS slots.
1625 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1626 float exponents. (rereported by Peter Seibel)
1627 * fixed a bug: #\Space (and other whitespace characters) are no
1628 longer considered to be macro characters in standard syntax by
1629 GET-MACRO-CHARACTER.
1630 * fixed bug: initialization of condition class metaobjects no longer
1631 causes an instance of the condition to be created. (reported by
1633 * fixed bug: it is now possible to have slots such that
1634 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1635 description is neither :INSTANCE nor :CLASS.
1636 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1637 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1638 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1639 non-standard SLOT-DEFINITION-ALLOCATION values.
1640 * fixed bugs: various race conditions handling exiting threads.
1641 CL-PPCRE's thread torture test now passes.
1642 * fixed arguably-a-bug: GC time proportional to number of threads in
1643 system even when most of them are idle
1644 * optimization: improved performance of BIT and SBIT on bit-vectors.
1645 * .fasl file incompatibility: The fasl file version number has
1646 been incremented because of changes associated with package locks.
1648 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1649 * minor incompatible change: the system no longer provides
1650 optimization hints (conditions which are TYPEP
1651 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1652 mode; these hints are still emitted when higher SPEED optimization
1654 * new contrib module: a sampling profiler (profiling by statistical
1655 sampling, rather than by instrumenting functions) is available as
1656 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1657 * the behaviour of the standard function ED is now customizeable by
1658 third parties through a hook variable: see ED's documentation
1659 string for information on the protocol.
1660 * the compiler no longer emits efficiency notes for (FUNCALL X)
1661 when the type of X is uncertain under default optimization
1663 * fixed bug 276: mutating a binding of a specialized parameter to a
1664 method to something that is not TYPEP the specializer is now
1666 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1667 exhibit the required behaviour on the x86 platform. (thanks to
1668 Peter van Eynde, Eric Marsden and Bruno Haible)
1669 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1670 even for difficult arguments. (reported by Peter Graves)
1671 * fixed bug 141a: the backquote printer now descends quoted
1673 * fixed another bug in backquote printing: no more destructive
1674 modification of the form's list structure. (reported by Brian
1676 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1677 move between its address being taken and the call to
1678 interrupt_thread, fixing a crashing race condition.
1679 * the SB-POSIX contrib implementation has been adjusted so that it
1680 no longer exhibits ridiculously poor performance when constructing
1681 instances corresponding to C structs.
1683 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1684 * minor incompatible change: the sb-grovel contrib now treats C
1685 structures as alien (in the sense of SB-ALIEN) objects rather than
1686 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1687 has implications for memory management of client code
1688 (sb-grovel-returned objects must now be manually managed) and for
1689 type safety (alien objects now have full types).
1690 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1691 used to control emission of compiler diagnostics, rather than the
1692 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1693 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1694 quality should be considered deprecated.
1695 * install.sh now installs the user manual as well
1696 * (not quite a new documentable feature, but worth considering in
1697 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1698 beginnings of a semantically meaningful condition hierarchy is
1699 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1701 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1702 designator as the defaults argument.
1703 * fixed bug: Displaced arrays whose displaced-to array has become
1704 too small now cause ARRAY-DIMENSION to signal an error, providing
1705 for safer bounds-checking. (reported by Bruno Haible)
1706 * fixed bug: DEFCLASS slot definitions with identical :READER and
1707 :WRITER names now signal a reasonable error. (reported by Thomas
1709 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1711 * fixed bug: Invalid slot specification errors now print correctly.
1712 (thanks to Zach Beane)
1713 * fixed bug 320: Shared to local slot value transfers in class
1714 redefinitions now happen corrently from superclasses as
1715 well. (reported by Bruno Haible)
1716 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1718 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1719 improper lists correctly. (reported by Bruno Haible)
1720 * fixed bug 313: source-transform for <fun-name> was erroneously
1721 applied to a call of a value of a variable with name <fun-name>.
1722 (reported by Antonio Menezes Leitao)
1723 * fixed bug 307: The obsolete instance protocol ensures that
1724 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1725 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1726 remove dead unknown-values globs from the middle of the stack.
1727 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1728 easier to resume long computations after using *BREAK-ON-SIGNALS*
1729 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1730 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1731 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1732 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1733 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1734 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1736 * fixed a (fairly theoretical) bug in string printing: if
1737 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1738 does not have array-element-type equal to the most general string
1740 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1741 function name. (thanks to Nikodemus Siivola)
1742 * fixed bug: compile-time format string checker failed on
1743 non-closed ~{. (reported by Thomas F Burdick)
1744 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1745 don't warn on legal loop code involving more than one aggregate
1746 boolean. (thanks to Nikodemus Siivola)
1747 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1748 in keyword position to MAKE-PACKAGE was not being reported as
1750 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1751 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1753 * fixed bugs 280 and 312: the checking for multiple definitions in a
1754 file is less likely to become confused by uses of inline
1756 * fixed bug: the #S reader macro performs the keyword coercion
1757 specified for slot names. (reported by Kalle Niemitalo)
1758 * fixed bug: lambda lists may contain symbols whose names start with
1759 & but are not lambda-list-keywords; their occurrence triggers a
1761 * fixed bug 321: define-method-combination argument lambda lists do
1762 not cause invalid code to be generated when &optional or &aux
1763 variables have default values. (reported by Bruno Haible)
1764 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1765 have CLOS classes; this appears also to have fixed reported
1766 instability in stack exhaustion detection.
1767 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1768 MUFFLE-WARNING when no associated restart is present is now
1770 * optimization: rearranged the expansion of various defining macros
1771 so that each expands into only one top-level form in a
1772 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1774 * optimization: used a previously little-used slot in symbols to
1775 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1777 * fixed some bugs revealed by Paul Dietz' test suite:
1778 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1780 ** Fixed an optimization bug related to inheritance of initforms
1781 from local to shared slots.
1782 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1783 ** CAST optimizer forgot to flush argument derived type.
1784 ** print/read consistency on floats is now orders of magnitude
1785 more likely. (thanks also to Bruno Haible for a similar report
1787 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1788 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1789 for objects of type REAL. Make it so.
1790 ** SXHASH is specified (infelicitously) to respect similarity,
1791 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1792 it so. (thanks to Markus Ziegler)
1793 ** on the Alpha, the compiler succeeds in compiling functions
1794 returning a known number of arguments greater than 63.
1795 ** fixed handling of invalid NIL arguments in keyword position
1797 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1799 ** provided a readably-printable representation for RANDOM-STATE
1801 ** ensured that pathnames, where they have a namestring, always
1802 print using #P"..." syntax.
1804 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1805 * Support for the forthcoming 2.0 version of the NetBSD kernel
1806 running on x86 hardware has been added. (thanks to Perry
1807 E. Metzger most immediately, and others for their past work)
1808 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1809 current release version; SBCL's OpenBSD support had been broken
1810 since about the time of OpenBSD's switch to ELF binary format.)
1811 * [placeholder for DX summary]
1812 ** user code with &REST lists declared dynamic-extent, under high
1813 speed or space and low safety and debug optimization policy.
1814 * The manual has been converted to Texinfo format and the debugger
1815 chapter from the cmucl manual has been added.
1816 * A facility has been added to extract documentation strings from
1817 sbcl and store them as Texinfo-formatted snippets for inclusion in
1818 the manual (via Texinfo's @include directive)
1819 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1820 the test case to Dave Roberts)
1821 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1822 pointers. (reported by Sean Ross)
1823 * bug fix: PROFILE output is printed nicely even for large numerical
1824 values. (thanks to Zach Beane)
1825 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1826 greater than 32 handle EOF correctly.
1827 * bug fix: on X86 an immediate argument of the IMUL instruction is
1828 correctly printed in disassembly. (thanks to Lutz Euler)
1829 * bug fix: class slots in redefined classes preserve their old
1830 values. (thanks to Bruno Haible and Nikodemus Siivola)
1831 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1832 succeeds. (reported by Marco Baringer)
1833 * fixed some bugs revealed by Paul Dietz' test suite:
1834 ** READ-SEQUENCE now works on ECHO-STREAMs.
1835 ** RATIONALIZE works more according to its specification. (thanks
1838 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1839 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1840 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1841 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1842 should matter to you only if you rebind the printer control
1843 variables and then find you want different bindings in the
1844 debugger than in the ordinary execution of your program.)
1845 * The runtime build system has been tweaked to support building
1846 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1847 assembler and linker. (thanks to Nikodemus Siivola)
1848 * Unbound, undefined, undeclared variables now trigger full
1849 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1850 more often programmer error than deliberate exploitation of undefined
1852 * optimization: the hash algorithm for strings has changed to one
1853 that is less vulnerable to spurious collisions. (thanks to Juho
1855 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1856 needless bounds checking. (thanks to Juho Snellman)
1857 * optimization: implemented multiplication as a modular
1858 (UNSIGNED-BYTE 32) operation on the PPC backend.
1859 * fixed some bugs revealed by Paul Dietz' test suite:
1860 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1862 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1863 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1864 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1866 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1868 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1870 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1871 their output stream on EOF from read.
1872 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1873 have been read to end-of-file.
1874 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1876 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1877 description of determination of which consecutive characters
1879 ** Printing the "Space" character with escaping on now yields "#\\ ",
1880 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1881 ** Reading floating-point numbers with *READ-BASE* set to a number
1882 less than 10 works correctly.
1883 ** Reading floating-point numbers with *READ-BASE* set to a number
1884 more than 10 works correctly.
1885 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1886 the readtable currently in effect.
1888 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1889 * minor incompatible change: parsing of namestrings on a physical
1890 (Unix) host has changed; numbers after the final #\. in a
1891 namestring are no longer interpreted as a version field. This is
1892 intented to be largely invisible to the user, except that the
1893 meaning of the namestring "*.*.*" has changed: it now refers to a
1894 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1895 should usually be replaced by
1896 (make-pathname :name :wild :type :wild :version :wild)
1897 with the added benefit that this is more likely to be portable.
1898 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1899 option to OPEN now signals an error if the file being opened
1900 exists; this may have an impact on existing code.
1901 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1902 signals received, on the PowerPC platforms (both Linux and
1903 Darwin). (thanks to Pierre Mai for pointing out the location of
1905 * several fixes on OS X: The system now builds and runs cleanly on
1906 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1907 (thanks to Brian Mastenbrook)
1908 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1909 second-resolution: integer multiples of 1/3600 between -24 and 24.
1910 (thanks to Vincent Arkesteijn)
1911 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1912 type when called with 1 argument; PEEK-CHAR checked type of
1913 PEEK-TYPE only after having read first character from a
1914 stream. (reported by Peter Graves)
1915 * bug fix: the garbage collector now has much better locality
1916 behaviour, and in particular no longer treats all memory as being
1917 exhausted when many small objects point to each other in a deeply
1919 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1920 recognized as being TYPEP their class.
1921 * bug fix: the PUSHNEW documentation string has been corrected.
1922 (thanks to Vincent Arkesteijn)
1923 * bug fix: defaulting of the value for the last of an atypically
1924 large number of multiple values being bound was not being
1925 performed correctly on the Alpha or PPC platforms
1926 * optimization: implemented multiplication as a modular
1927 (UNSIGNED-BYTE 32) operation on the x86 backend.
1928 * optimization: SEARCH on simple-base-strings can now be open-coded.
1929 (see also contrib/compiler-extras.lisp for inspiration for
1930 teaching the compiler about the Boyer-Moore algorithm).
1931 * value, returned by MAX (and MIN) called with several EQUALP, but
1932 not EQL, arguments now does not depend on compiler settings.
1933 * fixed some bugs revealed by Paul Dietz' test suite:
1934 ** in stack analysis liveness information is propagated from
1935 non-local entry points.
1936 ** pathwise CAST removing failed when the CAST node did not start
1938 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1939 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1941 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1942 host is already defined.
1943 ** RENAME-FILE works on streams instead of signalling an internal
1945 ** PEEK-CHAR uses the current readtable when determining whether
1946 or not a character is whitespace.
1947 ** MERGE-PATHNAMES handles the case when the pathname does not
1948 specify a name while the default-pathname specifies a version.
1949 ** Pathnames now stand a better chance of respecting print/read
1951 ** Attempting to use standardized file system operators with a
1952 pathname with invalid :DIRECTORY components signals a
1954 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1956 ** DIRECTORY on logical pathnames is more correct.
1957 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1958 signal a TYPE-ERROR if their argument is not a stream.
1959 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1960 designator argument does not designate a stream.
1961 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1962 examining the synonym.
1963 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1965 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1966 element-type larger than ([UN]SIGNED-BYTE 32).
1968 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1969 * When built with the :SB-FUTEX feature, threaded builds now take
1970 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1971 for faster/more reliable mutex and condition variable support.
1972 * Incompatible change (but one you probably shouldn't have been using
1973 anyway): the interface and code for arbitrating between multiple
1974 threads in the same user session has been redesigned.
1975 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1976 error when called without an explicit environment argument.
1977 (thanks to Nikodemus Siivola)
1978 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1979 confused about their position. (thanks to Adam Warner and Gerd
1981 * bug fix: Pretty printing backquoted forms with unquotations in the
1982 argument list position of various code constructs such as LAMBDA
1983 now works correctly. (reported by Paul Dietz)
1984 * bug fix: Pretty printing unquotations no longer loses all
1985 stream position information.
1986 * optimization: performance of string output streams is now less
1987 poor for multiple small sequence writes.
1988 * optimization: performance of CSUBTYPEP in the presence of complex
1989 expressions involving CONS and NOT many times has been improved.
1990 (reported by Paul Dietz)
1991 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1993 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1994 streams for the REPL, for the debugger, and so on. (thanks to
1996 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1997 * fixed some bugs revealed by Paul Dietz' test suite:
1998 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1999 is ignored if it is too small, rather than propagating through
2000 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2001 ** extremely complex negations of CONS types were not being
2002 sufficiently canonicalized, leading to inconsistencies in
2004 ** VALUES tranformer lost derived type.
2006 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2007 * fixed a bootstrapping bug: the build process no longer assumes
2008 that the various BOOLE-related constants have the same value in
2009 host and target lisps. (noted by Paul Dietz' test suite on an
2010 SBCL binary built from CLISP)
2011 * The system can now be dynamically linked on the MIPS platform,
2012 which enables dynamic loading of foreign code from Lisp. (thanks
2013 to Ralf Baechle for discussions on the MIPS ABI)
2014 * The system now records debugging information for its own source
2015 files in a filesystem-position-independent manner, relative to
2016 the "SYS" logical host.
2017 * fixed a compiler bug: MV-LET convertion did not check references
2018 to the "max args" entry point. (reported by Brian Downing)
2019 * tweaked disassembly notes to be less confident about proclaiming
2020 some instruction as an LRA. (thanks to Brian Downing)
2021 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2022 now each have their own history, command character, and other
2023 characteristics. (thanks to David Lichteblau)
2024 * fixed some bugs revealed by Paul Dietz' test suite:
2025 ** compiler failure in compiling LOGAND expressions including a
2027 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2028 shift greater than 32.
2029 ** FLUSH-DEST did not mark blocks for type check regeneration.
2030 ** HANDLER-CASE failed to accept declarations in handler clauses
2031 in some circumstances.
2033 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2034 * New code in contrib/sb-introspect (still probably not entirely
2035 stable yet) provides some support for smart Lisp development
2036 environments like SLIME.
2037 * The conditions signalled for errors occurring when loading .fasl
2038 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2039 in a way which should help ASDF recover gracefully.
2040 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2041 stuff has been cleaned up. If you code contrib/ stuff, this might
2042 affect you, and you can look at contrib/README, contrib/STANDARDS,
2043 and/or the 0.8.4.27 diff to check.
2044 * In full calls the compiler now does not generate checks for declared
2045 argument types for all arguments.
2046 * various threading fixes
2047 ** and some experimental patches which didn't make it into
2048 the main tree for this release, but which are shipped in
2049 contrib/experimental-thread.patch as a possible fix for some
2050 failures (deadlock, spinning...) in GC-intensive multithreaded
2052 * fixed PPC build problem (source code incompatibility of different
2053 library versions): added offsetof() hackery which attempts to divine
2054 where glibc maintainers put uc_mcontext today
2055 * fixed bug 282: compiler does not trust type assertions while passing
2056 arguments to a full call.
2057 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2058 &OPTIONAL VALUES type parameter.
2059 * fix bug 214: algorithm for noting rejected templates is now more
2060 similar to that of template seletion. (also reported by rydis on
2062 * fixed bug 141b: printing backquoted information readably and prettily
2063 inserts a space where necessary.
2064 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2065 (CHARACTER 10) are now reported as errors, rather than propagated
2066 as unknown types. (reported by piso on #lisp)
2067 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2068 with values NIL and :ERROR. (thanks to Milan Zamazal)
2069 * fixed bug 191c: CLOS now does proper keyword argument checking as
2070 described in CLHS 7.6.5 and 7.6.5.1.
2071 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2072 longer raise an error; further, using a list as a for-as-arithmetic
2073 counter now raises a meaningful error.
2074 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2075 now understood by sequence creation functions such as MAKE-SEQUENCE
2077 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2078 a STRING-INPUT-STREAM.
2079 * compiler enhancement: SIGNUM is now better able to derive the type
2081 * type declarations inside WITH-SLOTS are checked. (reported by
2083 * fixed some bugs revealed by Paul Dietz' test suite:
2084 ** incorrect optimization of TRUNCATE for a positive first
2085 argument and negative second.
2086 ** compiler failure in let-convertion during flushing dead code.
2087 ** compiler failure while deriving type of TRUNCATE on an
2088 interval, containing 0.
2089 ** ASH of a negative bignum by a negative bignum count now returns
2091 ** intersection of CONS types now canonicalizes properly, fixing
2092 inconsistencies in SUBTYPEP.
2094 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2095 * incompatible change: The --disable-debugger command line
2096 option now clobbers the debugger at a more fundamental
2097 level, by redefining #'INVOKE-DEBUGGER instead of by
2098 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2099 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2100 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2101 enter the debugger REPL and then suffer recursive errors
2102 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2103 while under the new system, BREAK in --disable-debugger mode
2104 terminates the system just as an unhandled error would.
2105 * fixed compiler performance when processing loops with a step >1;
2106 * bug fix: DOCUMENTATION now retrieves generic function
2107 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2108 support has been systematized, and now supports the methods
2109 specified by ANSI, along with a default method and a method for
2110 slot documentation. (reported by Nathan Froyd)
2111 * bug fix: effective methods associated with a generic function are
2112 no longer cached over a change of that generic function's method
2113 combination. (reported by Andreas Fuchs)
2114 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2115 names a non-existent directory. (thanks to Andreas Fuchs)
2116 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2117 the PPC platform, lead to stack corruption; this has been fixed.
2118 (reported by Rainer Joswig)
2119 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2120 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2121 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2122 arithmetic in general, is now much more reliable on the PPC
2124 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2125 platform now returns the right answer.
2126 * optimization: restored some effective method precomputation in
2127 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2128 precomputation is now tunable.
2129 * optimization: compiler-internal data structure use has been
2130 reviewed, and changes have been made that should improve the
2131 performance of the compiler by about 20%.
2132 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2133 simple-base-strings and simple-bit-vectors is improved.
2134 * optimization: the optimization of 32-bit logical and arithmetic
2135 functions introduced in version 0.8.3 on the x86 has been
2136 implemented on the mips, ppc and sparc platforms; an
2137 implementation of the same facility, but for 64-bit arithmetic,
2138 has been added for the alpha.
2139 * microoptimization: the compiler is better able to make use of the
2140 x86 LEA instruction for multiplication by constants.
2141 * bug fix: in some situations compiler did not report usage of
2142 generic arithmetic in (SPEED 3) policy.
2143 * bug 145b fix: compiler used wrong type specifier while converting
2144 MEMBER-types to numeric.
2145 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2147 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2148 index argument now works properly on the Alpha platform.
2149 * bug fix: floating point exception treatment on the Alpha platform
2151 * bug fix: FILE-POSITION works much better on string input and
2152 output streams. (thanks to Nikodemus Siivola)
2153 * bug fix: many threading/garbage collection symptoms sorted.
2154 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2155 might be pseudo-atomic.
2156 * internal change: Stopping for GC is now done with signals not
2157 ptrace. GC is now done in whichever thread wanted it, instead of
2159 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2161 * bug fix: result form in DO is not contained in the implicit
2163 * incompatible change: ICR structure is changed; the value part of
2164 CONTINUATION is now called LVAR; corresponding functions are
2165 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2166 * added type deriver for ISQRT (thanks to Robert E. Brown).
2167 * bug fix: better support for loading from the command line when an
2168 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2170 * fixed some bugs revealed by Paul Dietz' test suite:
2171 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2172 ** ROUND and FROUND now give the right answer when given very
2173 small float arguments.
2174 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2176 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2177 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2178 ** LCM should return a non-negative integer.
2179 ** PARSE-INTEGER returned the index of a terminator instead of the
2180 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2181 ** PARSE-INTEGER returned an incorrect index being applied to a
2183 ** LCM with two arguments of 0 returns 0 rather than signalling
2185 ** unsigned addition of a 32-bit constant with the high bit set no
2186 longer causes an internal compiler error on the x86.
2187 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2188 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2189 with negative last argument.
2190 ** byte specifiers with zero size and position no longer cause
2191 an error during type derivation.
2192 ** bignum multiplication on the Alpha platform now returns the
2194 * porting: The system now builds on SuSE AMD64, although it still
2195 generates a 32-bit binary.
2196 * .fasl file incompatibility: The fasl file version number has
2197 been incremented (because of the changes to internal compiler
2198 data structures referred to above).
2200 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2201 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2202 more accurately, on the Darwin kernel running on PowerPC hardware.
2203 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2204 * Compiler code deletion notes now signal a condition of type
2205 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2206 an associated MUFFLE-WARNING restart.
2207 * The compiler now performs limited argument count validation of
2208 constant format strings in FORMAT, and where appropriate in ERROR,
2209 CERROR and WARN. (thanks to Gerd Moellmann)
2210 * New ASDF-INSTALL contrib can be used for automatic download and
2211 installation of third-party Lisp code from CCLAN or other sites
2213 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2214 which forces another thread to execute a function supplied by the
2216 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2217 now accept and act upon their :ELEMENT-TYPE keyword argument.
2218 (reported by Martin Atzmueller, Edi Weitz)
2219 * bug fix: FILE-POSITION now accepts position designators up to
2220 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2221 is the greater. (thanks to Patrik Nordebo)
2222 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2223 Kalle Olavi Niemitalo)
2224 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2225 printable. (reported by Eric Marsden)
2226 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2227 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2228 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2229 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2230 (thanks to Stig Sandoe)
2231 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2232 UNREAD-CHAR now correctly includes the unread character in the
2233 target sequence. (thanks to Gerd Moellmann)
2234 * bug fix in threaded builds: the system can now be suspended and
2235 resumed by shell job control with minimal disruption.
2236 * bug fixes in times and timezones >2038AD
2237 * better handling of "where is GNU make?" problem in build scripts
2238 (thanks to Nikodemus Siivola)
2239 * new optimization: inside a named function any reference to a
2240 function with the same name is considered to be a self-reference;
2241 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2242 optimization quality.
2243 * new optimization on x86: logical functions and + now have
2244 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2245 used when the result is truncated to 32 bits.
2246 * VALUES declaration is partially enabled.
2247 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2248 * fixed some bugs revealed by Paul Dietz' test suite:
2249 ** The system now obeys the constraint imposed by
2250 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2251 types form a lattice under type intersection.
2252 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2253 ** ASSOC now ignores NIL elements in an alist.
2254 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2255 and (1+ MOST-POSITIVE-FIXNUM) answers.
2256 ** The addition of a method with invalid qualifiers to a generic
2257 function does not cause an error to be signalled immediately;
2258 a warning is signalled, and the error is generated only on
2259 calling the generic function.
2260 * changed .fasl file version number, in part to add type codes for
2261 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2262 obscure ANSI requirements
2264 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2265 * fixed bug 148: failure to inline-expand a local function left
2266 garbage, confusing the compiler.
2267 * fixed bugs 3cd: structure slot readers perform type check if the
2268 slot can have an invalid value (i.e. it is either not initialized
2269 or can be written with a less specific slot writer).
2270 * bug fix: the compiler now traps array references to elements off
2271 the end of an array; previously, the bounds checking in some
2272 circumstances could go off-by-one.
2273 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2275 * type declarations for array element types now obey the description
2276 on the CLHS page "Declaration TYPE", as per discussions on
2277 sbcl-help around 2003-05-08. This means that a declaration
2278 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2279 declaration, all references to BAR will be asserted or assumed
2280 (with THE, so dependent on compiler policy) to involve objects of
2281 type FOO. Note that no such declaration is implied in
2282 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2283 * declared types of functions from the "Conditions"
2284 chapter. (reported by Paul Dietz)
2285 * bug fix: CERROR accepts a function as its first argument.
2286 * bug fix: NTH an NTHCDR accept a bignum as index
2287 arguments. (reported by Adam Warner)
2288 * optimization: character compare routines now optimize comparing
2289 against a constant character. (reported by Gilbert Baumann)
2290 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2291 argument now works properly on the MIPS platform.
2292 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2293 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2294 declared to have a functional type, does not cause an error
2296 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2297 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2298 * fixed bug: if last continuation of a deleted block has a
2299 destination, this destination should be deleted too. (reported by
2301 * fixed a bug in the bootstrap process: the host compiler's values
2302 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2303 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2304 test case from Patrik Nordebo)
2305 * improved the ability of the disassembler on the PPC platform to
2306 provide helpful disassembly notes.
2307 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2308 the class in more cases than previously.
2309 * bug fix: FILE-POSITION now understands :START and :END for
2310 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2311 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2312 8) data. (thanks to David Lichteblau)
2313 * bug fix: it is possible to add a method to a generic function
2314 without lambda list.
2315 * bug fix: reader failed to signal END-OF-FILE inside an
2316 object representation. (reported by Nikodemus Siivola)
2317 * fixed some bugs revealed by Paul Dietz' test suite:
2318 ** LAST and [N]BUTLAST should accept a bignum.
2319 ** condition slot accessors are methods.
2320 ** (VECTOR NIL) is a subtype of STRING.
2322 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2323 * minor incompatible change: some nonsensical specialized lambda
2324 lists (used in DEFMETHOD) which were previously ignored now signal
2326 * minor incompatible change: the system is now aware of the types of
2327 variables in the COMMON-LISP package, and will signal errors for
2328 most violations of these type constraints (where previously they
2329 were silently accepted).
2330 * minor incompatible change: COMPILE-FILE now uses the freedom
2331 afforded (ANSI 3.2.2.3) to use derived function types for
2332 functions defined in the same file. This also permits the system
2333 to warn on static type mismatches and function
2334 redefinition. (Currently it does not work with high DEBUG level.)
2335 * minor incompatible change: VALUES declaration is disabled.
2336 * When issuing notes, the compiler now signals a condition of type
2337 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2338 restart for use in user handlers. It is expected that the
2339 COMPILER-NOTE condition will eventually become a condition
2340 supertype to a hierarchy of note types, which will then be
2341 handleable in a similar fashion. However, at the moment, no such
2342 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2343 but it's an implementation detail, not a classification for the
2345 * Changes in type checking closed the following bugs:
2346 ** type checking of unused values (192b, 194d, 203);
2347 ** template selection based on unsafe type assertions (192c, 236);
2348 ** type checking in branches (194bc).
2349 * A short form of VALUES type specifier has ANSI meaning (it has
2350 increased the number of situations when SBCL cannot perform type
2352 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2353 keywords or constants is permissible.
2354 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2355 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2356 classes. (thanks to Antonio Martinez)
2357 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2358 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2359 argument to a pretty-print stream. Instead, it leaves any such
2360 operations to DESCRIBE-OBJECT methods.
2361 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2363 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2364 lambda lists. (thanks to David Lichteblau)
2365 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2366 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2367 (thanks to Gerd Moellmann)
2368 * fixed bug 246: increased compilation speed of long
2369 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2371 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2373 * DEFCONSTANT now throws a condition of type
2374 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2375 constant to a non-EQL value; CONTINUE and ABORT restarts
2376 respectively change and preserve the value.
2377 * fixed bug 63: The code walker, part of the implementation of CLOS,
2378 is now better at handling symbol macros.
2379 * bug fix: There is no longer an internal implementation type named
2380 CL:LENGTH. (reported by Raymond Toy)
2381 * bug fix: In macro-like defining macros/special operators the
2382 implicit block does not enclose the lambda list.
2383 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2384 enclosed in parentheses) are not suitable as type specifiers, and
2385 their use properly signals an error now.
2386 * bug fix: An argument count mismatch for a type specifier in code
2387 being compiled no longer causes an unhandled error at compile
2388 time, but signals a compile-time warning.
2389 * fixed simple vector readable printing
2390 * bug fix: DESCRIBE takes more care over whether the class
2391 precedence list slot of a class is bound before accessing it.
2392 (reported by Markus Krummenacker)
2393 * bug fix: FORMATTER can successfully compile pretty-printer format
2394 strings which use variants of the ~* directive inside.
2395 * bug fix: SEARCH now applies its TEST predicate to the elements of
2396 the arguments in the correct order. (thanks to Wolfhard Buss)
2397 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2398 in type inference. (thanks to Robert E. Brown)
2399 * bug fix: Reading in symbols with an explicit package name of ""
2400 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2401 * fixed some bugs revealed by Paul Dietz' test suite:
2402 ** NIL is now allowed as a structure slot name.
2403 ** Arbitrary numbers, not just REALs, are allowed in certain
2404 circumstances in LOOP for-as-arithmetic clauses.
2405 ** Multiple class redefinitions before slot access no longer
2406 causes a type error.
2407 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2408 association between the name and a class.
2409 ** Generic functions with non-standard method-combination and over
2410 five methods all of which return constants no longer return NIL
2411 after the first few invocations. (thanks to Gerd Moellmann)
2412 ** CALL-NEXT-METHOD with no arguments now passes the original
2413 values of the arguments, even in the presence of assignment.
2414 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2416 ** DISASSEMBLE works with closures and funcallable instances.
2417 ** ADD-METHOD now returns the generic function, not the new method.
2418 ** FIND-METHOD signals an error if the lengths of the specializers
2419 is incompatible with the generic function, even if the ERRORP
2421 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2422 which its argument is a member.
2423 ** DEFCLASS only redefines the class named by its class-name
2424 argument if that name is the proper name of the class;
2425 otherwise, it creates a new class.
2426 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2427 of the UNBOUND-SLOT condition to the name of the slot.
2428 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2429 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2430 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2431 SLOT-MAKUNBOUND in the specified fashion.
2433 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2434 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2435 cross-compilation host. As a consequence, we can now bootstrap our
2436 way up to SBCL starting with a bare gcc toolchain and human-readable
2437 source code (first the source to CLISP, then the source to SBCL).
2438 * A contributed module containing a partial implementation of the
2439 simple-streams interface has been included. (thanks to Rudi
2441 * A contributed module implementing the RFC1321 Message Digest
2442 Algorithm, known as MD5, has been included.
2443 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2444 no longer has any effect, as the code controlled by this feature
2445 has been deleted. (As far as we know, no-one has ever built using
2446 this feature, and its semantics were confused in any case).
2447 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2448 work on conditions (as required by the ANSI specification),
2449 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2450 expected behaviour on conditions. Users should note, however,
2451 that such behaviour is not required by the ANSI specification,
2452 and so use of this behaviour may render their code unportable.
2453 * fixed some bugs revealed by Paul Dietz' test suite:
2454 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2456 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2457 ** SXHASH on condition objects no longer returns NIL.
2458 ** :ALLOCATION :CLASS slots are better treated; their values are
2459 updated on class redefinition, and initforms inherited from
2460 superclasses are applied.
2461 ** REMOVE-METHOD returns its generic function argument even when
2462 no method was removed.
2463 ** SHARED-INITIALIZE now initializes the values of the requested
2464 slots, including those with :ALLOCATION :CLASS.
2465 ** ALLOCATE-INSTANCE now works on structure classes defined via
2466 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2468 ** SLOT-EXISTS-P now works on conditions, as well as structures
2470 ** MAKE-LOAD-FORM now has the required methods on
2471 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2472 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2473 keyword, and now implements the SLOT-NAMES argument.
2474 ** methods with &OPTIONAL arguments no longer allow too many
2475 arguments to be passed in the call without error.
2476 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2477 option is consistent with the required arguments of the generic
2478 function lambda list.
2479 * bug fix: REQUIRE accepts a string designator. (Thanks to
2481 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2482 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2483 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2484 * bug fix: The compiler now checks for duplicated variables in macro
2486 * bug fix: SETQ on globals returns the correct value.
2487 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2488 not a COMPILER-ERROR (followed by some other strange error on
2489 choosing the CONTINUE restart).
2490 * bug fix: make.sh and friends are now more consistent in the way that
2491 they look for GNU "make".
2493 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2494 * experimental native threads support (on x86 Linux >=2.4 only).
2495 This is not compiled in by default: you need to add :SB-THREAD to
2496 the target features. See the "Beyond ANSI" chapter of the manual
2498 * fix for longstanding nonANSIism: The old distinction between
2499 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2500 The return value from CL:FIND-CLASS is now a CLOS class, and
2501 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2502 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2503 * An interface to the MetaObject Protocol, as described in Kiczales,
2504 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2505 MIT Press, 1991, is available from the SB-MOP package.
2506 * incompatible change: the SB-PCL package should now be considered
2507 a private implementation detail, and no longer a semi-private MOP
2509 * minor incompatible change: due to rearrangement for threads, the
2510 control stack and binding stack are now allocated at arbitrary
2511 addresses instead of being hardcoded per-port. Users affected by
2512 this probably have to be doing advanced things with shared
2513 libraries, and will know who they are.
2514 * minor incompatible change: Previously, all --eval forms used were
2515 processed with READ before any of them were processed with EVAL.
2516 Now each --eval form is processed with both READ and EVAL before
2517 the next --eval form is processed. (Thus package operations like
2518 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2519 work as the user might reasonably expect.)
2520 * minor incompatible change: *STANDARD-INPUT* is now only an
2521 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2523 * minor incompatible change: Y-OR-N-P is now character-oriented,
2524 not line oriented. Also, YES-OR-NO-P now works without errors.
2525 (thanks to Antonio Martinez)
2526 * sb-aclrepl module improvements: an integrated inspector, added
2527 repl features, and a bug fix to :trace command.
2528 * Known functions, which cannot be open coded by the backend, are
2529 considered to be able to check types of their arguments. (fixing
2530 a bug report by Nathan J. Froyd)
2531 * fixed a bug in computing method discriminating functions: It is
2532 now possible to define methods specialized on classes which have
2533 forward-referenced superclasses. (thanks to Gerd Moellmann)
2534 * fixed evaluation order in optional entries (reported by Gilbert
2536 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2537 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2538 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2539 required-by-AMOP NAME argument, as well as CLASS and
2540 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2541 * fixed bug 20: DEFMETHOD can define methods using names that are
2542 not the proper names of classes to designate class specializers.
2543 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2544 Unix file descriptors, instead of blowing up. (thanks to Antonio
2546 * Garbage collection refactoring: user-visible change is that a
2547 call to the GC function during WITHOUT-GCING will not do garbage
2548 collection until the end of the WITHOUT-GCING. If you were doing
2549 this you were probably losing anyway.
2550 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2551 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2552 TYPEP the latter but not the former.
2553 * The compiler issues a full WARNING for calls to undefined functions
2554 with names from the CL package.
2555 * MAP-INTO for a vector destination is open coded. (reported by
2556 Brian Downing on c.l.l)
2557 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2558 documentation string.
2559 * fixed some bugs revealed by Paul Dietz' test suite:
2560 ** COPY-ALIST now signals an error if its argument is a dotted
2562 ** Condition slots are now accessed more correctly in the presence
2563 of multiple initargs for a given slot.
2564 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2565 exclude restarts of the same name associated with a different
2567 ** DEFCLASS of forward-referenced classes with another
2568 forward-referenced class in the superclasses list no longer
2570 ** Condition slots are now initialized once each, not multiple
2571 times. (thanks to Gerd Moellmann)
2572 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2574 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2575 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2576 arguments contain duplicated elements.
2577 ** RESTART-CASE understands local macros.
2578 ** RESTART-CASE associates exactly its own restarts with a condition.
2579 ** ENDP in safe mode checks its argument to be of type LIST.
2580 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2581 in question is unbound.
2582 ** Optimized MAKE-INSTANCE functions no longer cause internal
2583 assertion failures in the presence of duplicate initargs.
2584 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2585 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2586 propagated even in the ctor optimized implementation of
2588 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2590 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2591 * a better implementation of SXHASH on (simple) bit vectors,
2592 measured both in execution speed and in distribution of results
2593 over the positive fixnums, has been installed. Likewise, a better
2594 implementation of EQUAL for simple bit vectors is now available.
2595 * fixed CEILING optimization for a divisor of form 2^k.
2596 * fixed bug 240 (emitting extra style warnings "using the lexical
2597 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2598 by Antonio Martinez)
2599 * fixed SXHASH, giving different results for NIL depending on type
2600 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2601 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2602 variable. (found by Rolf Wester)
2603 * SBCL does not ignore type declarations for special
2604 variables. (reported by rif on c.l.l 2003-03-05)
2605 * some bug fixes in contrib/sb-aclrepl/
2606 * fixed some bugs revealed by Paul Dietz' test suite:
2607 ** a bug in the CONS type specifier, whereby the CAR and CDR
2608 types got intertwined, has been fixed;
2609 ** the type system is now able to reason about the interaction
2610 between INTEGER and RATIO types more completely;
2611 ** APPEND, [N]REVERSE and NRECONC check that those their
2612 arguments, which must be proper lists, are really so;
2613 ** An array specialized to be unable to hold elements has been
2614 implemented, as required -- yes, really -- by ANSI;
2615 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2616 on malformed property lists;
2618 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2619 * incompatible packaging change: in line with Unix convention,
2620 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2621 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2622 * REQUIRE and PROVIDE are now optionally capable of doing something
2623 useful. See the documentation string for REQUIRE.
2624 * infrastructure for a managed SBCL contrib system: contributed
2625 modules in this release include:
2626 ** the ASDF system definition facility;
2627 ** an interface to the BSD Sockets API;
2628 ** an ACL-like convenience interface to the repl;
2629 (thanks to Kevin Rosenberg)
2630 ** an implementation of ROTATE-BYTE, with efficient implementation
2632 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2633 gives a closer approximation to the right answer than previously.
2634 (thanks to Raymond Toy)
2635 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2636 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2637 optional environment arguments, as required by ANSI.
2638 * fixed bugs in other functions taking environment objects, allowing
2639 calls with an explicit NIL environment argument to be compiled
2641 * fixed bug 228: primary return values from
2642 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2643 COMPILE or FUNCTION.
2644 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2645 :TYPE will no longer signal errors on innocuous objects.
2646 * fixed bug 231b: SETQ is better at respecting type declarations in
2647 the lexical environment.
2648 * fixed a bug in DEFCLASS: classes named by symbols with no or
2649 unprintable packages can now be defined.
2650 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2651 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2652 * fixed compiler failure related to checking types of functions.
2653 (reported by Robert E. Brown)
2654 * the compiler is now much more consistent in its error-checking
2655 treatment of bounding index arguments to sequence functions: in
2656 (SAFETY 3) code, errors will be signalled in almost all cases if
2657 invalid sequence bounding indices are passed to functions defined
2658 by ANSI to operate on sequences.
2659 * fixed a bug in the build procedure: documentation of SBCL-specific
2660 packages is now preserved and available in the final Lisp image.
2661 * lifted FDEFINITION lookup out of loops in the implementation of
2662 many list operations. (thanks to Robert E. Brown)
2663 * fixed a bug in the reader: the #n# reader macro now works for
2664 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2665 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2666 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2667 E. Brown and Tony Martinez respectively)
2668 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2669 SLOT-BOUNDP: in particular, it is now able to optimize them much
2670 better, and is now not vulnerable to having packages renamed.
2671 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2672 always called when a slot is not present in an instance. (thanks
2674 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2675 longer signal an error on unreading a character following EOF on
2676 the previous constituent stream. (thanks to Tony Martinez)
2677 * fixed some bugs revealed by Paul Dietz' test suite:
2678 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2679 not just nonnegative fixnums;
2680 ** the logical bit-array operators such as BIT-AND now accept an
2681 explicit NIL for their "opt-arg" argument (to indicate a
2682 freshly-consed result bit-array);
2683 ** ELT now signals an error on an invalid sequence index in safe
2685 ** the type system is now cleverer about negations of numeric
2686 types, and consequently understands the BIGNUM and RATIO types
2688 ** the type system is now cleverer about the interaction between
2689 INTEGER and RATIO types: while bugs still remain, many more
2690 cases are accurately computed;
2691 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2692 if it is in the last clause;
2693 ** CONSTANTLY now correctly returns a side-effect-free function in
2695 ** DECLARE is no longer treated as a special-operator; in
2696 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2697 * incremented fasl file version number due to the change in the
2698 DEFSTRUCT-SLOT-DESCRIPTION structure.
2700 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2701 * minor incompatible change: code processed by the "interpreter" or
2702 EVAL now has a compilation optimization policy of (DEBUG 2)
2703 (changed from (DEBUG 1)) to improve debuggability of interactive
2704 development, and to allow the use of the debug RETURN command in
2706 * an experimental implementation of the RETURN command for the
2707 debugger has been included. (thanks to Frederik Kuivinen)
2708 * fixed bug 62: constraints were not propagated into a loop.
2709 * fixed bug in embedded calls of SORT (reported and investigated by
2711 * fixed some bugs revealed by Paul Dietz' test suite:
2712 ** printing and reading of arrays with some dimensions having
2713 length 0 (thanks to Gerd Moellmann);
2714 ** BOA constructor with &AUX argument without a default value does
2715 not cause a type error;
2716 ** CONSTANTP now returns true for all self-evaluating objects.
2718 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2719 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2720 accessors that are related by inheritance, as specified in the
2721 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2722 Valtteri Vuorikoski)
2723 * The compiler is now able to inline functions that were defined in
2724 a complex lexical environment (e.g. inside a MACROLET).
2725 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2727 * Support for the upcoming FreeBSD-5.0 release has been included.
2728 (thanks to Dag-Erling Smorgrav)
2729 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2730 effect when it is not in a toplevel context.
2731 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2732 stand a better chance of being correct. (thanks to Gerd
2734 * fixed bug in COERCE, which now signals an error on coercing a
2735 rational to a bounded real type which excludes the expected
2737 * The compiler is now able to derive types more accurately from the
2738 COERCE and COMPILE functions.
2739 * fixed bug 223: functional binding is considered to be constant
2740 only for symbols in the CL package.
2741 * fixed bug 231: SETQ did not check the type of a variable being set
2742 (reported by Robert E. Brown)
2743 * A new optimization for MAKE-INSTANCE has been included, fixing
2744 various bugs (including relating to :ALLOCATION :CLASS slots and
2745 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2747 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2748 ** As required by ANSI, LOOP now disallows anonymous collection
2749 clauses such as COLLECT I in conjunction with aggregate boolean
2750 clauses such as THEREIS (= I 1);
2751 ** LOOP now signals an error when any variable is reused in the
2752 same loop (including the potentially useful construct analogous
2753 to WITH A = 1 WITH A = (1+ A);
2754 ** IT is only a special loop symbol within the first clause of a
2755 conditional loop clause;
2756 ** LOOP with a typed iteration variable over a hashtable now
2757 signals a type error iff it should.
2758 * fixed some other bugs revealed by Paul Dietz' test suite:
2759 ** FILE-STREAM now names the class previously known as FD-STREAM;
2760 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2761 argument) no longer signals an error;
2762 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2763 of the slot symbol, rather than using the current package
2764 ((:CONC-NAME "") continues to intern the slot's name in the
2766 * incremented fasl file version number, because of the incompatible
2767 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2768 of the new implementation of DEFINE-COMPILER-MACRO.
2770 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2771 * Support for building SBCL for MIPS platforms running in
2772 little-endian mode has now been checked in, and basic
2773 functionality on said platforms verified.
2774 * minor incompatible change: PCL now records the pathname of a file
2775 in which methods and the like are defined, rather than its
2777 * minor incompatible change: TRUENAME now considers the truename of
2778 a file naming a directory to be the pathname with :DIRECTORY
2779 component indicating that directory.
2780 * minor incompatible change: a NAMED clause in the extended form of
2781 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2782 reason for the previous behaviour is unclear.
2783 * more systematization and improvement of CLOS and MOP conformance
2784 in PCL (thanks to Gerd Moellman and Pierre Mai):
2785 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2787 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2788 primary methods with no specializers;
2789 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2791 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2792 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2793 FORWARD-REFERENCED-CLASSes; error reporting on
2794 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2796 ** SXHASH on CLOS instances now uses a slot internal to the
2797 instance to return different numbers on distinct instances,
2798 while preserving the same return value through invocations of
2800 ** DEFMETHOD signals errors when methods with longer incongruent
2801 lambda lists are added to generic functions;
2802 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2803 CLASS, as specified in AMOP;
2804 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2805 on the order returned by the primary method for classes of
2806 class STANDARD-CLASS;
2807 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2808 * fixed some bugs shown by Paul Dietz' test suite:
2809 ** DOLIST puts its body in TAGBODY;
2810 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2812 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2813 value producing form;
2814 ** if more variables are given to PROGV than values, extra
2815 variables are bound and made to have no value;
2816 ** NSUBSTITUTE on list arguments gets the right answer with
2818 ** ELT signals an error of type TYPE-ERROR when the index argument
2819 is not a valid sequence index;
2820 ** LOOP signals (at macroexpansion time) an error of type
2821 PROGRAM-ERROR when duplicate variable names are found;
2822 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2823 ** FUNCALL of special-operators now cause an error of type
2825 ** PSETQ now works as required in the presence of side-effecting
2826 symbol-macro places;
2827 ** NCONC accepts any object as its last argument;
2828 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2830 ** loop-for-as-package does not require a package to be explicitely
2832 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2834 * fixed bug 166: compiler preserves "there is a way to go"
2835 invariant when deleting code.
2836 * fixed bug 172: macro lambda lists with required arguments after
2837 &REST arguments now cause an error to be signalled. (thanks to
2839 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2840 bug. (thanks to Matthew Danish)
2841 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2843 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2844 arguments are lexically rebound. (thanks to Gerd Moellmann and
2846 * fixed bug 194: error messages are now more informative when there
2847 is no primary method applicable in a call to a generic
2848 function. (thanks to Gerd Moellmann)
2849 * fixed bug in command line argument checking (thanks to Julian
2851 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2852 (thanks to Matthew Danish)
2853 * incremented fasl file version number, because of the
2854 SXHASH-related changes in the layout of CLOS data structures
2856 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2857 * minor incompatible change: The runtime (the Unix executable named
2858 "sbcl") is now much pickier about the .core files it will load.
2859 Essentially it now requires .core files to descend from the same
2860 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2861 as the runtime does. (The intent is to prevent the crashes which
2862 can occur, and which can even be reported as mysterious failures,
2863 when people patch the sources or change the build parameters
2864 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2865 sbcl and .core files.)
2866 * fixed bug: VALUES-LIST is no longer optimized away.
2867 * fixed bug 142: The FFI conversion of C string values to Lisp
2868 string values no longer conses excessively. (thanks to Nathan
2869 Froyd porting Raymond Toy's fix to CMU CL)
2870 * began to systematize and improve MOP conformance in PCL (thanks to
2871 Nathan Froyd, Gerd Moellman and Pierre Mai):
2872 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2874 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2875 ** FINALIZE-INHERITANCE is now called on class finalization;
2876 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2877 argument precedence order.
2878 * fixed bug 202: The compiler no longer fails on functions whose
2879 derived types contradict their declared type.
2880 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2881 so it can be non-toplevel.
2882 * The fasl file version number has changed (because of the new
2883 implementation of DEFMACRO).
2884 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2885 safe code, that any length requirement by their type-specifier
2886 argument is valid. The exceptions to this are described in bug
2888 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2889 functions now check the types of their inputs as required by ANSI.
2890 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2891 introduced symbol is DECLAREd to be SPECIAL.
2892 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2893 * fixed bug: :COUNT argument to sequence functions may be negative
2894 * fixed bug: body of DO-SYMBOLS may contain declarations
2895 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2896 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2897 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2898 symbol macro only once
2899 * fixed printing of call frame when argument list is unavailable
2900 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2901 * compiler no longer signals WARNING on unknown keyword
2904 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2905 * A beta-quality port to the mips architecture running Linux,
2906 based on the old CMUCL backend, has been made. It has been tested
2907 on a big-endian kernel, and works sufficiently well to be able to
2908 rebuild itself; it has not been tested in little-endian mode.
2909 * fixed an inconsistency between gencgc.c and purify.c which made
2910 dumping/loading .core files unreliable
2911 * fixed bug 120a: The compiler now deals correctly with IFs where
2912 the consequent is the same as the alternative, instead of
2913 misderiving the return type. (thanks to Alexey Dejneka)
2914 * fixed bug 113: Logical pathnames are now dumpable (the logical
2915 host is resolved at load-time, throwing an error if it is not
2917 * fixed bug 174: FORMAT's error message is slightly clearer when a
2918 non-printing character is used in a format directive.
2919 * fixed several bugs in compiler checking of type declarations, i.e.
2920 violations of the Python "declarations are assertions" principle
2921 (thanks to Alexey Dejneka)
2922 * fixed several bugs in PCL's error checking (thanks to Gerd
2924 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2926 * fixed bug in compilation of functions as first class values
2927 (thanks to Antonio Martinez-Shotton)
2928 * The compiler's handling TYPE-ERRORs which it can prove will
2929 inevitably happen at runtime has been cleaned up and corrected
2930 in several ways. (thanks to Alexey Dejneka)
2931 * improved argument type checking for various basic arithmetic
2932 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2933 lavished on them in the past that they can be compiled in many
2934 ways in different special cases
2935 * fixed bug 181: compiler checks validity of user supplied type
2937 * cleaned up code flushing in optimization: Function calls which
2938 should signal errors for safety purposes (e.g. which ANSI says
2939 should signal errors when their arguments are of incorrect type)
2940 are no longer optimized away.
2941 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2942 * incremented fasl file version number, because changes in the
2943 implementation of sequence functions like COERCE caused
2944 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2945 in old inline expansions) to become undefined. (Actually these
2946 changes were later undone, so we might very well be binary
2947 compatible with 0.7.7 after all, but leaving the version number
2948 incremented seemed like the simplest and most conservative
2951 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2952 * An alpha-quality port to the parisc architecture running Linux,
2953 based on the old CMUCL backend, has been made. This, even more so
2954 than the other backends, should be considered still a work in
2955 progress; known problems include that the Linux kernel in 64-bit
2956 mode does not propagate the correct sigcontext structure to
2957 userspace, and consequently SBCL on a parisc64 kernel will not
2959 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2960 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2961 suppressed.) Also now that the compiler is looking at declarations
2962 in the environment, it checks optimization declarations as well,
2963 and suppresses inlining when (> DEBUG SPEED).
2964 * More fixes have been made to treatment of floating point exception
2965 treatment and other Unix signals. In particular, floating point
2966 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2967 * The detection and handling of control stack exhaustion (infinite
2968 or very deeply nested recursion) has changed. Stack exhaustion
2969 detection is now done by write-protecting pages at the OS level
2970 and applies at all optimization settings; when found, a
2971 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2972 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2973 be caught using IGNORE-ERRORS.
2974 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2975 correctly when transforming recursive LABELS functions to LETs.
2976 (thanks to Alexey Dejneka)
2977 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2978 that are names of constants or global variables.
2979 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2980 alien routines with docstrings.
2981 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2982 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2984 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2985 (thanks to Eric Marsden)
2986 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2987 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2988 * Bug fix: The SPARC backend can now compile functions involving
2989 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2990 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2991 to a functions expecting a PATHNAME-DESIGNATOR.
2992 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2993 lambda lists. (thanks to Alexey Dejneka)
2994 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2995 functions so that EXPORTing the name of the class doesn't cause
2996 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2997 to Antonio Martinez for reporting this)
2998 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2999 constant-folded without failing an assertion. (thanks to Einar
3000 Floystad Dorum for reporting this)
3001 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3002 types are now dealt with more correctly by the compiler.
3003 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3004 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3005 behaviour analogous to RENAME-FILE. This puts its behaviour more
3006 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3008 * The fasl file version number has changed again. (because of the
3009 bug fix involving the names of PCL MAKE-INSTANCE functions)
3011 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3012 * bug fix: Floating point exceptions are treated much more
3013 consistently on the x86/Linux and PPC/Linux platforms.
3014 * Array initialization with :INITIAL-ELEMENT is now much faster for
3015 cases when the compiler cannot open code the array creation, but
3016 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3017 array accesses have also seen a speed increase.
3018 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3019 specified. (This was at the root of some bad interactions between
3020 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3022 * bug fix: Internal error arguments for undefined functions are now
3023 computed correctly on the PPC/Linux platform.
3024 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3025 Raymond Toy's patch for CMU CL)
3026 * Support for the Solaris 9 operating environment has been included
3027 (thanks to Daniel Merritt)
3028 * A very ugly but hopefully complete draft of the missing FFI chapter
3029 of the manual has been created by reformatting the corresponding
3030 CMU CL manual chapter into (currently very ugly and incoherent)
3031 DocBook and bringing it up to date for SBCL behavior. Thus, the
3032 manual is now essentially complete, at least by my extreme
3033 once-and-only-once standards, whereby it's acceptable to refer to
3034 the doc strings of SB-EXT functions as the primary documentation.
3035 * The fasl file version number has changed again, due to cleanup of
3036 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3037 is no longer a static symbol.)
3039 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3040 * SBCL now builds with OpenMCL (version 0.12) as the
3041 cross-compilation host; also, more progress has been made toward
3042 bootstrapping under CLISP.
3043 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3045 * bug 158 fixed: The compiler can now deal with integer loop
3046 increments different from 1; fixing this turned out also to fix
3048 * bug 169 fixed: no more bogus warnings about using lexical bindings
3049 despite the presence of perfectly good SPECIAL declarations (thanks
3050 to David Lichteblau)
3051 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3052 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3053 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3055 * bug 180 fixed: Method combination specifications no longer ignore
3056 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3057 * bug fix: Structure type predicate functions now check their argument
3058 count as they should.
3059 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3060 correctly. (thanks to Pierre Mai)
3061 * minor incompatible change: The --noprogrammer option is deprecated
3062 in favor of the new --disable-debugger option, which is very similar.
3063 (The major difference is that it takes effect at a slightly different
3064 time at startup, causing handling of errors in --sysinit and
3065 --userinit files will be affected differently.) The
3066 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3067 been added to allow this functionality to be controlled from ordinary
3068 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3069 maintainers, who might want to run non-interactive scripts to
3070 build SBCL cores which will later be used interactively.)
3071 * minor incompatible change: The LOAD function no longer, when given
3072 a wild pathname to load, loads all files matching that pathname.
3073 Instead, an error of type FILE-ERROR is signalled.
3075 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3076 * bug 147 fixed: The compiler preserves its block link/count
3077 invariants more correctly now so that it doesn't crash. (thanks
3079 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3081 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3082 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3083 no-value-for-this-character as NIL (as specified by ANSI).
3084 * HOST-NAMESTRING on physical pathnames now returns a string that is
3085 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3086 (thanks to Christophe Rhodes)
3087 * The Alpha port handles icache flushing more correctly. (thanks to
3089 * More progress has been made toward bootstrapping under CLISP. (thanks
3090 to Christophe Rhodes)
3091 * The fasl file format has changed again, because dynamic loading
3092 on OpenBSD (which has non-ELF object files) motivated some cleanups
3093 in the way that foreign symbols are transformed and passed around.
3094 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3095 is no longer treated as whitespace by the reader, but instead as
3096 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3097 |A
\7fB|, instead of A as it used to.
3099 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3100 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3101 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3102 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3103 this as of 0.7.1.45, but was left out of the previous news section
3104 (thanks to Dan Barlow)
3105 * SBCL now runs on the Solaris operating system on SPARC architectures
3106 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3107 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3108 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3109 * SPARC backend cleanups, allowing builds of cores optimized for V8
3110 and V9 SPARCS, and also emission of code targeted to a particular
3111 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3113 * SBCL is closer to bootstrapping under CLISP, thanks to various
3114 fixes by Christophe Rhodes.
3115 * The fasl file format has changed again, to allow the compiler's
3116 INFO database to support symbol macros.
3117 * The user manual (in doc/) is formatted into HTML more nicely.
3118 (thanks to coreythomas)
3119 * The system is smarter about SUBTYPEP relationships, especially
3120 those involving NOT types (including types such as ATOM which are
3121 represented internally using NOT types). Thus SUBTYPEP is less
3122 likely to return (VALUES NIL NIL) in general, and in particular
3123 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3125 * The fasl file format has changed again, because the internal
3126 representation of types now includes a new slot to support the new
3127 SUBTYPEP-of-NOT-types logic.
3128 * (not a change in the main branch of SBCL, but a related prototype
3129 which can hopefully be merged into the main branch of SBCL in the
3130 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3131 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3132 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3133 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3134 dispatch macro character combinations. (thanks to Alexey Dejneka)
3135 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3136 that superficially look like logical namestrings correctly.
3137 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3139 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3140 default supplied-p) for &optional and &key arguments. (thanks to
3143 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3144 * incompatible change: The compiler is now less aggressive about
3145 tail call optimization, doing it only when (> SPACE DEBUG) or
3146 (> SPEED DEBUG). (This is an incompatible change because there are
3147 programs which relied on the old CMU-CL-style behavior to optimize
3148 away their unbounded recursion which will now die of stack overflow.)
3149 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3150 for non-GENCGC systems has been increased to 20M (since that
3151 seems much closer to the likely performance optimum for modern
3152 systems than the old 4M value was)
3153 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3154 and *DEBUG-PRINT-LEVEL*
3155 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3156 of CMU CL's support for SPARC, and various endianness and other
3157 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3158 * new syntactic sugar for the Unix command line: --load foo.bar is now
3159 an alternate notation for --eval '(load "foo.bar")'.
3161 ** The system now detects stack overflow and handles it gracefully,
3162 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3163 optimization settings. (This is a good thing in general, and
3164 its introduction in this version should be particularly timely
3165 for anyone whose code fails because of suppression of tail
3167 ** The system now hunts for the C variable "environ" in a more
3168 devious way, to avoid segfaults when the C library version differs
3169 between compile time and run time. (thanks to Christophe Rhodes)
3170 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3171 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3172 ** The compiler no longer issues bogus style warnings for undefined
3173 classes in the same source file as the DEFCLASSes which defined
3174 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3176 ** fixes in CONDITION class precedence list for undefined function
3177 errors (thanks to Alexei Dejneka)
3178 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3179 correctly. (thanks to Dan Barlow)
3180 ** portability fixes aiming at bootstrapping under CLISP (thanks
3181 to Dave McDonald and Christophe Rhodes)
3182 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3183 ** fixes in type translation and and type inference (thanks to
3185 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3186 ** various fixes in the new ports (thanks to Dan Barlow)
3187 * several changes related to debugging:
3188 ** suppression of tail recursion, as noted above
3189 ** stack overflow detection, as noted above
3190 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3191 is now the default. (For some time encapsulation has been more
3192 reliable than the breakpoint-based :ENCAPSULATE NIL
3193 implementation, at least on X86 systems; and I just noticed that
3194 encapsulation also seems closer to the spirit of the ANSI
3197 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3199 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3200 up properly again. (There was a packaging bug in 0.7.0 which
3201 left their definitions in SB-SYS::LOAD-FOREIGN and
3202 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3203 vital for most things which interface to C-level interfaces,
3204 like extensions working with sockets or databases or
3205 Perl-compatible regexes or whatever, and the need to fix
3206 this bug is the main reason that 0.7.1 was released so
3208 ** DEFGENERIC is now choosier about the methods it redefines, so that
3209 reLOADing a previously-LOADed file containing DEFGENERICs does
3210 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3211 cycle now works as it should. (thanks to Alexey Dejneka)
3212 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3213 fixed by Christophe Rhodes. (He actually submitted this patch
3214 months ago, and I delayed until after 0.7.0.)
3215 ** Bug 111 (internal compiler confusion about runtime checks on
3216 FUNCTION types) was fixed by Alexey Dejneka.
3217 * Some internal cleanups (getting rid of variables which aren't
3218 needed now that the byte interpreter is gone) caused the fasl
3219 file format number to change again.
3221 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3222 * major incompatible change: The default fasl file extension, i.e. the
3223 default extension for files produced by COMPILE-FILE, has changed
3224 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3226 ** There are many changes in the implementation of the compiler.
3227 SBCL is now essentially a compiler-only implementation of ANSI
3228 Common Lisp. EVAL still "interprets" a few special cases, but
3229 almost all the interesting cases are handled by creating
3230 a LAMBDA expression, calling COMPILE on it, then calling
3231 FUNCALL on the result.
3232 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3233 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3234 Since the code is newer, there might still be some new bugs
3235 (though not as many as before Martin Atzmueller's fixes:-). But
3236 the new code is substantially simpler and clearer, and hopefully
3237 any remaining bugs will be simpler, less fundamental, and more
3238 fixable then the bugs in the old code.
3239 ** The revised compiler is still a little unsteady on its feet.
3241 *** The debugging information it produces (particularly the names
3242 of FUNCTION objects) is sometimes much less useful than what
3243 the old compiler produced.
3244 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3245 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3246 within a MACROLET) has been temporarily weakened.
3247 ** There are new compiler optimizations for various functions:
3248 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3249 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3250 *** the math functions TRUNCATE, FLOOR, and CEILING
3251 *** the function-of-all-trades COERCE
3252 Mostly these should be transparent, but there's one
3253 potentially-annoying problem (bug 117): when the compiler
3254 inline-expands a function and does type analysis on the result,
3255 it can create control paths which have type mismatches, and
3256 when it can't prove that those control paths aren't taken,
3257 it will issue WARNINGs about the type mismatches. This is
3258 a particular problem in practice for the new sequence functions.
3259 It's not clear how this should be fixed, and for now, a
3260 workaround is given in the entry for 117 in the BUGS file.
3261 ** (Because of the interaction between the two previous items --
3262 occasional inlining problems and new inline expansions -- some
3263 of the new sequence function optimizations won't really kick in
3264 completely until debugging information, and then inlining, are
3265 straightened out in some future version.)
3266 * minor incompatible changes:
3267 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3268 DIRECTORY no longer implicitly promotes NIL slots of its
3269 pathname argument to :WILD. In particular, when you ask for the
3270 contents of a directory (which you used to be able to do without
3271 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3272 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3273 ** changes in behavior that ANSI explicitly defines to be
3274 implementation dependent:
3275 *** The new compiler-only implementation still conforms with ANSI,
3276 but acts a little different than before. Besides the obvious
3277 changes in performance tradeoffs (that the cost per form passed
3278 to EVAL has gone up, and the cost per form executed by EVAL
3279 has gone down), the behavior of the system changes a little
3280 because there are no longer any interpreted function objects.
3281 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3282 e.g. doing COMPILE on the output of interactive DEFUN is
3284 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3286 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3287 has changed from (:CL) to NIL.
3288 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3289 e.g. #\Newline and #\Tab, don't have names specified in the
3290 ANSI Common Lisp standard, is now based on their ASCII symbolic
3291 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3292 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3293 are no longer used for output.
3294 ** changes in internal implementation constants:
3295 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3296 4 million. (If your application spends a lot of time GCing and
3297 you have a lot of RAM, you might want to experiment with
3298 increasing it even more.)
3299 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3300 However, almost all old code should still continue to work without
3301 immediate update, as SB-C-CALL is now a (deprecated) nickname
3303 ** Old operator names in the style DEF-FOO are now deprecated in
3304 favor of new corresponding names DEFINE-FOO, for consistency with
3305 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3306 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3307 internal symbols, but a few supported extensions like
3308 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3309 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3310 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3311 etc. as you get deeper into recursive calls to the debugger
3312 command loop, instead of the old "5]", "5]]", "5]]]"
3313 sequence. (I was motivated to do this when squabbles between
3314 ILISP and SBCL left me very deeply nested in the debugger. In the
3315 short term, this change will probably provoke more ILISP/SBCL
3316 squabbles, but hopefully it will be an improvement in the long run.)
3317 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3318 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3319 function, since declaiming return types involving aliens is
3320 (1) annoyingly messy to do by hand and (2) vital to efficient
3321 compilation of code which calls such functions.
3322 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3323 longer reexported by the SB-EXT package. They're solely useful
3324 for alien code, so it seems more logical that you should get
3325 them from the SB-ALIEN package, not in SB-EXT.
3326 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3327 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3328 features. Instead, the code that they used to control is always
3329 built into the system.
3330 * many other bug fixes
3331 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3332 advantage of the new EVAL-WHEN stuff and to clean them up in
3333 general, and they are now more ANSI-compliant in a number of
3334 ways. Martin Atzmueller is responsible for a lot of this.
3335 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3337 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3338 *** correct ERROR type for various file operations
3339 *** some fixes for Lisp streams
3340 *** DEFMETHOD syntax checking
3341 *** changing old weird representation of debug information as
3342 strings (which, among their other deficiencies, don't transform
3343 correctly when you rename packages, and don't change their
3344 print representation when you change things like *PACKAGE*
3345 and *PRINT-LENGTH*) to symbols and lists of symbols
3346 He also made several improvements and fixed several bugs in DESCRIBE.
3347 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3349 *** misbehavior of WRITE-STRING/WRITE-LINE
3350 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3351 and several other LOOP problems as well
3352 *** DIRECTORY when similar filenames are present
3353 *** DEFGENERIC with :METHOD options
3354 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3355 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3356 *** argument ordering in FIND with :TEST option
3357 *** mishandled package designator argument in APROPOS-LIST
3358 *** various problems in the backquote readmacro
3359 *** a bug in APROPOS
3360 *** probably some others that I'm not describing very well here,
3361 since the CVS log documents them by reference to sbcl-devel
3362 messages, and the SourceForge archives aren't working well.:-(
3363 ** Dan Barlow improved the Alpha port (and is making progress on the
3364 PPC port, for those of you who think different).
3365 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3366 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3367 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3368 continues to work on the SPARC port (for those of you in a position
3369 to look down upon our little PC-compatible boxes from a great height).
3370 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3371 and uses that copy, rather than the current dynamic value, when
3372 it's trying to decide whether to truncate output. Thus e.g.
3373 (let ((*print-lines* 50))
3374 (pprint-logical-block (stream nil)
3376 (let ((*print-lines* 8))
3377 (print (aref possiblybigthings i) stream)))))
3378 should now truncate the logical block only at 50 lines, instead of
3379 often truncating it at 8 lines, as it did before.
3380 * The doc/cmucl/ directory, containing old CMU CL documentation
3381 from the time of the fork, is no longer part of the base system.
3382 SourceForge has shut down its anonymous FTP service, and with it
3383 my original plan for distributing the old CMU CL documentation
3384 there. For now, if you need these files you can download an old
3385 SBCL source release and extract them from it.
3386 * The fasl file version number changed again, for dozens of reasons,
3387 some of which are apparent above.
3389 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3390 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3391 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3392 patch, so that DISASSEMBLE output is much nicer.
3393 * The code in the SB-PROFILE package now seems reasonably stable.
3394 I still haven't decided what the final interface should look like
3395 (I'd like PROFILE to interact cleanly with TRACE, since both
3396 facilities use function encapsulation) but if you have a need
3397 for profiling now, you can probably use it successfully with
3398 the current CMU-CL-style interface.
3399 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3400 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3401 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3402 link now returns the dangling link itself, and for similar
3403 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3404 link itself. (In these cases the old code signalled an error and
3405 looped endlessly, respectively.) Thus, DIRECTORY now works even
3406 in the presence of dangling and cyclic symbolic links.
3407 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3408 is now a supported extension again, since the consensus on
3409 sbcl-devel was that it can be useful for ordinary development
3410 work, not just for debugging SBCL itself.
3411 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3412 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3413 that currently-defined functions might be redefined later with
3414 different return types.
3415 * Hash tables can be printed readably, as inspired by CMU CL code
3416 of Eric Marsden and SBCL code of Martin Atzmueller.
3417 * better error handling in CLOS method combination, thanks to
3418 Martin Atzmueller porting Pierre Mai's CMU CL patches
3419 * more overflow fixes for >16Mbyte I/O buffers
3420 * A bug in READ has been fixed, so that now a single Ctrl-D
3421 character suffices to cause end-of-file on character streams.
3422 In particular, now you only need one Ctrl-D at the command
3423 line (not two) to exit SBCL.
3424 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3426 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3427 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3428 does the right thing.
3429 * The implementation of some type tests, especially for CONDITION
3430 types, is now tidier and maybe faster, due to CMU CL code
3431 originally by Douglas Crosher, ported by Martin Atzmueller.
3432 * Some math functions have been fixed, and there are new
3433 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3434 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3435 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3436 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3437 likely to be built into sbcl-0.7.0.)
3438 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3439 CL, ported to SBCL by Martin Atzmueller)
3440 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3441 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3442 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3443 * There's a new slam.sh hack to shorten the edit/compile/debug
3444 cycle for low-level changes to SBCL itself, and a new
3445 :SB-AFTER-XC-CORE target feature to control the generation of
3446 the after-xc.core file needed by slam.sh.
3447 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3448 COMPILE-FILE is no longer supported, so that now every function
3449 gets an entry point, so that block compilation looks a little
3450 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3451 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3452 returns the number of bytes consed since the system started,
3453 rather than the number consed since the first time the function
3454 was called. (The new definition parallels ANSI functions like
3455 CL:GET-INTERNAL-RUN-TIME.)
3456 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3457 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3458 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3459 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3460 * Fasl file version numbers are now independent of the target CPU,
3461 since historically most system changes which required version
3462 number changes have affected all CPUs equally. Similarly,
3463 the byte fasl file version is now equal to the ordinary
3466 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3467 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3468 is no longer recognized. I apologize for this, because it was
3469 listed in SB-EXT as a supported extension, but I found that
3470 its existing behavior was poorly specified, as well as incorrectly
3471 specified, and it looked like too much of a mess to straighten it
3472 out. I have enough on my hands trying to get ANSI stuff to work..
3473 * many patches ported from CMU CL by Martin Atzmueller, with
3474 half a dozen bug fixes in pretty-printing and the debugger, and
3475 half a dozen others elsewhere
3476 * fixed bug 13: Floating point infinities are now supported again.
3477 They might still be a little bit flaky, but thanks to bug reports
3478 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3479 as flaky as they were.
3480 * The --noprogrammer command line option is now supported. (Its
3481 behavior is slightly different in detail from what the old man
3482 page claimed it would do, but it's still appropriate under the
3483 same circumstances that the man page talks about.)
3484 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3485 are now supported, and enabled by default. Thus, the compiler can
3486 handle many floating point and complex operations much less
3487 inefficiently. (Thus e.g. you can implement a complex FFT
3489 * The compiler now detects type mismatches between DECLAIM FTYPE
3490 and DEFUN better, and implements CHECK-TYPE more correctly, and
3491 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3492 thanks to patches from Martin Atzmueller.
3493 * various fixes to make the cross-compiler more portable to
3494 ANSI-conforming-but-different cross-compilation hosts (notably
3495 Lispworks for Windows, following bug reports from Arthur Lemmens)
3496 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3497 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3498 fixed thanks to Pierre Mai's CMU CL patches.
3499 * fixes to keep the system from overflowing internal counters when
3500 it tries to use i/o buffers larger than 16M bytes
3501 * fixed bug 45a: Various internal functions required to support
3502 complex special functions have been merged from CMU CL sources.
3503 (When I was first setting up SBCL, I misunderstood a compile-time
3504 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3505 * improved support for type intersection and union, fixing bug 12
3506 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3507 more obscure bugs as well
3508 * some steps toward byte-compiling non-performance-critical
3509 parts of the system, courtesy of patches from Martin Atzmueller
3510 * Christophe Rhodes has made some debian packages of sbcl at
3511 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3512 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3513 stable, but are nonetheless usable. When he's ready, I'd be happy
3514 to add them to the SourceForge "File Releases" section. (And if
3515 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3516 * new fasl file format version number (because of changes in
3517 internal representation of (OR ..) types to accommodate the new
3518 support for (AND ..) types, among other things)
3520 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3521 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3523 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3525 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3526 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3527 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3528 * DESCRIBE now works on structure objects again.
3529 * Most function call argument type mismatches are now handled as
3530 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3531 know whether the function will be redefined before the call is
3532 executed. (The compiler could flag local calls with full WARNINGs,
3533 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3534 it doesn't keep track of enough information to know whether calls
3535 are local in this sense.)
3536 * Compiler output is now more verbose, with messages truncated
3537 later than before. (There should be some supported way for users
3538 to override the default verbosity, but I haven't decided how to
3539 provide it yet, so this behavior is still controlled by the internal
3540 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3541 src/compiler/ir1util.lisp.)
3542 * Fasl file format version numbers have increased again, because
3543 support for the Gray streams extension changes the layout of the
3544 system's STREAM objects.
3545 * The Gray subclassable streams extension now works, thanks to a
3546 patch from Martin Atzmueller.
3547 * The full LOAD-FOREIGN extension (not just the primitive
3548 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3549 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3550 but like most other programs, it defaults to copying the Unix
3551 environment from the original process instead of starting the
3552 new process in an empty environment.
3553 * Extensions which manipulate the Unix environment now support
3554 an :ENVIRONMENT keyword option which doesn't smash case or
3555 do other bad things. The CMU-CL-style :ENV option is retained
3556 for porting convenience.
3557 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3558 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3560 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3562 * A patch from Martin Atzmueller seems to have solved the SIGINT
3563 problem, and as far as we know, signal-handling now works cleanly.
3564 (If you find any new bugs, please report them!)
3565 * The system no longer defaults Lisp source file names to types
3566 ".l", ".cl", or ".lsp", but only to ".lisp".
3567 * The compiler no longer uses special default file extensions for
3568 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3569 seems to expect a single default extension for all compiled code,
3570 and there's no compelling reason to try to stretch the standard
3571 to allow two different extensions.) Instead, byte-compiled files
3572 default to the same extension as native-compiled files.
3573 * Fasl file format version numbers have increased again, because
3574 a rearrangement of internal implementation packages made some
3575 dumped symbols in old fasl files unreadable in new cores.
3576 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3577 many fewer weird special cases.
3578 * Bug #17 (differing COMPILE-FILE behavior between logical and
3579 physical pathnames) has been fixed, and some related misbehavior too,
3580 thanks to a patch from Martin Atzmueller.
3581 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3582 by Tim Moore, ported to SBCL by Martin Atzmueller.
3583 * Martin Atzmueller fixed several filesystem-related problems,
3584 including bug #36, in part by porting CMU CL patches, which were
3585 written in part by Paul Werkowski.
3586 * More compiler warnings in src/runtime/ are gone, thanks to
3587 more patches from Martin Atzmueller.
3588 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3591 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3593 * DESCRIBE now works on CONDITION objects.
3594 * The debugger now handles errors which arise when trying to print
3595 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3597 * The build system now uses an additional file, customize-target-features.lisp,
3598 to allow local modifications to the target *FEATURES* list. (The point of
3599 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3600 debugging features enabled, without having to worry about propagating it
3601 into everyone's system when I do a "cvs update".) When no
3602 customize-target-features.lisp file exists, the target *FEATURES* list
3603 should be constructed the same way as before.
3604 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3605 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3606 warn about behavior which is undefined under the ANSI spec. Note: This
3607 is specified by ANSI, but it's not very popular with programmers.
3608 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3609 macro in the SBCL sources for an example of a workaround which you
3610 might use to make portable ANSI-standard code which does what you want.
3611 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3612 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3613 when it's used as a non-toplevel form. (This is required in order
3614 to implement the DEFCONSTANT-EQX macro.)
3615 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3616 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3617 funny things, due to bugs in the implementation of EVAL-WHEN
3618 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3619 non-toplevel DEFCONSTANTs are uncommon.)
3620 * The core file version number and fasl file version number have been
3621 incremented, because the old noncompliant DEFCONSTANT behavior involved
3622 calling functions which no longer exist, and because I also took the
3623 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3624 * fixed bug 1 (error handling before read-eval-print loop starts), and
3625 redid debugger restarts and related debugger commands somewhat while
3627 ** The QUIT debugger command is gone, since it did something
3628 rather different than the SB-EXT:QUIT command, and since it never
3629 worked properly outside the main toplevel read/eval/print loop.
3630 Invoking the new TOPLEVEL restart provides the same functionality.
3631 ** The GO debugger command is also gone, since you can just invoke
3632 the CONTINUE restart directly instead.
3633 ** The TOP debugger command is also gone, since it's redundant with the
3634 FRAME 0 command, and since it interfered with abbreviations for the
3636 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3638 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3639 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3640 some porting work by Martin Atzmueller)
3641 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3642 with a boatload of patches which he ported to SBCL. Now that those
3644 ** The system tries to make sure that its low-priority messages
3645 are prefixed by semicolons, to help people who like to use
3646 syntax highlighting in their ILISP buffer. (This patch
3647 was originally due to Raymond Toy.)
3648 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3649 patches originally written by Raymond Toy.
3650 ** The compiler understands coercion between single-value and
3651 multiple-VALUES type expressions better, getting rid of some very
3652 weird behavior, thanks to patches originally by Robert MacLachlan
3653 and Douglas Crosher.
3654 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3655 lambda lists, thanks to a patch originally by Pierre Mai.
3656 ** The system no longer bogusly warns about "abbreviated type
3658 ** The compiler gets less confused by inlining and RETURN-FROM,
3659 thanks to some patches originally by Tim Moore.
3660 ** The system no longer hangs when dumping circular lists to fasl
3661 files, thanks to a patch originally from Douglas Crosher.
3662 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3663 undefined function error.
3664 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3665 in structure slot initforms) for now, documented workaround instead:-|
3666 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3667 * fixed bug 5: added stubs for various Gray stream functions called
3668 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3669 installed, at least appropriate type errors are generated
3670 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3671 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3672 consistently in DEFMETHOD forms.
3673 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3674 it doesn't seem to affect SBCL after all
3675 * The C runtime system now builds with better optimization and many
3676 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3678 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3680 * The system is now under CVS at SourceForge (instead of the
3681 CVS repository on my home machine).
3682 * The new signal handling code has been tweaked to treat register
3683 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3684 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3685 downstream, e.g. in debug-int.lisp, has implicit dependencies
3686 on the unsignedness of integer representation of machine words,
3687 and that caused the system to bomb out with infinite regress
3688 when trying to recover from type errors involving signed values,
3689 e.g. (BUTLAST '(1 2 3) -1).)
3690 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3691 (This was one of the bugs Peter Van Eynde reported back in July.)
3692 * The system now uses code inspired by Colin Walters' O(N)
3693 implementation of MAP (from the cmucl-imp@cons.org mailing
3694 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3695 inline the MAP operation, and there is more than one
3696 sequence argument to the MAP call (so that it can't just
3697 do ETYPECASE once and for all based on the type of the
3698 single sequence argument). (The old non-inline implementation
3699 of the general M-argument sequence-of-length-N case required
3700 O(M*N*N) time when any of the sequence arguments were LISTs.)
3701 * The QUIT :UNIX-CODE keyword argument has been renamed to
3702 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3703 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3705 * Raymond Wiker's patches to port dynamic loading from Linux to
3706 FreeBSD have been added.
3707 * The BUGS file is now more nearly up to date, thanks in large part
3708 to Martin Atzmueller's review of it.
3709 * The debugger now flushes standard output streams before it begins
3710 its output ("debugger invoked" and so forth).
3711 * The core version number and fasl file version number have both
3712 been incremented, because of incompatible changes in the layout
3714 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3715 used to not be called for a saved Lisp image.)
3716 * Martin Atzmueller's version of a patch to fix a compiler crash,
3717 as posted on sbcl-devel 13 September 2000, has been installed.
3718 * Instead of installing Martin Atzmueller's patch for the
3719 compiler transform for SUBSEQ, I deleted the compiler transform,
3720 and transforms for some similar consing operations.
3721 * A bug in signal handling which kept TRACE from working on OpenBSD
3723 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3724 away by constant folding
3725 * The system now defines its address space constants in one place
3726 (in the Lisp sources), and propagates them automatically elsewhere
3727 (through GENESIS and the sbcl.h file). Therefore, patching the
3728 address map is less unnecessarily tedious and error-prone. The
3729 Lisp names of address space constants have also been systematized.
3730 * CVS tags like dollar-Header-dollar have been removed from
3731 the sources, because they have never saved me trouble and
3732 they've been source of trouble working with patches and other
3733 diff-related operations.
3734 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3735 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3737 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3739 * The system has been ported to OpenBSD.
3740 * The system now compiles with a simple "sh make.sh" on the systems
3741 that it's supported on. I.e., now you no longer need to tweak
3742 text in the target-features.lisp-expr and symlinks in src/runtime/
3743 by hand, the make.sh takes care of it for you.
3744 * The system is no longer so grossly inefficient when compiling code
3745 involving vectors implemented as general (not simple) vectors (VECTOR T),
3746 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3747 which dares to use the various sequence functions on non-simple
3748 vectors, takes less of a performance hit.
3749 * There is now a primitive type predicate VECTOR-T-P
3750 to test for the (VECTOR T) type, so that e.g.
3751 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3752 can now be compiled with some semblance of efficiency. (The old code
3753 turned the type declaration into a full call to %TYPEP at runtime!)
3754 * AREF on (VECTOR T) is still not fast, since it's still compiled
3755 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3756 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3757 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3758 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3759 without making full calls to SUBTYPEP at runtime.
3760 (Some analogous efficiency issues for non-simple vectors specialized to
3761 element types other than T, or for non-simple multidimensional arrays,
3762 have not been addressed. They could almost certainly be handled the
3763 same way if anyone is motivated to do so.)
3764 * The changes in array handling break binary compatibility, so
3765 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3766 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3767 instead of (VALUES T T).
3768 * By following the instructions that Dan Barlow posted to sbcl-devel
3769 on 2 July 2000, I was able to enable primitive dynamic object
3770 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3771 functionality is not implemented (since it calls ld to resolve
3772 library references automatically, requiring RUN-PROGRAM for its
3773 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3774 try to resolve library references) is now supported.
3775 * The system now flushes the standard output streams when it terminates,
3776 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3777 them at several other probably-convenient times, e.g. in each pass of
3778 the toplevel read-eval-print loop, and after evaluating a form given
3779 as an "--eval" command-line option. (These changes were motivated by a
3780 discussion of stream flushing issues on cmucl-imp in August 2000.)
3781 * The source transform for TYPEP of array types no longer assumes
3782 that an array whose element type is a not-yet-defined type
3783 is implemented as an array of T, but instead punts, so that the
3784 type will be interpreted at runtime.
3785 * There is now some support for cross-compiling in make.sh: each of
3786 the phases of make.sh has its own script. (This should be transparent
3787 to people doing ordinary, non-cross-compile builds.)
3788 * Since my laptop doesn't have hundreds of megabytes of memory like
3789 my desktop machine, I became more motivated to do some items on
3790 my to-do list in order to reduce the size of the system a little:
3791 ** Arrange for various needed-only-at-cold-init things to be
3792 uninterned after cold init. To support this, those things have
3793 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3794 symbols with such names are now uninterned after cold init).
3795 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3796 fasl files for cold load.
3797 ** Remove the old compiler structure pooling code (which used to
3798 be conditional on the target feature :SB-ALLOC) completely.
3799 ** Redo the representation of some data in cold init to be more compact.
3800 (I also looked into supporting byte compiled code at bootstrap time,
3801 which would probably reduce the size of the system a lot, but that
3802 looked too complicated, so I punted for now.)
3803 * The maximum signal nesting depth in the src/runtime/ support code has
3804 been reduced from 4096 to 256. (I don't know any reason for the very
3805 large old value. If the new smaller value turns out to break something,
3806 I'll probably just bump it back up.)
3807 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3809 * Many, many bugs reported by Peter Van Eynde have been added to
3810 the BUGS list; some have even been fixed.
3811 * While enabling dynamic object file loading, I tried to make the
3812 code easier to understand, renaming various functions and variables
3813 with less ambiguous names, and changing some function calling
3814 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3815 * While trying to figure out how to do the OpenBSD port, I tried to
3816 clean up some of the code in src/runtime/. In particular, I dropped
3817 support for non-POSIX signal handling, added various comments,
3818 tweaked the code to reduce the number of compilation warnings, and
3819 renamed some files to increase consistency.
3820 * To support the new automatic configuration functionality in make.sh,
3821 the source file target-features.lisp-expr has been replaced with the
3822 source file base-target-features.lisp-expr and the machine-generated
3823 file local-target-features.lisp-expr.
3824 * fixed a stupid quoting error in make.sh so that using CMU CL
3825 "lisp -batch" as cross-compilation host works again
3827 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3829 * DESCRIBE no longer tries to call itself recursively to describe
3830 bound/fbound values, so that it no longer fails on symbols which are
3831 bound to themselves (like keywords, T, and NIL).
3832 * DESCRIBE now works on generic functions.
3833 * The printer now prints less-screwed-up representations of closures
3834 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3835 * A private symbol is used instead of the :EMPTY keyword previously
3836 used to mark empty slots in hash tables. Thus
3837 (DEFVAR *HT* (MAKE-HASH-TABLE))
3838 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3839 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3840 now does what ANSI says that it should. (You can still get
3841 similar noncompliant behavior if bang on the hash table
3842 implementation with all the symbols you get back from
3843 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3844 This breaks binary compatibility, since tests for equality to
3845 :EMPTY are wired into things like the macroexpansion of
3846 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3848 * There's now a minimal placeholder implementation for CL:STEP,
3849 as required by ANSI.
3850 * An obscure bug in the interaction of the normal compiler, the byte
3851 compiler, inlining, and structure predicates has been patched
3852 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3853 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3855 * Missing ordinary arguments in a macro call are now detected even
3856 when the macro lambda list contains &KEY or &REST.
3857 * The debugger no longer complains about encountering the top of the
3858 stack when you type "FRAME 0" to explicitly instruct it to go to
3859 the top of the stack. And it now prints the frame you request even
3860 if it's the current frame (instead of saying "You are here.").
3861 * As specified by ANSI, the system now always prints keywords
3862 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3864 * The default initial SIZE of HASH-TABLEs is now smaller.
3865 * Type information from CLOS class dispatch is now propagated
3866 into DEFMETHOD bodies, so that e.g.
3867 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3869 is now basically equivalent to
3870 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3871 (DECLARE (TYPE SINGLE-FLOAT X))
3873 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3874 operation, without having to do run-time type dispatch.
3875 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3876 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3877 * Since I don't seem to be making big file reorganizations very often
3878 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3879 is overflowing my ability to conveniently back them up), I've finally
3880 checked the system into CVS. (The CVS repository is on my home system,
3881 not at SourceForge -- putting it on SourceForge might come later.)
3882 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3883 high-level GC-NOTIFY-FOO functions send their output. (There's
3884 still very little control of where low-level verbose GC functions
3885 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3886 less than it used to -- the GC-NOTIFY-FOO functions are now under
3887 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3888 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3889 in only one place in the source code, and propagates it automatically
3890 everywhere that it's needed. Thus e.g. when I bump the version from
3891 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3892 * The C source files now include boilerplate legalese and documentation
3893 at the head of each file (just as the Lisp source files already did).
3894 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3895 to his page will be replaced with a link to his new CLiki service.
3897 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3899 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3900 * The build process now looks for GNU make under the default name "gmake",
3901 instead of "make" as it used to. If GNU make is not available as "gmake"
3902 on your system, you can change this default behavior by setting the
3903 GNUMAKE environment variable.
3904 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3905 can build without error under CMU CL.
3907 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3909 * There is now a partial SBCL user manual (with some new text and some
3910 text cribbed from the CMU CL manual).
3911 * The beginnings of a profiler have been added (starting with the
3912 CMU CL profiler and simplifying and cleaning up). Eventually the
3913 main interface should be through the TRACE macro, but for now,
3914 it's still accessed through vaguely CMU-CL-style functions and macros
3915 exported from the package SB-PROFILE.
3916 * Some problems left over from porting CMU CL to the new
3917 cross-compilation bootstrap process have been cleaned up:
3918 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3919 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3920 part when getting rid of the compiler *BACKEND* stuff.)
3921 ** The value of *NULL-TYPE* was screwed up, because it was
3922 being initialized before the type system knew the final
3923 definition of the 'NULL type. This screwed up several key
3924 optimizations in the compiler, causing inefficiency in all sorts
3925 of places. (I found it because I wanted to understand why
3926 GET-INTERNAL-RUN-TIME was consing.)
3927 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3928 PROCLAIM FTYPE information. Unfortunately this broke binary
3929 compatibility again, since now the forms output by DEFGENERIC
3930 to refer to functions which didn't exist in 0.6.3.
3931 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3932 can use the new (as of 0.6.3) transform for SOME into MAP into
3934 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3935 values to (INTEGER 0 1000000), so that the time code will no longer
3936 occasionally get blown up by Linux returning 1000000 microseconds
3937 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3938 its output conform to the ANSI spec. (Alas, this makes its output
3939 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3940 * A full call to MAP NIL with a single sequence argument no longer conses.
3941 * fixes to problems pointed out by Martin Atzmueller:
3942 * The manual page no longer talks about multiprocessing as though
3943 it were currently supported.
3944 * The ILISP support patches have been removed from the distribution,
3945 because as of version 5.10.1, ILISP now supports SBCL without us
3946 having to maintain patches.
3947 * added a modified version of Raymond Toy's recent CMU CL patch for
3948 EQUALP comparison of HASH-TABLE
3950 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3952 * The system still can't cross-compile itself with
3953 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3954 entails), but at least it can compile more of itself that way
3955 than it used to be able to, and various buglets which were uncovered
3956 by trying to cross-compile itself that way have now been fixed.
3957 * This release breaks binary compatibility again. This time
3958 at least I've incremented the FASL file format version to 2, so that the
3959 problem can be detected reliably instead of just causing weird errors.
3960 * various new style warnings:
3961 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3962 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3963 ** using the lexical binding of a variable named in the *FOO* style
3964 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3965 as specified by ANSI.
3966 * *RANDOM-STATE* is no longer automatically initialized from
3967 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3968 default behavior of the system is to repeat its behavior every time
3969 it's run. If you'd like to change this behavior, you can always
3970 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3971 old convention there was no comparably easy way to get the system to
3972 repeat its behavior every time it was run.
3973 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3974 FTYPE declarations has been removed, in favor of their ANSI
3975 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3976 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3977 the types of their sequence arguments can be determined at compile time.
3978 This is done through a new open code expansion for MAP which eliminates
3979 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3980 argument types can be determined at compile time.
3981 * The optimizer now transforms COERCE into an identity operation when it
3982 can prove that the coerced object is already of the correct type. (This
3983 can be a win for machine generated code, including the output of other
3984 optimization transforms, such as the MAP transform above.)
3985 * Credit information has been moved from source file headers into CREDITS.
3986 * Source file headers have been made more standard.
3987 * The CASE macro now compiles without complaining even when it has
3990 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3992 * (Note that the way that the PCL macroexpansions were rewritten
3993 to accommodate the change in DEFGENERIC below breaks binary
3994 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3995 not run under sbcl-0.6.2. Once we get out of alpha releases,
3996 i.e. hit release 1.0.0, we'll probably try to maintain binary
3997 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3998 and sbcl-1.4.4. Until then, however, it might be fairly common
3999 for maintenance releases to break binary compatibility.)
4000 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4002 * The reporting of unhandled signals has been changed to print some
4003 explanatory text as well as the report form. (Previously only
4004 the report form was printed.)
4005 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4006 it defines, so that the compiler no longer issues undefined function
4007 warnings for compiled-but-not-yet-loaded generic functions.
4008 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4009 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4010 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4011 section of the ANSI spec are supported.
4012 * The "" nickname for the "KEYWORD" package has been removed.
4013 The reader still handles symbol tokens which begin with a package marker
4014 as keywords, but it doesn't expose its mechanism for doing so in the
4015 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4016 * The system now issues STYLE-WARNINGs for contradictory TYPE
4017 proclamations. (Warnings for contradictory FTYPE proclamations would
4018 be nice too, but those can't be done usefully unless the type system
4019 is made smarter about FUNCTION types.)
4020 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4021 systematized, so that "*target-*.lisp is supposed to exist only on the
4022 target and imply that there's a related file which exists on the
4023 host, and *host-*.lisp is supposed to exist only on the host and imply
4024 that there's a related file which exists on the target. This involves a
4025 lot of renaming. Hopefully the acute confusion caused by the renaming
4026 will be justified by the reduction in chronic confusion..
4027 ** runtime-type.lisp -> early-target-type.lisp
4028 ** target-type.lisp -> late-target-type.lisp
4029 ** early-host-format.lisp -> early-format.lisp
4030 ** late-host-format.lisp -> late-format.lisp
4031 ** host-error.lisp -> misc-error.lisp
4032 ** early-error.lisp -> early-target-error.lisp
4033 ** late-error.lisp -> late-target-error.lisp
4034 ** host-defboot.lisp -> early-defboot.lisp
4035 ** code/misc.lisp -> code/target-misc.lisp
4036 ** code/host-misc.lisp -> code/misc.lisp
4037 ** code/numbers.lisp -> code/target-numbers.lisp
4038 ** code/early-numbers.lisp -> numbers.lisp
4039 ** early-host-type.lisp -> early-type.lisp
4040 ** late-host-type.lisp -> late-type.lisp
4041 ** host-typep.lisp -> typep.lisp
4042 ** load.lisp -> target-load.lisp
4043 ** host-load.lisp -> load.lisp
4044 ** host-disassem.lisp -> disassem.lisp
4045 ** host-insts.lisp -> insts.lisp
4046 ** byte-comp.lisp -> target-byte-comp.lisp
4047 ** host-byte-comp.lisp -> byte-comp.lisp
4048 ** host-signal.lisp -> signal.lisp
4049 ** host-defstruct.lisp -> defstruct.lisp
4050 ** late-target-type.lisp -> deftypes-for-target.lisp
4051 Furthermore, several other previously target-only files foo.lisp (e.g.
4052 hash-table.lisp and random.lisp) have been split into a target-and-host
4053 foo.lisp file and a target-only target-foo.lisp file, with their key type
4054 definitions in the target-and-host part, so that the cross-compiler will
4055 know more about target types.
4056 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4057 gone away. In their place are various *BACKEND-FOO* variables
4058 corresponding to the slots of the old structure.
4059 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4060 into the target SBCL has been fixed.
4061 * The chill.lisp system for loading cold code into a running SBCL
4063 * Support for the CMU CL "scavenger hook" extension has been removed.
4064 (It was undocumented and unused in the CMU CL sources that SBCL was
4065 derived from, and stale in sbcl-0.6.1.)
4066 * Various errors in the cross-compiler type system were detected
4067 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4068 (enabling various consistency checks). Many of them were fixed,
4069 but some hard problems remain, so the compiler is back to
4070 running without *TYPE-SYSTEM-INITIALIZED* for now.
4071 * As part of the cross-compiler type system cleanup, I implemented
4072 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4073 * I have started adding UNCROSS calls throughout the type system
4074 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4075 on cross-compiler input files will be able to go away, and various
4077 * CONSTANTP now returns true for quoted forms (as explicitly required
4080 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4082 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4083 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4084 (MAKE-LIST -1). In the longer term, it ought to become true
4085 that declarations are assertions even at SAFETY 1. For now, it's not
4086 quite true even at SAFETY 3, but it's at least more nearly true..
4087 (Note that this change seems to increases the size of the system by
4088 O(5%) and to decrease the speed of the compiler by 20% or more.)
4089 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4090 for the problem of printing dozens of lines of distracting information
4091 about low-level system machinery as part of the top stack frame
4092 on entry to the debugger when an undefined function was called.
4093 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4094 is not reset to COMMON-LISP-USER.
4095 * Compilation of stuff related to dyncount.lisp has been made conditional
4096 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4097 smaller. The various dyncount-related symbols have been moved into
4098 a new "SB-DYNCOUNT" package.
4099 * tty-inspect.lisp has been renamed to inspect.lisp.
4100 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4101 feature has gone away. (When we eventually port to other flavors of
4102 libc and/or Unix, we'll try to make the differences between flavors
4103 invisible at the user level.)
4104 * Various other *FEATURES* tags, and/or their associated conditionals,
4105 have been removed if obsolescent, or given better documentation, or
4106 sometimes given more-mnemonic names.
4108 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4110 * tidied up "make.sh" script
4111 * tidied up system directory structure
4112 * better "clean.sh" behavior
4113 * added doc/FOR-CMUCL-DEVELOPERS
4114 * many many small tweaks to output format, e.g. removing possibly-confusing
4115 trailing #\. character in DESCRIBE-INSTANCE
4116 * (EQUALP #\A 'A) no longer signals an error.
4117 * new hashing code, including EQUALP hashing
4118 * tidied up Lisp initialization and toplevel
4119 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4120 * command line argument processing
4121 * added POSIX-GETENV function to deal with Unix-ish environment variables
4122 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4123 terminating SBCL on EOF
4124 * non-verbose GC by default
4125 * There is no more "sbcl" shell script; the sbcl file is now the C
4126 runtime executable (just like CMU CL).
4127 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4129 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4130 symbol and package information as Lisp native symbol and package objects
4131 instead of strings naming symbols and strings naming packages. This way,
4132 whenever packages are renamed (as in warm init), debug information is
4133 transformed along with everything else.
4134 * tweaked the optimization policy declarations which control the building
4135 of SBCL itself. Now, among other things, the system no longer saves
4136 source location debugging information. (This helps two problems at once
4137 by reducing SBCL size and by keeping SBCL from trying to look for its
4138 sources -- which may not exist -- when reporting errors.)
4139 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4140 debugging and testing purposes
4141 * cleaned up printing, making the printer call PRINT-OBJECT for
4142 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4143 methods, giving nearly-ANSI behavior
4144 * converted almost all special variables to use *FOO* naming convention
4145 * deleted PARSE-TIME functionality, since it can be done portably
4146 * moved some files out of cold init into warm init
4147 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4149 * regularized formatting of source files
4150 * added an install.sh script
4151 * fixed ridiculous memory usage of cross-compiler by making
4152 compiler/alloc.lisp not try to do pooling unless it can hook
4153 itself into the GC of the cross-compilation host. Now the system
4154 builds nicely on my old laptop.
4155 * added :SB-ALLOC in target-features.lisp-expr
4156 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4157 was not implemented)
4158 * re-did condition handling and note reporting in the compiler. Notes
4159 are no longer handled by signalling conditions. Style warnings
4160 and warnings are handled more correctly and reported in such a way
4161 that it's easy to find one or the other in your output (so that you
4162 can e.g. figure out which of many problems caused COMPILE-FILE to
4164 * changed the severity of several compiler warnings from full WARNING
4165 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4166 compiler note reporting so that it doesn't use the condition system
4167 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4168 COMPILE-FILE command)
4169 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4170 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4171 no longer supported, since their implementation was deeply intertwingled
4172 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4173 * removed bogus "support" for compiler macros named (SETF FOO), and
4174 removed the compiler macro for SETF INFO (but only after making a fool
4175 of myself on the cmucl-imp mailing list by posting a bogus patch for
4176 DEFINE-COMPILER-MACRO..)
4177 * Compiled files containing forms which have side effects on the Lisp
4178 reader (such as DEFPACKAGE forms) are now handled more correctly.
4179 (Compiler queuing of top level lambdas has been suppressed by setting
4180 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4181 * deleted various currently-unused source files, e.g. gengc.lisp. They
4182 may be added back at some point e.g. when porting to other architectures,
4183 but until they are it's distracting to distribute them and to try to
4185 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4186 there were so many of them they're just distractions, and UNCROSS is
4187 known to be able to handle the current sources
4188 * moved PROFILE functionality into TRACE, so that it will be clear
4189 how the wrapping and unwrapping of functions when you profile them
4190 interacts with the wrapping and unwrapping of functions when you
4191 trace them. (Actually, the functionality isn't there yet, but at least
4192 the interface specification is there. Hopefully, the functionality will
4193 arrive with some future maintenance release.)
4194 * removed host-oops.lisp
4195 * changed signature of QUIT function to allow UNIX-CODE argument
4196 * fixed READ-SEQUENCE bug
4197 * tweaked verbose GC output so that it looks more like the progress
4198 output that ANSI specifies for functions like LOAD
4199 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4200 * added <http://sbcl.sourceforge.com> to the banner information printed by